Top Python libraries of 2023 | Tryolabs

link管理

链接快照平台

输入网页链接，自动生成快照
标签化管理网页链接

相关文章推荐

风流倜傥的小笼包 · error when i run ...· 2 天前 ·

风流倜傥的杨桃 · Is there a python ...· 昨天 ·

没有腹肌的碗 · Anaconda使用、conda的环境管理和 ...· 8 小时前 ·

痛苦的筷子 · Anaconda数据分析&人工智能教程（10 ...· 8 小时前 ·

从未表白的棒棒糖 · python - Flask - ...· 4 小时前 ·

拉风的勺子 · 2024年最新mosdns对接 ...· 2 月前 ·

体贴的水煮肉 · Ociperlimab Plus ...· 4 月前 ·

玩篮球的韭菜 · boxGrob function - ...· 5 月前 ·

腹黑的鼠标 · gradle assemble ...· 6 月前 ·

笑点低的荒野 · android AudioTrack ...· 1 年前 ·

</noscript><div id="__next" data-reactroot=""><div class="css-4fd7nd evajj133 headroom-wrapper"><div style="position:relative;top:0;left:0;right:0;z-index:1;-webkit-transform:translate3D(0, 0, 0);-ms-transform:translate3D(0, 0, 0);transform:translate3D(0, 0, 0)" class="headroom headroom--unfixed"><div class="MuiPaper-root MuiPaper-transparent ex2z6672 css-wmc1wj"><div><span class="MuiTypography-root MuiTypography-body1 css-wmpgkn"><img src="/icons/navy-blue/rocket.svg" width="35" height="25" alt="Notification" style="transform:translate(0, 4px)"/><b class="MuiTypography-root MuiTypography-body1 css-wmpgkn">LLMOps guide | </b>Elevate LLM performance with operational excellence. <b class="css-w0pxt2 ex2z6670"><span style="margin:0 5px 0 0">Dive in!</span><span style="box-sizing:border-box;display:inline-block;overflow:hidden;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;position:relative;max-width:100%"><span style="box-sizing:border-box;display:block;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;max-width:100%"><img style="display:block;max-width:100%;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0" alt="" aria-hidden="true" src="data:image/svg+xml,%3csvg%20xmlns=%27http://www.w3.org/2000/svg%27%20version=%271.1%27%20width=%2726%27%20height=%2722%27/%3e"/></span><img alt="icon" srcset="https://tryolabs.imgix.net/icons/navy-blue/arrow-narrow-right.svg?auto=format&amp;fit=max&amp;w=32 1x, https://tryolabs.imgix.net/icons/navy-blue/arrow-narrow-right.svg?auto=format&amp;fit=max&amp;w=64 2x" src="https://tryolabs.imgix.net/icons/navy-blue/arrow-narrow-right.svg?auto=format&amp;fit=max&amp;w=64" decoding="async" data-nimg="intrinsic" class="img" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%"/></span></b></span><button class="css-v87u9x ex2z6671"><img src="/icons/navy-blue/cross.svg" width="100%" height="100%" alt="Close"/></button></div></div><div class="MuiContainer-root MuiContainer-maxWidthLg evajj130 css-1xkflu"><div class="MuiPaper-root MuiPaper-elevation MuiPaper-rounded MuiPaper-elevation0 css-zmnxdm"><div><div class="css-1og3vwv evajj132"><div class="css-1y673bg e17615560"><a class="" href="/"><span style="box-sizing:border-box;display:inline-block;overflow:hidden;width:45px;height:45px;background:none;opacity:1;border:0;margin:0;padding:0;position:relative"><img alt="Tryolabs logo" src="data:image/gif;base64,R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAAALAAAAAABAAEAAAIBRAA7" decoding="async" data-nimg="fixed" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%"/><noscript><img alt="Tryolabs logo" srcset="https://tryolabs.imgix.net/logos/tryolabs-iso.svg?auto=format&amp;fit=max&amp;w=48 1x, https://tryolabs.imgix.net/logos/tryolabs-iso.svg?auto=format&amp;fit=max&amp;w=96 2x" src="https://tryolabs.imgix.net/logos/tryolabs-iso.svg?auto=format&amp;fit=max&amp;w=96" decoding="async" data-nimg="fixed" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%" loading="lazy"/></noscript></span></a></div><div class="MuiTypography-root MuiTypography-h5 css-17ulk1b">|</div><a href="/blog"><div class="MuiTypography-root MuiTypography-h3 css-174bfc3"> Blog</div></a></div></div><nav><div class="MuiPaper-root MuiPaper-elevation MuiPaper-rounded MuiPaper-elevation0 e18d6qwm8 css-1rbrm99"><div class="css-1u2dhfu e18d6qwm7"><span class="MuiTypography-root MuiTypography-overline dropbtn css-edpvdu">Services<svg class="MuiSvgIcon-root MuiSvgIcon-fontSizeMedium icon css-vubbuv" focusable="false" aria-hidden="true" viewbox="0 0 24 24" data-testid="KeyboardArrowDownIcon"><path d="M7.41 8.59 12 13.17l4.59-4.58L18 10l-6 6-6-6 1.41-1.41z"/></svg></span><div class="MuiPaper-root MuiPaper-elevation MuiPaper-rounded MuiPaper-elevation0 e18d6qwm6 css-1fp1nqp"><div style="cursor:pointer"><div class="css-xpwfow e18d6qwm5"><div class="css-tv35oe e18d6qwm0"><h6 class="MuiTypography-root MuiTypography-subtitle1 css-y55gkz">Predictive Analytics<span class="navbar-triangle"><span style="box-sizing:border-box;display:inline-block;overflow:hidden;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;position:relative;max-width:100%"><span style="box-sizing:border-box;display:block;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;max-width:100%"><img style="display:block;max-width:100%;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0" alt="" aria-hidden="true" src="data:image/svg+xml,%3csvg%20xmlns=%27http://www.w3.org/2000/svg%27%20version=%271.1%27%20width=%2730%27%20height=%2730%27/%3e"/></span><img alt="navbar-triangle" srcset="https://tryolabs.imgix.net/misc/navbar-triangle.svg?auto=format&amp;fit=max&amp;w=32 1x, https://tryolabs.imgix.net/misc/navbar-triangle.svg?auto=format&amp;fit=max&amp;w=64 2x" src="https://tryolabs.imgix.net/misc/navbar-triangle.svg?auto=format&amp;fit=max&amp;w=64" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%"/></span></span></h6></div><button class="arrow css-26bz4 ec4d3xb0"><div class="css-1drbbg7 e7gbu0f0"><span style="box-sizing:border-box;display:inline-block;overflow:hidden;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;position:relative;max-width:100%"><span style="box-sizing:border-box;display:block;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;max-width:100%"><img style="display:block;max-width:100%;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0" alt="" aria-hidden="true" src="data:image/svg+xml,%3csvg%20xmlns=%27http://www.w3.org/2000/svg%27%20version=%271.1%27%20width=%27100%27%20height=%2740%27/%3e"/></span><img alt="Arrow" src="data:image/gif;base64,R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAAALAAAAAABAAEAAAIBRAA7" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%"/><noscript><img alt="Arrow" srcset="https://tryolabs.imgix.net/icons/arrow-stick.svg?auto=format&amp;fit=max&amp;w=128 1x, https://tryolabs.imgix.net/icons/arrow-stick.svg?auto=format&amp;fit=max&amp;w=256 2x" src="https://tryolabs.imgix.net/icons/arrow-stick.svg?auto=format&amp;fit=max&amp;w=256" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%" loading="lazy"/></noscript></span></div></button><div class="css-1owhdxx e18d6qwm3"><div><div class="css-au3ace e18d6qwm2"><a class="" href="/services/price-optimization"><div class="css-1273sb3 e18d6qwm1"><span style="box-sizing:border-box;display:inline-block;overflow:hidden;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;position:relative;max-width:100%"><span style="box-sizing:border-box;display:block;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;max-width:100%"><img style="display:block;max-width:100%;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0" alt="" aria-hidden="true" src="data:image/svg+xml,%3csvg%20xmlns=%27http://www.w3.org/2000/svg%27%20version=%271.1%27%20width=%2730%27%20height=%2730%27/%3e"/></span><img alt="icon" src="data:image/gif;base64,R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAAALAAAAAABAAEAAAIBRAA7" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%"/><noscript><img alt="icon" srcset="https://tryolabs.imgix.net/icons/navy-blue/price-check.svg?auto=format&amp;fit=max&amp;w=32 1x, https://tryolabs.imgix.net/icons/navy-blue/price-check.svg?auto=format&amp;fit=max&amp;w=64 2x" src="https://tryolabs.imgix.net/icons/navy-blue/price-check.svg?auto=format&amp;fit=max&amp;w=64" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%" loading="lazy"/></noscript></span><div class="MuiTypography-root MuiTypography-body1 css-wmpgkn"><strong style="font-weight:400">Price Optimization</strong><p class="MuiTypography-root MuiTypography-caption css-ryfcqn">Dynamic and automated</p></div></div></a></div><div class="css-au3ace e18d6qwm2"><a class="" href="/services/demand-forecasting"><div class="css-1273sb3 e18d6qwm1"><span style="box-sizing:border-box;display:inline-block;overflow:hidden;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;position:relative;max-width:100%"><span style="box-sizing:border-box;display:block;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;max-width:100%"><img style="display:block;max-width:100%;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0" alt="" aria-hidden="true" src="data:image/svg+xml,%3csvg%20xmlns=%27http://www.w3.org/2000/svg%27%20version=%271.1%27%20width=%2730%27%20height=%2730%27/%3e"/></span><img alt="icon" src="data:image/gif;base64,R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAAALAAAAAABAAEAAAIBRAA7" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%"/><noscript><img alt="icon" srcset="https://tryolabs.imgix.net/icons/navy-blue/bar-chart-flag.svg?auto=format&amp;fit=max&amp;w=32 1x, https://tryolabs.imgix.net/icons/navy-blue/bar-chart-flag.svg?auto=format&amp;fit=max&amp;w=64 2x" src="https://tryolabs.imgix.net/icons/navy-blue/bar-chart-flag.svg?auto=format&amp;fit=max&amp;w=64" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%" loading="lazy"/></noscript></span><div class="MuiTypography-root MuiTypography-body1 css-wmpgkn"><strong style="font-weight:400">Demand Forecasting</strong><p class="MuiTypography-root MuiTypography-caption css-ryfcqn">Estimate and predict</p></div></div></a></div><div class="css-au3ace e18d6qwm2"><a class="" href="/services/predictive-maintenance"><div class="css-1273sb3 e18d6qwm1"><span style="box-sizing:border-box;display:inline-block;overflow:hidden;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;position:relative;max-width:100%"><span style="box-sizing:border-box;display:block;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;max-width:100%"><img style="display:block;max-width:100%;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0" alt="" aria-hidden="true" src="data:image/svg+xml,%3csvg%20xmlns=%27http://www.w3.org/2000/svg%27%20version=%271.1%27%20width=%2730%27%20height=%2730%27/%3e"/></span><img alt="icon" src="data:image/gif;base64,R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAAALAAAAAABAAEAAAIBRAA7" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%"/><noscript><img alt="icon" srcset="https://tryolabs.imgix.net/icons/navy-blue/tools.svg?auto=format&amp;fit=max&amp;w=32 1x, https://tryolabs.imgix.net/icons/navy-blue/tools.svg?auto=format&amp;fit=max&amp;w=64 2x" src="https://tryolabs.imgix.net/icons/navy-blue/tools.svg?auto=format&amp;fit=max&amp;w=64" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%" loading="lazy"/></noscript></span><div class="MuiTypography-root MuiTypography-body1 css-wmpgkn"><strong style="font-weight:400">Predictive Maintenance</strong><p class="MuiTypography-root MuiTypography-caption css-ryfcqn">Failure prevention</p></div></div></a></div><div class="css-au3ace e18d6qwm2"><a class="" href="/services/product-matching"><div class="css-1273sb3 e18d6qwm1"><span style="box-sizing:border-box;display:inline-block;overflow:hidden;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;position:relative;max-width:100%"><span style="box-sizing:border-box;display:block;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;max-width:100%"><img style="display:block;max-width:100%;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0" alt="" aria-hidden="true" src="data:image/svg+xml,%3csvg%20xmlns=%27http://www.w3.org/2000/svg%27%20version=%271.1%27%20width=%2730%27%20height=%2730%27/%3e"/></span><img alt="icon" src="data:image/gif;base64,R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAAALAAAAAABAAEAAAIBRAA7" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%"/><noscript><img alt="icon" srcset="https://tryolabs.imgix.net/icons/navy-blue/search-refraction.svg?auto=format&amp;fit=max&amp;w=32 1x, https://tryolabs.imgix.net/icons/navy-blue/search-refraction.svg?auto=format&amp;fit=max&amp;w=64 2x" src="https://tryolabs.imgix.net/icons/navy-blue/search-refraction.svg?auto=format&amp;fit=max&amp;w=64" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%" loading="lazy"/></noscript></span><div class="MuiTypography-root MuiTypography-body1 css-wmpgkn"><strong style="font-weight:400">Product Matching</strong><p class="MuiTypography-root MuiTypography-caption css-ryfcqn">Smart benchmarking</p></div></div></a></div></div></div></div></div><div style="cursor:pointer"><div class="css-xpwfow e18d6qwm5"><div class="css-tv35oe e18d6qwm0"><h6 class="MuiTypography-root MuiTypography-subtitle1 css-y55gkz">Computer Vision<span class="navbar-triangle"><span style="box-sizing:border-box;display:inline-block;overflow:hidden;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;position:relative;max-width:100%"><span style="box-sizing:border-box;display:block;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;max-width:100%"><img style="display:block;max-width:100%;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0" alt="" aria-hidden="true" src="data:image/svg+xml,%3csvg%20xmlns=%27http://www.w3.org/2000/svg%27%20version=%271.1%27%20width=%2730%27%20height=%2730%27/%3e"/></span><img alt="navbar-triangle" srcset="https://tryolabs.imgix.net/misc/navbar-triangle.svg?auto=format&amp;fit=max&amp;w=32 1x, https://tryolabs.imgix.net/misc/navbar-triangle.svg?auto=format&amp;fit=max&amp;w=64 2x" src="https://tryolabs.imgix.net/misc/navbar-triangle.svg?auto=format&amp;fit=max&amp;w=64" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%"/></span></span></h6></div><button class="arrow css-26bz4 ec4d3xb0"><div class="css-1drbbg7 e7gbu0f0"><span style="box-sizing:border-box;display:inline-block;overflow:hidden;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;position:relative;max-width:100%"><span style="box-sizing:border-box;display:block;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;max-width:100%"><img style="display:block;max-width:100%;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0" alt="" aria-hidden="true" src="data:image/svg+xml,%3csvg%20xmlns=%27http://www.w3.org/2000/svg%27%20version=%271.1%27%20width=%27100%27%20height=%2740%27/%3e"/></span><img alt="Arrow" src="data:image/gif;base64,R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAAALAAAAAABAAEAAAIBRAA7" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%"/><noscript><img alt="Arrow" srcset="https://tryolabs.imgix.net/icons/arrow-stick.svg?auto=format&amp;fit=max&amp;w=128 1x, https://tryolabs.imgix.net/icons/arrow-stick.svg?auto=format&amp;fit=max&amp;w=256 2x" src="https://tryolabs.imgix.net/icons/arrow-stick.svg?auto=format&amp;fit=max&amp;w=256" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%" loading="lazy"/></noscript></span></div></button><div class="css-1owhdxx e18d6qwm3"><div><div class="css-au3ace e18d6qwm2"><a class="" href="/services/video-analytics"><div class="css-1273sb3 e18d6qwm1"><span style="box-sizing:border-box;display:inline-block;overflow:hidden;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;position:relative;max-width:100%"><span style="box-sizing:border-box;display:block;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;max-width:100%"><img style="display:block;max-width:100%;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0" alt="" aria-hidden="true" src="data:image/svg+xml,%3csvg%20xmlns=%27http://www.w3.org/2000/svg%27%20version=%271.1%27%20width=%2730%27%20height=%2730%27/%3e"/></span><img alt="icon" src="data:image/gif;base64,R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAAALAAAAAABAAEAAAIBRAA7" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%"/><noscript><img alt="icon" srcset="https://tryolabs.imgix.net/icons/navy-blue/camera.svg?auto=format&amp;fit=max&amp;w=32 1x, https://tryolabs.imgix.net/icons/navy-blue/camera.svg?auto=format&amp;fit=max&amp;w=64 2x" src="https://tryolabs.imgix.net/icons/navy-blue/camera.svg?auto=format&amp;fit=max&amp;w=64" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%" loading="lazy"/></noscript></span><div class="MuiTypography-root MuiTypography-body1 css-wmpgkn"><strong style="font-weight:400">Video Analytics</strong><p class="MuiTypography-root MuiTypography-caption css-ryfcqn">Video to insights</p></div></div></a></div></div></div></div></div><div style="cursor:pointer"><div class="css-xpwfow e18d6qwm5"><div class="css-tv35oe e18d6qwm0"><h6 class="MuiTypography-root MuiTypography-subtitle1 css-y55gkz">Custom AI<span class="navbar-triangle"><span style="box-sizing:border-box;display:inline-block;overflow:hidden;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;position:relative;max-width:100%"><span style="box-sizing:border-box;display:block;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;max-width:100%"><img style="display:block;max-width:100%;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0" alt="" aria-hidden="true" src="data:image/svg+xml,%3csvg%20xmlns=%27http://www.w3.org/2000/svg%27%20version=%271.1%27%20width=%2730%27%20height=%2730%27/%3e"/></span><img alt="navbar-triangle" srcset="https://tryolabs.imgix.net/misc/navbar-triangle.svg?auto=format&amp;fit=max&amp;w=32 1x, https://tryolabs.imgix.net/misc/navbar-triangle.svg?auto=format&amp;fit=max&amp;w=64 2x" src="https://tryolabs.imgix.net/misc/navbar-triangle.svg?auto=format&amp;fit=max&amp;w=64" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%"/></span></span></h6></div><button class="arrow css-26bz4 ec4d3xb0"><div class="css-1drbbg7 e7gbu0f0"><span style="box-sizing:border-box;display:inline-block;overflow:hidden;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;position:relative;max-width:100%"><span style="box-sizing:border-box;display:block;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;max-width:100%"><img style="display:block;max-width:100%;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0" alt="" aria-hidden="true" src="data:image/svg+xml,%3csvg%20xmlns=%27http://www.w3.org/2000/svg%27%20version=%271.1%27%20width=%27100%27%20height=%2740%27/%3e"/></span><img alt="Arrow" src="data:image/gif;base64,R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAAALAAAAAABAAEAAAIBRAA7" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%"/><noscript><img alt="Arrow" srcset="https://tryolabs.imgix.net/icons/arrow-stick.svg?auto=format&amp;fit=max&amp;w=128 1x, https://tryolabs.imgix.net/icons/arrow-stick.svg?auto=format&amp;fit=max&amp;w=256 2x" src="https://tryolabs.imgix.net/icons/arrow-stick.svg?auto=format&amp;fit=max&amp;w=256" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%" loading="lazy"/></noscript></span></div></button><div class="css-1owhdxx e18d6qwm3"><div><div class="css-au3ace e18d6qwm2"><a class="" href="/services/custom-ai"><div class="css-1273sb3 e18d6qwm1"><span style="box-sizing:border-box;display:inline-block;overflow:hidden;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;position:relative;max-width:100%"><span style="box-sizing:border-box;display:block;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;max-width:100%"><img style="display:block;max-width:100%;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0" alt="" aria-hidden="true" src="data:image/svg+xml,%3csvg%20xmlns=%27http://www.w3.org/2000/svg%27%20version=%271.1%27%20width=%2730%27%20height=%2730%27/%3e"/></span><img alt="icon" src="data:image/gif;base64,R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAAALAAAAAABAAEAAAIBRAA7" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%"/><noscript><img alt="icon" srcset="https://tryolabs.imgix.net/icons/navy-blue/target-2.svg?auto=format&amp;fit=max&amp;w=32 1x, https://tryolabs.imgix.net/icons/navy-blue/target-2.svg?auto=format&amp;fit=max&amp;w=64 2x" src="https://tryolabs.imgix.net/icons/navy-blue/target-2.svg?auto=format&amp;fit=max&amp;w=64" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%" loading="lazy"/></noscript></span><div class="MuiTypography-root MuiTypography-body1 css-wmpgkn"><strong style="font-weight:400">Custom AI Solutions</strong><p class="MuiTypography-root MuiTypography-caption css-ryfcqn">AI strategy and adoption</p></div></div></a></div><div class="css-au3ace e18d6qwm2"><a class="" href="/services/data-engineering"><div class="css-1273sb3 e18d6qwm1"><span style="box-sizing:border-box;display:inline-block;overflow:hidden;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;position:relative;max-width:100%"><span style="box-sizing:border-box;display:block;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;max-width:100%"><img style="display:block;max-width:100%;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0" alt="" aria-hidden="true" src="data:image/svg+xml,%3csvg%20xmlns=%27http://www.w3.org/2000/svg%27%20version=%271.1%27%20width=%2730%27%20height=%2730%27/%3e"/></span><img alt="icon" src="data:image/gif;base64,R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAAALAAAAAABAAEAAAIBRAA7" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%"/><noscript><img alt="icon" srcset="https://tryolabs.imgix.net/icons/navy-blue/database.svg?auto=format&amp;fit=max&amp;w=32 1x, https://tryolabs.imgix.net/icons/navy-blue/database.svg?auto=format&amp;fit=max&amp;w=64 2x" src="https://tryolabs.imgix.net/icons/navy-blue/database.svg?auto=format&amp;fit=max&amp;w=64" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%" loading="lazy"/></noscript></span><div class="MuiTypography-root MuiTypography-body1 css-wmpgkn"><strong style="font-weight:400">Data Engineering</strong><p class="MuiTypography-root MuiTypography-caption css-ryfcqn">Raw into actionable data</p></div></div></a></div><div class="css-au3ace e18d6qwm2"><a class="" href="/services/mlops"><div class="css-1273sb3 e18d6qwm1"><span style="box-sizing:border-box;display:inline-block;overflow:hidden;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;position:relative;max-width:100%"><span style="box-sizing:border-box;display:block;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;max-width:100%"><img style="display:block;max-width:100%;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0" alt="" aria-hidden="true" src="data:image/svg+xml,%3csvg%20xmlns=%27http://www.w3.org/2000/svg%27%20version=%271.1%27%20width=%2730%27%20height=%2730%27/%3e"/></span><img alt="icon" src="data:image/gif;base64,R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAAALAAAAAABAAEAAAIBRAA7" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%"/><noscript><img alt="icon" srcset="https://tryolabs.imgix.net/icons/navy-blue/mlops.svg?auto=format&amp;fit=max&amp;w=32 1x, https://tryolabs.imgix.net/icons/navy-blue/mlops.svg?auto=format&amp;fit=max&amp;w=64 2x" src="https://tryolabs.imgix.net/icons/navy-blue/mlops.svg?auto=format&amp;fit=max&amp;w=64" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%" loading="lazy"/></noscript></span><div class="MuiTypography-root MuiTypography-body1 css-wmpgkn"><strong style="font-weight:400">MLOps</strong><p class="MuiTypography-root MuiTypography-caption css-ryfcqn">Optimize your AI pipeline</p></div></div></a></div></div></div></div></div><div style="cursor:pointer"><div class="css-xpwfow e18d6qwm5"><div class="css-tv35oe e18d6qwm0"><h6 class="MuiTypography-root MuiTypography-subtitle1 css-y55gkz">Labs<span class="navbar-triangle"><span style="box-sizing:border-box;display:inline-block;overflow:hidden;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;position:relative;max-width:100%"><span style="box-sizing:border-box;display:block;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;max-width:100%"><img style="display:block;max-width:100%;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0" alt="" aria-hidden="true" src="data:image/svg+xml,%3csvg%20xmlns=%27http://www.w3.org/2000/svg%27%20version=%271.1%27%20width=%2730%27%20height=%2730%27/%3e"/></span><img alt="navbar-triangle" srcset="https://tryolabs.imgix.net/misc/navbar-triangle.svg?auto=format&amp;fit=max&amp;w=32 1x, https://tryolabs.imgix.net/misc/navbar-triangle.svg?auto=format&amp;fit=max&amp;w=64 2x" src="https://tryolabs.imgix.net/misc/navbar-triangle.svg?auto=format&amp;fit=max&amp;w=64" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%"/></span></span></h6></div><button class="arrow css-26bz4 ec4d3xb0"><div class="css-1drbbg7 e7gbu0f0"><span style="box-sizing:border-box;display:inline-block;overflow:hidden;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;position:relative;max-width:100%"><span style="box-sizing:border-box;display:block;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;max-width:100%"><img style="display:block;max-width:100%;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0" alt="" aria-hidden="true" src="data:image/svg+xml,%3csvg%20xmlns=%27http://www.w3.org/2000/svg%27%20version=%271.1%27%20width=%27100%27%20height=%2740%27/%3e"/></span><img alt="Arrow" src="data:image/gif;base64,R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAAALAAAAAABAAEAAAIBRAA7" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%"/><noscript><img alt="Arrow" srcset="https://tryolabs.imgix.net/icons/arrow-stick.svg?auto=format&amp;fit=max&amp;w=128 1x, https://tryolabs.imgix.net/icons/arrow-stick.svg?auto=format&amp;fit=max&amp;w=256 2x" src="https://tryolabs.imgix.net/icons/arrow-stick.svg?auto=format&amp;fit=max&amp;w=256" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%" loading="lazy"/></noscript></span></div></button><div class="css-1owhdxx e18d6qwm3"><div><div class="css-au3ace e18d6qwm2"><a class="" href="/services/edge-computing"><div class="css-1273sb3 e18d6qwm1"><span style="box-sizing:border-box;display:inline-block;overflow:hidden;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;position:relative;max-width:100%"><span style="box-sizing:border-box;display:block;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;max-width:100%"><img style="display:block;max-width:100%;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0" alt="" aria-hidden="true" src="data:image/svg+xml,%3csvg%20xmlns=%27http://www.w3.org/2000/svg%27%20version=%271.1%27%20width=%2730%27%20height=%2730%27/%3e"/></span><img alt="icon" src="data:image/gif;base64,R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAAALAAAAAABAAEAAAIBRAA7" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%"/><noscript><img alt="icon" srcset="https://tryolabs.imgix.net/icons/navy-blue/ai-chip.svg?auto=format&amp;fit=max&amp;w=32 1x, https://tryolabs.imgix.net/icons/navy-blue/ai-chip.svg?auto=format&amp;fit=max&amp;w=64 2x" src="https://tryolabs.imgix.net/icons/navy-blue/ai-chip.svg?auto=format&amp;fit=max&amp;w=64" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%" loading="lazy"/></noscript></span><div class="MuiTypography-root MuiTypography-body1 css-wmpgkn"><strong style="font-weight:400">Edge Computing</strong><p class="MuiTypography-root MuiTypography-caption css-ryfcqn">IoT, real-time data</p></div></div></a></div><div class="css-au3ace e18d6qwm2"><a class="" href="/services/transformers"><div class="css-1273sb3 e18d6qwm1"><span style="box-sizing:border-box;display:inline-block;overflow:hidden;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;position:relative;max-width:100%"><span style="box-sizing:border-box;display:block;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;max-width:100%"><img style="display:block;max-width:100%;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0" alt="" aria-hidden="true" src="data:image/svg+xml,%3csvg%20xmlns=%27http://www.w3.org/2000/svg%27%20version=%271.1%27%20width=%2730%27%20height=%2730%27/%3e"/></span><img alt="icon" src="data:image/gif;base64,R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAAALAAAAAABAAEAAAIBRAA7" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%"/><noscript><img alt="icon" srcset="https://tryolabs.imgix.net/icons/navy-blue/robot.svg?auto=format&amp;fit=max&amp;w=32 1x, https://tryolabs.imgix.net/icons/navy-blue/robot.svg?auto=format&amp;fit=max&amp;w=64 2x" src="https://tryolabs.imgix.net/icons/navy-blue/robot.svg?auto=format&amp;fit=max&amp;w=64" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%" loading="lazy"/></noscript></span><div class="MuiTypography-root MuiTypography-body1 css-wmpgkn"><strong style="font-weight:400">Transformers</strong><p class="MuiTypography-root MuiTypography-caption css-ryfcqn">Optimization and fine-tuning</p></div></div></a></div></div></div></div></div></div></div><div class="css-1u2dhfu e18d6qwm7"><span class="MuiTypography-root MuiTypography-overline dropbtn css-edpvdu">Industry cases<svg class="MuiSvgIcon-root MuiSvgIcon-fontSizeMedium icon css-vubbuv" focusable="false" aria-hidden="true" viewbox="0 0 24 24" data-testid="KeyboardArrowDownIcon"><path d="M7.41 8.59 12 13.17l4.59-4.58L18 10l-6 6-6-6 1.41-1.41z"/></svg></span><div class="MuiPaper-root MuiPaper-elevation MuiPaper-rounded MuiPaper-elevation0 e18d6qwm6 css-1fp1nqp"><a class="" href="/customers"><div style="cursor:pointer;position:relative"><div class="css-xpwfow e18d6qwm5"><div class="css-tv35oe e18d6qwm0"><h6 class="MuiTypography-root MuiTypography-subtitle1 css-y55gkz">Overview</h6></div></div></div></a><a class="" href="/customers/mercadolibre"><div style="cursor:pointer;position:relative"><div class="css-xpwfow e18d6qwm5"><div class="css-tv35oe e18d6qwm0"><h6 class="MuiTypography-root MuiTypography-subtitle1 css-y55gkz">E-commerce &amp; Retail</h6></div></div></div></a><a class="" href="/customers/allianz-gi"><div style="cursor:pointer;position:relative"><div class="css-xpwfow e18d6qwm5"><div class="css-tv35oe e18d6qwm0"><h6 class="MuiTypography-root MuiTypography-subtitle1 css-y55gkz">Insurance</h6></div></div></div></a><a class="" href="/customers/foot-levelers"><div style="cursor:pointer;position:relative"><div class="css-xpwfow e18d6qwm5"><div class="css-tv35oe e18d6qwm0"><h6 class="MuiTypography-root MuiTypography-subtitle1 css-y55gkz">Manufacturing</h6></div></div></div></a><a class="" href="/customers/ses"><div style="cursor:pointer;position:relative"><div class="css-xpwfow e18d6qwm5"><div class="css-tv35oe e18d6qwm0"><h6 class="MuiTypography-root MuiTypography-subtitle1 css-y55gkz">Telecom</h6></div></div></div></a><a class="" href="/customers/halliburton"><div style="cursor:pointer;position:relative"><div class="css-xpwfow e18d6qwm5"><div class="css-tv35oe e18d6qwm0"><h6 class="MuiTypography-root MuiTypography-subtitle1 css-y55gkz">Oil &amp; Gas</h6></div></div></div></a><a class="" href="/customers/cfl"><div style="cursor:pointer;position:relative"><div class="css-xpwfow e18d6qwm5"><div class="css-tv35oe e18d6qwm0"><h6 class="MuiTypography-root MuiTypography-subtitle1 css-y55gkz">Transportation</h6></div></div></div></a><a class="" href="/customers/linc"><div style="cursor:pointer;position:relative"><div class="css-xpwfow e18d6qwm5"><div class="css-tv35oe e18d6qwm0"><h6 class="MuiTypography-root MuiTypography-subtitle1 css-y55gkz">AI4SG</h6></div></div></div></a></div></div><div class="css-1u2dhfu e18d6qwm7"><span class="MuiTypography-root MuiTypography-overline dropbtn css-edpvdu">About<svg class="MuiSvgIcon-root MuiSvgIcon-fontSizeMedium icon css-vubbuv" focusable="false" aria-hidden="true" viewbox="0 0 24 24" data-testid="KeyboardArrowDownIcon"><path d="M7.41 8.59 12 13.17l4.59-4.58L18 10l-6 6-6-6 1.41-1.41z"/></svg></span><div class="MuiPaper-root MuiPaper-elevation MuiPaper-rounded MuiPaper-elevation0 e18d6qwm6 css-1fp1nqp"><a class="" href="/about"><div style="cursor:pointer;position:relative"><div class="css-xpwfow e18d6qwm5"><div class="css-tv35oe e18d6qwm0"><h6 class="MuiTypography-root MuiTypography-subtitle1 css-y55gkz">Company</h6></div></div></div></a><a class="" href="/partnerships"><div style="cursor:pointer;position:relative"><div class="css-xpwfow e18d6qwm5"><div class="css-tv35oe e18d6qwm0"><h6 class="MuiTypography-root MuiTypography-subtitle1 css-y55gkz">Partnerships</h6></div></div></div></a><a class="" href="/team"><div style="cursor:pointer;position:relative"><div class="css-xpwfow e18d6qwm5"><div class="css-tv35oe e18d6qwm0"><h6 class="MuiTypography-root MuiTypography-subtitle1 css-y55gkz">Team</h6></div></div></div></a><a class="" href="/careers"><div style="cursor:pointer;position:relative"><div class="css-xpwfow e18d6qwm5"><div class="css-tv35oe e18d6qwm0"><h6 class="MuiTypography-root MuiTypography-subtitle1 css-y55gkz">Careers</h6></div></div></div></a><a class="" href="/contact-offices"><div style="cursor:pointer;position:relative"><div class="css-xpwfow e18d6qwm5"><div class="css-tv35oe e18d6qwm0"><h6 class="MuiTypography-root MuiTypography-subtitle1 css-y55gkz">Contact</h6></div></div></div></a></div></div><div class="css-1u2dhfu e18d6qwm7"><span class="MuiTypography-root MuiTypography-overline dropbtn css-edpvdu">Resources<svg class="MuiSvgIcon-root MuiSvgIcon-fontSizeMedium icon css-vubbuv" focusable="false" aria-hidden="true" viewbox="0 0 24 24" data-testid="KeyboardArrowDownIcon"><path d="M7.41 8.59 12 13.17l4.59-4.58L18 10l-6 6-6-6 1.41-1.41z"/></svg></span><div class="MuiPaper-root MuiPaper-elevation MuiPaper-rounded MuiPaper-elevation0 e18d6qwm6 css-1fp1nqp"><a class="" href="/guides"><div style="cursor:pointer;position:relative"><div class="css-xpwfow e18d6qwm5"><div class="css-tv35oe e18d6qwm0"><h6 class="MuiTypography-root MuiTypography-subtitle1 css-y55gkz">Guides</h6></div></div></div></a><a class="" href="/blog/collections"><div style="cursor:pointer;position:relative"><div class="css-xpwfow e18d6qwm5"><div class="css-tv35oe e18d6qwm0"><h6 class="MuiTypography-root MuiTypography-subtitle1 css-y55gkz">Collections</h6></div></div></div></a><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/tryolabs" style="color:unset;text-decoration:none" target="_blank"><div style="cursor:pointer;position:relative"><div class="css-xpwfow e18d6qwm5"><div class="css-tv35oe e18d6qwm0"><h6 class="MuiTypography-root MuiTypography-subtitle1 css-y55gkz">Open source</h6></div></div><div class="css-22i38x e18d6qwm4"><span style="box-sizing:border-box;display:inline-block;overflow:hidden;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;position:relative;max-width:100%"><span style="box-sizing:border-box;display:block;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;max-width:100%"><img style="display:block;max-width:100%;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0" alt="" aria-hidden="true" src="data:image/svg+xml,%3csvg%20xmlns=%27http://www.w3.org/2000/svg%27%20version=%271.1%27%20width=%2715%27%20height=%2715%27/%3e"/></span><img alt="" src="data:image/gif;base64,R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAAALAAAAAABAAEAAAIBRAA7" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%"/><noscript><img alt="" srcset="https://tryolabs.imgix.net/icons/navy-blue/out.svg?auto=format&amp;fit=max&amp;w=16 1x, https://tryolabs.imgix.net/icons/navy-blue/out.svg?auto=format&amp;fit=max&amp;w=32 2x" src="https://tryolabs.imgix.net/icons/navy-blue/out.svg?auto=format&amp;fit=max&amp;w=32" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%" loading="lazy"/></noscript></span></div></div></a></div></div><div class="css-1u2dhfu e18d6qwm7"><a class="" href="/blog"><span class="MuiTypography-root MuiTypography-overline dropbtn css-edpvdu">Blog</span></a></div></div></nav></div></div></div></div><div textlength="71" class="css-f2w656 e111ix0u0"/><div class="css-1bs84lh e11jq86a5"><section class="MuiPaper-root MuiPaper-transparent section e11jq86a3 css-1s2d737" backgroundrounded="frenchMiddle" color="frenchMiddle" backgroundalpha="1"><span style="box-sizing:border-box;display:block;overflow:hidden;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;top:0;left:0;bottom:0;right:0"><img imageopacity="0.7" alt="/assets/blog/top-python-libraries-2023/python-23-be70c04ae0.png" sizes="100vw" srcset="https://tryolabs.imgix.net/assets/blog/top-python-libraries-2023/python-23-be70c04ae0.png?auto=format&amp;fit=max&amp;w=640 640w, https://tryolabs.imgix.net/assets/blog/top-python-libraries-2023/python-23-be70c04ae0.png?auto=format&amp;fit=max&amp;w=750 750w, https://tryolabs.imgix.net/assets/blog/top-python-libraries-2023/python-23-be70c04ae0.png?auto=format&amp;fit=max&amp;w=828 828w, https://tryolabs.imgix.net/assets/blog/top-python-libraries-2023/python-23-be70c04ae0.png?auto=format&amp;fit=max&amp;w=1080 1080w, https://tryolabs.imgix.net/assets/blog/top-python-libraries-2023/python-23-be70c04ae0.png?auto=format&amp;fit=max&amp;w=1200 1200w, https://tryolabs.imgix.net/assets/blog/top-python-libraries-2023/python-23-be70c04ae0.png?auto=format&amp;fit=max&amp;w=1920 1920w, https://tryolabs.imgix.net/assets/blog/top-python-libraries-2023/python-23-be70c04ae0.png?auto=format&amp;fit=max&amp;w=2048 2048w, https://tryolabs.imgix.net/assets/blog/top-python-libraries-2023/python-23-be70c04ae0.png?auto=format&amp;fit=max&amp;w=3840 3840w" src="https://tryolabs.imgix.net/assets/blog/top-python-libraries-2023/python-23-be70c04ae0.png?auto=format&amp;fit=max&amp;w=3840" decoding="async" data-nimg="fill" class="css-1pjovlw e11jq86a2" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%;object-fit:cover"/></span><div class="MuiContainer-root MuiContainer-maxWidthLg e11jq86a0 css-169ahgk" style="z-index:10"><div initial="hidden" variants="[object Object]" viewport="[object Object]" class="css-1emghr6 e11jq86a4"><div><div class="css-q6lpcd eu5qz580"><div style="z-index:3"><div class="MuiTypography-root MuiTypography-overline MuiTypography-alignLeft MuiTypography-gutterBottom css-1w78c24">blog</div><span class="MuiTypography-root MuiTypography-h1Small css-f69qjv">Top Python libraries of 2023</span><p class="MuiTypography-root MuiTypography-body2 MuiTypography-gutterBottom css-13gb4dq">Wed, Dec 20, 2023</p></div></div></div></div></div></section></div><div class="css-1bs84lh e11jq86a5"><section class="MuiPaper-root MuiPaper-transparent section e11jq86a3 css-1grnmav" backgroundrounded="plainWhite" color="plainWhite" backgroundalpha="1"><div class="MuiContainer-root MuiContainer-maxWidthLg e11jq86a0 css-169ahgk" style="z-index:10"><div initial="hidden" variants="[object Object]" viewport="[object Object]" class="css-1emghr6 e11jq86a4"><div><p class="MuiTypography-root MuiTypography-body1 eu5qz583 css-12dqwqn" sectioncolor="plainWhite"/><div><div class="css-1fnba78 e10y6y5s1"><div><div class="MuiTypography-root MuiTypography-overline MuiTypography-gutterBottom css-m6hp5p">Authors</div><div class="css-x9mxmb e10y6y5s0"><a class="" href="/blog/authors/alan-descoins"><div class="css-1rudm8l e1mkolzr2"><div class="MuiAvatar-root MuiAvatar-circular MuiAvatar-colorDefault e1mkolzr1 css-1nnvw21"><span style="box-sizing:border-box;display:block;overflow:hidden;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;top:0;left:0;bottom:0;right:0"><img alt="Alan Descoins" src="data:image/gif;base64,R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAAALAAAAAABAAEAAAIBRAA7" decoding="async" data-nimg="fill" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%;filter:blur(20px);background-size:cover;background-image:url(&quot;data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAAQAAAADCAIAAAA7ljmRAAAACXBIWXMAAAsTAAALEwEAmpwYAAAAKklEQVQImWNgYGD4//HGo8W9DBDw/8CalSVJIFZJV/npA6vWzO8OSw0DABr9D1Z0leyqAAAAAElFTkSuQmCC&quot;);background-position:0% 0%"/><noscript><img alt="Alan Descoins" sizes="100vw" srcset="https://tryolabs.imgix.net/assets/team/alan-descoins/face-955b55617a.png?auto=format&amp;fit=max&amp;w=640 640w, https://tryolabs.imgix.net/assets/team/alan-descoins/face-955b55617a.png?auto=format&amp;fit=max&amp;w=750 750w, https://tryolabs.imgix.net/assets/team/alan-descoins/face-955b55617a.png?auto=format&amp;fit=max&amp;w=828 828w, https://tryolabs.imgix.net/assets/team/alan-descoins/face-955b55617a.png?auto=format&amp;fit=max&amp;w=1080 1080w, https://tryolabs.imgix.net/assets/team/alan-descoins/face-955b55617a.png?auto=format&amp;fit=max&amp;w=1200 1200w, https://tryolabs.imgix.net/assets/team/alan-descoins/face-955b55617a.png?auto=format&amp;fit=max&amp;w=1920 1920w, https://tryolabs.imgix.net/assets/team/alan-descoins/face-955b55617a.png?auto=format&amp;fit=max&amp;w=2048 2048w, https://tryolabs.imgix.net/assets/team/alan-descoins/face-955b55617a.png?auto=format&amp;fit=max&amp;w=3840 3840w" src="https://tryolabs.imgix.net/assets/team/alan-descoins/face-955b55617a.png?auto=format&amp;fit=max&amp;w=3840" decoding="async" data-nimg="fill" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%" loading="lazy"/></noscript></span></div><div class="css-15u9uoz e1mkolzr0"><div class="MuiTypography-root MuiTypography-subtitle1 css-1vep6qo">Alan Descoins</div><span class="MuiTypography-root MuiTypography-caption css-okque8">Chief Executive Officer (CEO)</span></div></div></a><a class="" href="/blog/authors/pablo-alfaro"><div class="css-1rudm8l e1mkolzr2"><div class="MuiAvatar-root MuiAvatar-circular MuiAvatar-colorDefault e1mkolzr1 css-1nnvw21"><span style="box-sizing:border-box;display:block;overflow:hidden;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;top:0;left:0;bottom:0;right:0"><img alt="Pablo Alfaro" src="data:image/gif;base64,R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAAALAAAAAABAAEAAAIBRAA7" decoding="async" data-nimg="fill" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%;filter:blur(20px);background-size:cover;background-image:url(&quot;data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAAQAAAADCAIAAAA7ljmRAAAACXBIWXMAAAsTAAALEwEAmpwYAAAAKUlEQVQImWOoqan5//X5/zuXGCDg7+Ujd/ZsArHiY+Kby8pzU5O9fQMAaYsQgpw5jooAAAAASUVORK5CYII=&quot;);background-position:0% 0%"/><noscript><img alt="Pablo Alfaro" sizes="100vw" srcset="https://tryolabs.imgix.net/assets/team/pablo-alfaro/face-39e3b7a86c.png?auto=format&amp;fit=max&amp;w=640 640w, https://tryolabs.imgix.net/assets/team/pablo-alfaro/face-39e3b7a86c.png?auto=format&amp;fit=max&amp;w=750 750w, https://tryolabs.imgix.net/assets/team/pablo-alfaro/face-39e3b7a86c.png?auto=format&amp;fit=max&amp;w=828 828w, https://tryolabs.imgix.net/assets/team/pablo-alfaro/face-39e3b7a86c.png?auto=format&amp;fit=max&amp;w=1080 1080w, https://tryolabs.imgix.net/assets/team/pablo-alfaro/face-39e3b7a86c.png?auto=format&amp;fit=max&amp;w=1200 1200w, https://tryolabs.imgix.net/assets/team/pablo-alfaro/face-39e3b7a86c.png?auto=format&amp;fit=max&amp;w=1920 1920w, https://tryolabs.imgix.net/assets/team/pablo-alfaro/face-39e3b7a86c.png?auto=format&amp;fit=max&amp;w=2048 2048w, https://tryolabs.imgix.net/assets/team/pablo-alfaro/face-39e3b7a86c.png?auto=format&amp;fit=max&amp;w=3840 3840w" src="https://tryolabs.imgix.net/assets/team/pablo-alfaro/face-39e3b7a86c.png?auto=format&amp;fit=max&amp;w=3840" decoding="async" data-nimg="fill" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%" loading="lazy"/></noscript></span></div><div class="css-15u9uoz e1mkolzr0"><div class="MuiTypography-root MuiTypography-subtitle1 css-1vep6qo">Pablo Alfaro</div><span class="MuiTypography-root MuiTypography-caption css-okque8">Tech Expert</span></div></div></a></div></div><div><div class="MuiTypography-root MuiTypography-overline MuiTypography-gutterBottom css-m6hp5p">Share</div><nav class="css-1b3zjki e12pko990"><div><a href="https://twitter.com/intent/tweet?url=https%3A%2F%2Ftryolabs.com%2Fblog%2Ftop-python-libraries-2023&amp;text=Top%20Python%20libraries%20of%202023&amp;via=tryolabs" target="_blank" rel="noreferrer"><span style="box-sizing:border-box;display:inline-block;overflow:hidden;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;position:relative;max-width:100%"><span style="box-sizing:border-box;display:block;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;max-width:100%"><img style="display:block;max-width:100%;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0" alt="" aria-hidden="true" src="data:image/svg+xml,%3csvg%20xmlns=%27http://www.w3.org/2000/svg%27%20version=%271.1%27%20width=%2732%27%20height=%2732%27/%3e"/></span><img alt="twitter" src="data:image/gif;base64,R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAAALAAAAAABAAEAAAIBRAA7" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%"/><noscript><img alt="twitter" srcset="https://tryolabs.imgix.net/logos/twitter.svg?auto=format&amp;fit=max&amp;w=32 1x, https://tryolabs.imgix.net/logos/twitter.svg?auto=format&amp;fit=max&amp;w=64 2x" src="https://tryolabs.imgix.net/logos/twitter.svg?auto=format&amp;fit=max&amp;w=64" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%" loading="lazy"/></noscript></span></a></div><div><a href="https://www.linkedin.com/shareArticle?mini=true&amp;url=https%3A%2F%2Ftryolabs.com%2Fblog%2Ftop-python-libraries-2023&amp;title=Top%20Python%20libraries%20of%202023&amp;source=Tryolabs" target="_blank" rel="noreferrer"><span style="box-sizing:border-box;display:inline-block;overflow:hidden;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;position:relative;max-width:100%"><span style="box-sizing:border-box;display:block;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;max-width:100%"><img style="display:block;max-width:100%;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0" alt="" aria-hidden="true" src="data:image/svg+xml,%3csvg%20xmlns=%27http://www.w3.org/2000/svg%27%20version=%271.1%27%20width=%2732%27%20height=%2732%27/%3e"/></span><img alt="linkedin" src="data:image/gif;base64,R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAAALAAAAAABAAEAAAIBRAA7" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%"/><noscript><img alt="linkedin" srcset="https://tryolabs.imgix.net/logos/linkedin.svg?auto=format&amp;fit=max&amp;w=32 1x, https://tryolabs.imgix.net/logos/linkedin.svg?auto=format&amp;fit=max&amp;w=64 2x" src="https://tryolabs.imgix.net/logos/linkedin.svg?auto=format&amp;fit=max&amp;w=64" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%" loading="lazy"/></noscript></span></a></div><div><a href="https://www.facebook.com/sharer/sharer.php?u=https%3A%2F%2Ftryolabs.com%2Fblog%2Ftop-python-libraries-2023" target="_blank" rel="noreferrer"><span style="box-sizing:border-box;display:inline-block;overflow:hidden;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;position:relative;max-width:100%"><span style="box-sizing:border-box;display:block;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;max-width:100%"><img style="display:block;max-width:100%;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0" alt="" aria-hidden="true" src="data:image/svg+xml,%3csvg%20xmlns=%27http://www.w3.org/2000/svg%27%20version=%271.1%27%20width=%2732%27%20height=%2732%27/%3e"/></span><img alt="facebook" src="data:image/gif;base64,R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAAALAAAAAABAAEAAAIBRAA7" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%"/><noscript><img alt="facebook" srcset="https://tryolabs.imgix.net/logos/facebook.svg?auto=format&amp;fit=max&amp;w=32 1x, https://tryolabs.imgix.net/logos/facebook.svg?auto=format&amp;fit=max&amp;w=64 2x" src="https://tryolabs.imgix.net/logos/facebook.svg?auto=format&amp;fit=max&amp;w=64" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%" loading="lazy"/></noscript></span></a></div></nav></div></div></div><div class="css-1fmljgm eu5qz582"><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">Welcome to the 9th and not-yet-quite-a-decade edition of our yearly Top Python Libraries list! As we journey through the constantly evolving landscape of Python development, it's time again to spotlight the <strong>standout libraries</strong> and <strong>tools</strong> that have caught our attention this year. Since our first edition in 2015, we’ve dedicated ourselves to exploring the depths of the Python ecosystem, and this year is no exception. In fact, it's special — 2023 marks the boom of <strong>Generative AI</strong> and <strong>Large Language Models</strong> (LLMs), a trend that has significantly influenced our picks.</p><p class="MuiTypography-root MuiTypography-body1 ebhkqfd0 css-cl9oo2"/><div>See the other posts in this series: <a href="/blog/2022/12/26/top-python-libraries-2022">2022</a>, <a href="/blog/2021/12/21/top-python-libraries-2021">2021</a>, <a href="/blog/2020/12/21/top-10-python-libraries-of-2020">2020</a>, <a href="/blog/2019/12/10/top-10-python-libraries-of-2019">2019</a>, <a href="/blog/2018/12/28/top-10-python-libraries-2018">2018</a>, <a href="/blog/2017/12/19/top-10-python-libraries-of-2017">2017</a>, <a href="/blog/2016/12/20/top-10-python-libraries-of-2016">2016</a>, <a href="/blog/2015/12/15/top-10-python-libraries-of-2015">2015</a>.</div><div class="css-1lnth3f e179on1a1"><div><div class="MuiTypography-root MuiTypography-body1 css-wmpgkn"><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">Our criteria remain consistent: we're on the lookout for libraries and tools that have been launched or risen to prominence over the past year. To qualify for our top picks, they need to be well-maintained, innovative, and, of course, cool enough to spark your interest. This year, you'll notice an emphasis on libraries that align with the LLM wave, reflecting how these technologies are reshaping the Python (or should we say programming?) landscape.</p></div><div class="css-1a7loz9 e179on1a0"><span style="box-sizing:border-box;display:inline-block;overflow:hidden;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;position:relative;max-width:100%"><span style="box-sizing:border-box;display:block;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;max-width:100%"><img style="display:block;max-width:100%;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0" alt="" aria-hidden="true" src="data:image/svg+xml,%3csvg%20xmlns=%27http://www.w3.org/2000/svg%27%20version=%271.1%27%20width=%2735%27%20height=%2735%27/%3e"/></span><img alt="lightbulb.svg" src="data:image/gif;base64,R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAAALAAAAAABAAEAAAIBRAA7" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%"/><noscript><img alt="lightbulb.svg" srcset="https://tryolabs.imgix.net/icons/white/lightbulb.svg?auto=format&amp;fit=max&amp;w=48 1x, https://tryolabs.imgix.net/icons/white/lightbulb.svg?auto=format&amp;fit=max&amp;w=96 2x" src="https://tryolabs.imgix.net/icons/white/lightbulb.svg?auto=format&amp;fit=max&amp;w=96" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%" loading="lazy"/></noscript></span></div></div></div><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">But fear not, diversity in our selection is still key! Although getting increasingly more difficult every year, we tried to ensure that our full list is not just for AI enthusiasts but also contains gems that can be valuable for a wide range of Python developers, including some niche applications.</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">Our aim with this post is also to ignite conversations about the remarkable libraries and tools that we might have overlooked.</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">So, with eager anticipation, let's dive into the world of Python in 2023!</p><h3 class="MuiTypography-root MuiTypography-h3 MuiTypography-alignLeft css-1a86icr" style="margin:0.5em 0 0.3em 0" id="the-10-main-picks">The 10 main picks</h3><h4 class="MuiTypography-root MuiTypography-h4 MuiTypography-alignLeft css-18lh8g2" style="margin:0.8em 0 0.3em 0" id="1-litellm--call-any-llm-using-openai-format-and-more">1. <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/BerriAI/litellm" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">LiteLLM</a> — call any LLM using OpenAI format, and more</h4><div class="css-1ximumr e1883nak1"><div><span style="box-sizing:border-box;display:inline-block;overflow:hidden;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;position:relative;max-width:100%"><span style="box-sizing:border-box;display:block;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;max-width:100%"><img style="display:block;max-width:100%;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0" alt="" aria-hidden="true" src="data:image/svg+xml,%3csvg%20xmlns=%27http://www.w3.org/2000/svg%27%20version=%271.1%27%20width=%272400%27%20height=%27754%27/%3e"/></span><img alt="Open-source library to simplify LLM completion + embedding calls" src="data:image/gif;base64,R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAAALAAAAAABAAEAAAIBRAA7" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%;filter:blur(20px);background-size:cover;background-image:url(&quot;data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAAQAAAACCAIAAADwyuo0AAAACXBIWXMAABYlAAAWJQFJUiTwAAAAI0lEQVQImWM4cu4SAzO3rpbW1atXGf7+/59XWJKYmPD27VsAn58NquzQZpsAAAAASUVORK5CYII=&quot;);background-position:0% 0%"/><noscript><img alt="Open-source library to simplify LLM completion + embedding calls" srcset="https://tryolabs.imgix.net/assets/blog/top-python-libraries-2023/litellm-fe27bdcfd2.png?auto=format&amp;fit=max&amp;w=3840 1x" src="https://tryolabs.imgix.net/assets/blog/top-python-libraries-2023/litellm-fe27bdcfd2.png?auto=format&amp;fit=max&amp;w=3840" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%" loading="lazy"/></noscript></span></div></div><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">Imagine a world where you're not tethered to a specific model provider, where <strong>switching between LLMs</strong> is as easy as changing one or <strong>zero lines of code</strong>, contrary to rewriting your entire stack using tools like <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://www.langchain.com/" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">LangChain</a>. LiteLLM makes this a reality.</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">LiteLLM's intuitive and non-intrusive design allows seamless integration with various models, providing a <strong>unified format</strong> for inputs and outputs, regardless of the underlying LLM. This flexibility is a game-changer for developers who need to adapt quickly to the rapidly changing landscape of large language models.</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">But LiteLLM isn't just about flexibility; it's about efficiency and control. Features like <strong>streaming responses</strong> and mapping <strong>exceptions</strong> to OpenAI’s format demonstrate its robustness. The ability to track and <strong>manage costs</strong> across different LLMs is particularly noteworthy, ensuring that your AI explorations don't lead to unexpected bills. This level of financial control is vital in a domain where API calls can quickly accumulate costs.</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">Furthermore, LiteLLM comes with a <strong>proxy API server</strong>, enhancing its versatility in integrating on codebases without changing <em>a single line of code</em> ✨<strong>.</strong></p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">Another differentiating feature is the built-in <strong>load balancing</strong> and <strong>rate limiting</strong> features. These capabilities ensure optimal performance and resource utilization, especially when dealing with high volumes of requests across multiple inference calls. Load balancing efficiently distributes incoming API calls, preventing any single model provider from becoming a bottleneck. Meanwhile, the rate limiting function plays a crucial role in maintaining stable and controlled access to the models, safeguarding against excessive usage and potential overloads (forget <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://developer.mozilla.org/en-US/docs/Web/HTTP/Status/429" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">HTTP 429</a>).</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">Of course, the devil is in the details: prompting will work differently for different LLM providers, so you will need to do some work if you truly want to utilize different providers. But it’s good to have a layer that you can rely on to make the switch as easy as possible.</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">Cheers and congrats to the LiteLLM authors. Whether you're a seasoned AI developer or just starting out playing with Generative AI, LiteLLM is a library that belongs in your toolkit!</p><h4 class="MuiTypography-root MuiTypography-h4 MuiTypography-alignLeft css-18lh8g2" style="margin:0.8em 0 0.3em 0" id="2-pyapp--deploy-self-contained-python-applications-anywhere">2. <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/ofek/pyapp" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">PyApp</a> — deploy self contained Python applications anywhere</h4><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">Python's surge in popularity over the years is no secret. Its blend of simplicity and effectiveness, coupled with a vast and friendly community, has made it one of the most widely used programming languages globally. Yet, <strong>distributing Python applications</strong> remains a <strong>long-standing challenge</strong>. Unlike other languages that allow distribution of standalone compiled binaries, Python has struggled in this area, especially when targeting non-technical users. Traditionally, Python applications are shared via Docker container images, a method not always accessible to all users.</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">Enter PyApp: a utility library that simplifies the distribution and installation of Python applications. It achieves this by embedding Python within a <strong>self-installing package</strong> that is compatible across <strong>all operating systems</strong>. But it doesn't stop there! PyApp also supports self-updating capabilities, and can be customized to accommodate a variety of use cases.</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">With PyApp, specifying dependencies for pip installation via <code>requirements.txt</code> files is a breeze, and additional artifacts can be seamlessly embedded into the package using <code>.in</code> files. The library handles the installation process, creating a dedicated directory for Python and the app's required dependencies. On subsequent runs, PyApp checks for the existence of this installation directory, optimizing command-line interface responsiveness.</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">It differentiates from other tools you might know like <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://pyinstaller.org/en/stable/" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">PyInstaller</a> in the fact that PyApp <strong>doesn’t do dependency discovery</strong> but rather relies on an <strong>explicit declaration of dependencies</strong> which seems to solve some <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://news.ycombinator.com/item?id=38630516" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">issues</a> users have been facing.</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">In a nutshell, PyApp presents a straightforward yet potent solution for delivering Python applications to your users. It streamlines distribution and installation procedures, eliminating the hurdles typically associated with these processes. As we look forward to 2024, PyApp is undoubtedly a tool worth adding to your Python arsenal, making the distribution of Python apps as easy as pie!</p><h4 class="MuiTypography-root MuiTypography-h4 MuiTypography-alignLeft css-18lh8g2" style="margin:0.8em 0 0.3em 0" id="3-taipy--build-uis-for-data-apps-even-in-production">3. <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/Avaiga/taipy" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">Taipy</a> — build UIs for data apps, even in production</h4><div class="css-1ximumr e1883nak1"><div><span style="box-sizing:border-box;display:inline-block;overflow:hidden;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;position:relative;max-width:100%"><span style="box-sizing:border-box;display:block;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;max-width:100%"><img style="display:block;max-width:100%;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0" alt="" aria-hidden="true" src="data:image/svg+xml,%3csvg%20xmlns=%27http://www.w3.org/2000/svg%27%20version=%271.1%27%20width=%272400%27%20height=%27682%27/%3e"/></span><img alt="Open-source Python library for building production-ready front-end &amp; back-end in no time" src="data:image/gif;base64,R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAAALAAAAAABAAEAAAIBRAA7" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%;filter:blur(20px);background-size:cover;background-image:url(&quot;data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAAQAAAABCAIAAAB2XpiaAAAACXBIWXMAABYlAAAWJQFJUiTwAAAAFUlEQVQImWP4/PGTEJcgAwPr////ASa3BgbsIOf8AAAAAElFTkSuQmCC&quot;);background-position:0% 0%"/><noscript><img alt="Open-source Python library for building production-ready front-end &amp; back-end in no time" srcset="https://tryolabs.imgix.net/assets/blog/top-python-libraries-2023/tapiy-a9c82648b2.png?auto=format&amp;fit=max&amp;w=3840 1x" src="https://tryolabs.imgix.net/assets/blog/top-python-libraries-2023/tapiy-a9c82648b2.png?auto=format&amp;fit=max&amp;w=3840" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%" loading="lazy"/></noscript></span></div></div><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">In the wise words of fashion designer Giorgio Armani, <em>"Elegance is not about being noticed, it's about being remembered".</em> This philosophy is the guiding star of Taipy, a Python library that empowers data scientists to weave <strong>engaging narratives</strong> from their data.</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">In the realm of data science, the <strong>art of storytelling</strong> can be as crucial as the underlying models. And having the right tool can make all the difference. That is Taipy’s game. Taipy is a <strong>low-code</strong> Python library that enables data scientists to build <strong>interactive Web UIs</strong> for their machine learning products, without the need to master web stack tools. It's all about allowing data scientists to focus on what they do best - extracting insights from data.</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">Taipy's magic lies in its ability to <strong>bind variables and expressions</strong> to the state of <strong>visual components</strong> in the UI. This binding happens in-place when declaring the visual components, drastically <strong>reducing boilerplate</strong> code needed to develop your app. And speaking of visual components, Taipy offers a rich array of options, including plots, sliders, and image grids, to help you build next-gen UIs.</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">But wait, didn’t we have <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://streamlit.io/" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">Streamlit</a> (our <a class="" style="text-decoration:none;color:#3C5BFF" href="/blog/2019/12/10/top-10-python-libraries-of-2019">pick #7 of 2019</a>) for that? Well, that’s where Taipy shines. Unlike Streamlit, which reruns code for the entire page upon detecting an input change, <strong>Taipy employs callbacks to update only the components affected by the change</strong> (much like <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://shiny.posit.co/r/articles/build/understanding-reactivity/" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">shiny does</a> for R for those coming from our brother language) enhancing responsiveness and accelerating development.</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">But the innovation doesn't stop there. Taipy also provides tools for <strong>visually composing machine learning pipelines</strong>, speeding up development while offering an easy way to explain at a high level the steps your model takes. Importantly, data nodes, the inputs to these visual pipelines, can be dynamically changed, <strong>enabling interactive exploration and what-if scenarios</strong>.</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">Whether you choose to self-host your Taipy applications or use Taipy Cloud for serverless deployment, you're in for a seamless experience. Taipy helps you unveil the stories hidden in your data and create memorable, interactive machine learning workflows. Check it out today!</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">Here’s their <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://seattle2023.pydata.org/cfp/talk/WX3D8G/" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">2023 PyData Seattle</a> presentation and here’s their <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://news.ycombinator.com/item?id=38473615" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">launch post on HackerNews</a>; be sure to check the discussion for some interesting information nuggets on the lib.</p><h4 class="MuiTypography-root MuiTypography-h4 MuiTypography-alignLeft css-18lh8g2" style="margin:0.8em 0 0.3em 0" id="4-mlx--machine-learning-on-apple-silicon-with-numpy-like-api">4. <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/ml-explore/mlx" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">MLX</a> — machine learning on Apple silicon with NumPy-like API</h4><div class="css-1ximumr e1883nak1"><div><span style="box-sizing:border-box;display:inline-block;overflow:hidden;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;position:relative;max-width:100%"><span style="box-sizing:border-box;display:block;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;max-width:100%"><img style="display:block;max-width:100%;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0" alt="" aria-hidden="true" src="data:image/svg+xml,%3csvg%20xmlns=%27http://www.w3.org/2000/svg%27%20version=%271.1%27%20width=%272400%27%20height=%27682%27/%3e"/></span><img alt="Array framework for machine learning on Apple silicon." src="data:image/gif;base64,R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAAALAAAAAABAAEAAAIBRAA7" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%;filter:blur(20px);background-size:cover;background-image:url(&quot;data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAAQAAAABCAIAAAB2XpiaAAAACXBIWXMAABYlAAAWJQFJUiTwAAAAFElEQVQImWP4//8/AwODp6cnkAEAKy0G1vaK6gMAAAAASUVORK5CYII=&quot;);background-position:0% 0%"/><noscript><img alt="Array framework for machine learning on Apple silicon." srcset="https://tryolabs.imgix.net/assets/blog/top-python-libraries-2023/mlx-58b6a7338d.png?auto=format&amp;fit=max&amp;w=3840 1x" src="https://tryolabs.imgix.net/assets/blog/top-python-libraries-2023/mlx-58b6a7338d.png?auto=format&amp;fit=max&amp;w=3840" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%" loading="lazy"/></noscript></span></div></div><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">In 2021 Apple introduced it’s <strong>M1 architecture</strong>, now further refined into the current M3, a revolutionary piece of hardware that brought along <strong>performance and efficiency improvements</strong>, as well as a <strong>unified memory buffer</strong> that can be accessed by the CPU, GPU and NPU (Apple’s Neural Engine) without the need to copy information between processing units.</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">In 2023, the spotlight is on Apple's MLX, a groundbreaking <strong>array framework specifically designed for machine learning on Apple silicon</strong>. Brought to you by Apple's own machine learning research team, MLX is setting new standards for the development and deployment of machine learning models on Mac hardware.</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">MLX is designed to retain <strong>familiarity</strong>: its Python API is a close cousin of NumPy's, and it also offers a fully-featured C++ API that mirrors its Python counterpart. <strong>Higher-level packages</strong> like <code>mlx.nn</code> and <code>mlx.optimizers</code> follow the footsteps of PyTorch, making the <strong>construction of complex models</strong> a breeze.</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">But MLX is more than just a familiar face. It introduces innovative features like <strong>composable function transformations</strong>, enabling <strong>automatic differentiation, automatic vectorization, and computation graph optimization</strong>. It also employs <strong>lazy computation</strong>, materializing arrays only when needed, and dynamic graph construction, making <strong>debugging</strong> more intuitive and <strong>avoiding slow compilations</strong> due to changes in function argument shapes.</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">MLX's multi-device capability allows operations to run on any supported processing unit, be it CPU, GPU or NPU. And here's the game-changer: MLX's unified memory model. Unlike other frameworks, <strong>MLX arrays exist in shared memory</strong>, enabling operations on any supported device <strong>without the need for time-expensive data transfers</strong>.</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">This user-friendly yet efficient framework is designed by machine learning researchers for machine learning researchers. Inspired by the likes of NumPy, PyTorch, Jax, and ArrayFire, MLX aims to make it easy for researchers to extend and improve the framework, fostering rapid exploration of new ideas.</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">As we move towards 2024, MLX is definitely a toolkit to keep an eye on. Whether you're an Apple aficionado, a machine learning enthusiast, or both, MLX is poised to revolutionize your machine learning journey and utilize the capabilities of Apple silicon to the fullest.</p><h4 class="MuiTypography-root MuiTypography-h4 MuiTypography-alignLeft css-18lh8g2" style="margin:0.8em 0 0.3em 0" id="5-unstructured--the-ultimate-toolkit-for-text-preprocessing">5. <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/Unstructured-IO/unstructured" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">Unstructured</a> — the ultimate toolkit for text preprocessing</h4><div class="css-1ximumr e1883nak1"><div><span style="box-sizing:border-box;display:inline-block;overflow:hidden;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;position:relative;max-width:100%"><span style="box-sizing:border-box;display:block;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;max-width:100%"><img style="display:block;max-width:100%;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0" alt="" aria-hidden="true" src="data:image/svg+xml,%3csvg%20xmlns=%27http://www.w3.org/2000/svg%27%20version=%271.1%27%20width=%272400%27%20height=%27984%27/%3e"/></span><img alt="Open-source components for pre-processing text documents" src="data:image/gif;base64,R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAAALAAAAAABAAEAAAIBRAA7" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%;filter:blur(20px);background-size:cover;background-image:url(&quot;data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAAQAAAACCAIAAADwyuo0AAAACXBIWXMAABYlAAAWJQFJUiTwAAAAH0lEQVQImWP4//8/g6krg60PiPHr1y8GCRUGaXUgAwCwrQ1Yu+IUgwAAAABJRU5ErkJggg==&quot;);background-position:0% 0%"/><noscript><img alt="Open-source components for pre-processing text documents" srcset="https://tryolabs.imgix.net/assets/blog/top-python-libraries-2023/unstructed-ddd336acfa.png?auto=format&amp;fit=max&amp;w=3840 1x" src="https://tryolabs.imgix.net/assets/blog/top-python-libraries-2023/unstructed-ddd336acfa.png?auto=format&amp;fit=max&amp;w=3840" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%" loading="lazy"/></noscript></span></div></div><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">In the vast landscape of AI, a model is only <strong>as good as the data it feeds on</strong>. But what happens when your data is a chaotic jumble of <strong>raw documents</strong>, scattered across <strong>various formats</strong>? Enter Unstructured, a comprehensive solution for all your text preprocessing needs.</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">Unstructured is the master key to unlock the potential of your raw data, capable of <strong>digesting anything thrown at it</strong> — PDFs, HTML files, Word documents, even images. It's like a gourmet chef for your data, slicing and dicing it into palatable pieces for your AI models.</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">Unstructured offers a smorgasbord of functions, neatly categorized into six types:</p><ol style="margin:0 0 0.5em 0"><li style="margin:0.8em 0"><strong>Partitioning Functions</strong> that transform raw documents into structured and standardized formats.</li><li style="margin:0.8em 0"><strong>Cleaning Functions</strong> that scrub away unwanted text, leaving behind only the relevant nuggets.</li><li style="margin:0.8em 0"><strong>Staging Functions</strong> that format your data, readying it for ML tasks like inference or labeling.</li><li style="margin:0.8em 0"><strong>Extraction Functions</strong> that mine for valuable information within documents, like emails, IP addresses, or pattern-matched texts.</li><li style="margin:0.8em 0"><strong>Chunking Functions</strong> that break down documents into bite-sized sections, perfect for similarity searches and RAG.</li><li style="margin:0.8em 0"><strong>Embedding Functions</strong> that convert preprocessed text into vectors, offering easy-to-use interfaces.</li></ol><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">But that's not all. Unstructured also comes equipped with <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://unstructured-io.github.io/unstructured/ingest/index.html" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">Connectors</a>, a set of tools for seamless <strong>integration with various storage platforms</strong>. From GCS and S3 to Google Drive, One Drive, Notion, and many <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://unstructured-io.github.io/unstructured/ingest/source_connectors.html" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">more</a>. These Connectors work both ways, allowing you to read and write information to and from a plethora of sources.</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">Unstructured is your go-to Python library for <strong>effortless text extraction</strong> from documents. It simplifies <strong>text cleansing</strong>, handling everything from bullet removal to emoji management and language translation. Its extractor functions <strong>parse cleansed text</strong>, retrieving valuable information. You can use built-in extractors or custom templates for structured text.</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">The library's staging functions save preprocessed documents in <strong>diverse formats</strong>, compatible with various tools like Label Studio and Prodigy. It also features a chunker that groups texts by their title building <strong>cohesive, related sections</strong> and <strong>embedding functions</strong> for easy text vectorizing using popular LLM models.</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">With Unstructured in your toolkit, your journey into the world of Large Language Models is set to be a rewarding one. Get ready to unlock the full potential of your data in 2024!</p><h4 class="MuiTypography-root MuiTypography-h4 MuiTypography-alignLeft css-18lh8g2" style="margin:0.8em 0 0.3em 0" id="6-zenml-and-automlops--portable-production-ready-mlops-pipelines">6. <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/zenml-io/zenml" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">ZenML</a> and <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/GoogleCloudPlatform/automlops" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">AutoMLOps</a> — portable, production-ready MLOps pipelines</h4><div class="css-1ximumr e1883nak1"><div><span style="box-sizing:border-box;display:inline-block;overflow:hidden;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;position:relative;max-width:100%"><span style="box-sizing:border-box;display:block;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;max-width:100%"><img style="display:block;max-width:100%;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0" alt="" aria-hidden="true" src="data:image/svg+xml,%3csvg%20xmlns=%27http://www.w3.org/2000/svg%27%20version=%271.1%27%20width=%272400%27%20height=%27678%27/%3e"/></span><img alt="Zenml: open-source MLOps framework for creating portable, production-ready machine learning pipelines &amp; AutoMLOps: a service that generates, provisions, and deploys CI/CD integrated MLOps pipelines, bridging the gap between Data Science and DevOps" src="data:image/gif;base64,R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAAALAAAAAABAAEAAAIBRAA7" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%;filter:blur(20px);background-size:cover;background-image:url(&quot;data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAAQAAAABCAIAAAB2XpiaAAAACXBIWXMAABYlAAAWJQFJUiTwAAAAFUlEQVQImWNgYeD//+9/VGKum08EACAqBSJ2tvwIAAAAAElFTkSuQmCC&quot;);background-position:0% 0%"/><noscript><img alt="Zenml: open-source MLOps framework for creating portable, production-ready machine learning pipelines &amp; AutoMLOps: a service that generates, provisions, and deploys CI/CD integrated MLOps pipelines, bridging the gap between Data Science and DevOps" srcset="https://tryolabs.imgix.net/assets/blog/top-python-libraries-2023/zenml-gcp-06214f6ef4.png?auto=format&amp;fit=max&amp;w=3840 1x" src="https://tryolabs.imgix.net/assets/blog/top-python-libraries-2023/zenml-gcp-06214f6ef4.png?auto=format&amp;fit=max&amp;w=3840" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%" loading="lazy"/></noscript></span></div></div><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">Alright, we're bending the rules a bit here. Two libraries, one entry. But trust us, it’ll be worth it.</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">MLOps is a crucial link between machine learning model <strong>development</strong> and <strong>operations</strong>. It enables a <strong>smooth transition from experimental to production-ready</strong> models. Without MLOps, organizations face navigating a minefield of inefficiencies, wasted resources, and unreliable predictions. It ensures <strong>model reproducibility, scalability, and monitoring</strong>, transforming brilliant ideas from mere theories into <strong>valuable solutions</strong>. Thus, effective <strong>MLOps is the backbone</strong> of successful machine learning initiatives.</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">However, setting up an effective MLOps pipeline can often feel like a Herculean task. The sheer range of <strong>services</strong> and <strong>skills</strong> required can make the process daunting and time-consuming. Furthermore, custom developments often become so <strong>intertwined with application logic</strong> that <strong>maintaining application code becomes a challenging endeavor</strong>.</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">These picks are designed to alleviate these issues and put MLOps best practices front and center of your development.</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/zenml-io/zenml" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">ZenML</a> provides a fast-lane for <strong>transitioning from experimental code to a structured, production-ready pipeline with minimal changes</strong>. This <strong>vendor-agnostic framework</strong> excels at decoupling application code from infrastructure tooling, allowing you to specify different service providers for each component of your infrastructure.</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">With ZenML, your pipeline’s steps are <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/zenml-io/zenml#-3-create-a-pipeline" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">annotated</a> and linked together to form the actual <strong>pipeline</strong>. Then, you configure a <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://docs.zenml.io/user-guide/starter-guide/understand-stacks" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">Zen Stack</a>, specifying the necessary tooling and infrastructure. This stack consists of two main components: the <strong>orchestrator</strong> and the <strong>artifact store</strong>. The orchestrator, which could be a cloud execution service or a simple Python interpreter, runs your pipeline's code. The artifact store is where your data will reside. And the best part? ZenML supports numerous <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://www.zenml.io/integrations" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">integrations</a>, offering additional functionality like <strong>feature stores, data validation, and experiment trackers</strong>.</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">In the other corner, we have <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/GoogleCloudPlatform/automlops" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">Google's AutoMLOps</a>, which shares ZenML's approach of <strong>annotating model development code as components and pipelines</strong>. However, AutoMLOps takes it a step further by automatically generating the necessary code to <strong>containerize</strong> your components, <strong>provision</strong> them to Google’s cloud, and <strong>deploy</strong> those containers to the allocated resources. All of this follows an <strong>automated, reproducible CI/CD process</strong> that helps accelerate your path towards <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://cloud.google.com/architecture/mlops-continuous-delivery-and-automation-pipelines-in-machine-learning#mlops_level_2_cicd_pipeline_automation" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">MLOps maturity level 2</a>.</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">The deployed pipelines can then make use of Google’s extended <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://services.google.com/fh/files/misc/practitioners_guide_to_mlops_whitepaper.pdf" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">MLOps suite of functionalities</a>, ranging from <strong>data processing</strong> and <strong>storage</strong>, to <strong>experimentation</strong>, <strong>training</strong> and <strong>offline evaluation</strong> to <strong>online serving</strong> with configurable <strong>auto-scaling</strong> capabilities and <strong>testing</strong> as well as <strong>monitoring</strong> for drift both in your input data and in your models’ predictions. This enables use of advanced capabilities like <strong>continuous training and deployment</strong>, enhancing your applications’ accuracy and <strong>catching errors</strong> before they can have a real impact on the end-user or business.</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">Uniquely tailored to leverage Google Cloud Services and Vertex AI, AutoMLOps is your magic carpet ride through the cloud. While it's true that this tool is unabashedly <strong>Google-centric</strong>, it's a small concession for the vast benefits you'll reap. If you're open to navigating the Google Cloud, prepare to have a significant portion of your journey streamlined, making your cloud-based endeavors more efficient and enjoyable than ever.</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">With these two powerhouses in your toolkit, you're well-equipped to navigate the complexities of MLOps, ensuring your machine learning initiatives are not only successful but also efficient and manageable.</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">Say goodbye to the days of wrestling with infrastructure and application code, and say hello to the era of streamlined, effective MLOps.</p><h4 class="MuiTypography-root MuiTypography-h4 MuiTypography-alignLeft css-18lh8g2" style="margin:0.8em 0 0.3em 0" id="7-whisperx--speech-recognition-with-word-level-timestamps--diarization">7. <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/m-bain/whisperX" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">WhisperX</a> — speech recognition with word-level timestamps &amp; diarization</h4><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">Have you crossed paths with <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/openai/whisper" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">Whisper</a>, the trailblazing speech recognition model from <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/openai/whisper" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">OpenAI</a>? It's like a multilingual maestro, adept at a concert of tasks such as speech <strong>transcription, timestamping, speech-to-text translation and voice detection</strong> — all in various languages.</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">Now, let me introduce you to its turbo-charged cousin, WhisperX. It's like Whisper after a double shot of espresso, boasting <strong>more accurate timestamping</strong>, <strong>multiple speaker detection</strong>, and a <strong>reduction in hallucinations</strong> (not the psychedelic kind, but the kind where it starts pouring words when no-one is speaking) by <strong>enhancing the voice activity detection</strong>. The best part? It does all this while being faster and less memory-hungry.</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">WhisperX has a secret weapon called <strong>voice activity detection preprocessing</strong>. It's like a bouncer that only lets in audio segments that contain voice activity, and it's much cheaper than letting everyone in for a full automatic speech recognition party. This approach also ensures that the audio is sliced into neat chunks, avoiding any awkward cuts mid-speech and boosting the reliability of whole word detection. Plus, the timestamps are tidied up to match these boundaries, making them even more accurate. The final touch? A phoneme classifier model that aligns word boundaries based on a fixed, per-language dictionary. It's like having a grammar stickler on your team.</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">Curious to see it in action? Check out an English sample of the results on their <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/m-bain/whisperX" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">GitHub repository</a>.</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">The beauty of WhisperX is that the chunks can be processed <strong>in parallel</strong>, like a well-oiled assembly line, which enables <strong>batched inference</strong>. And thanks to a switch to the <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/guillaumekln/faster-whisper" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">Faster Whisper</a> backend, the RAM usage for the whisper-large model has been trimmed down to <strong>less than 8GB</strong>. This means it can now play nice with a <strong>wider range of graphics cards</strong> while also boasting a whopping <strong>70x speedup</strong> on the same model. Like putting your speech recognition on a high-speed treadmill.</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">For the curious minds who love to dive into the nitty-gritty, the authors have generously shared their <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://www.robots.ox.ac.uk/~vgg/publications/2023/Bain23/bain23.pdf" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">pre-print paper</a>.</p><h4 class="MuiTypography-root MuiTypography-h4 MuiTypography-alignLeft css-18lh8g2" style="margin:0.8em 0 0.3em 0" id="8-autogen--llm-conversational-collaborative-suite">8. <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/microsoft/autogen" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">AutoGen</a> — LLM conversational collaborative suite</h4><div class="css-1ximumr e1883nak1"><div><span style="box-sizing:border-box;display:inline-block;overflow:hidden;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;position:relative;max-width:100%"><span style="box-sizing:border-box;display:block;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;max-width:100%"><img style="display:block;max-width:100%;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0" alt="" aria-hidden="true" src="data:image/svg+xml,%3csvg%20xmlns=%27http://www.w3.org/2000/svg%27%20version=%271.1%27%20width=%272400%27%20height=%271236%27/%3e"/></span><img alt="Framework that enables the development of LLM applications using multiple agents that can converse with each other to solve tasks" src="data:image/gif;base64,R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAAALAAAAAABAAEAAAIBRAA7" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%;filter:blur(20px);background-size:cover;background-image:url(&quot;data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAAQAAAACCAIAAADwyuo0AAAACXBIWXMAABYlAAAWJQFJUiTwAAAAI0lEQVQImWP49v0HAyOrpp5Re9dEhn///4fFRPqHhJVU1AAAmNwLf+1KbwwAAAAASUVORK5CYII=&quot;);background-position:0% 0%"/><noscript><img alt="Framework that enables the development of LLM applications using multiple agents that can converse with each other to solve tasks" srcset="https://tryolabs.imgix.net/assets/blog/top-python-libraries-2023/llm-conversational-71fa511abc.png?auto=format&amp;fit=max&amp;w=3840 1x" src="https://tryolabs.imgix.net/assets/blog/top-python-libraries-2023/llm-conversational-71fa511abc.png?auto=format&amp;fit=max&amp;w=3840" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%" loading="lazy"/></noscript></span><span class="MuiTypography-root MuiTypography-caption MuiTypography-alignCenter css-1q3bokj">AutoGen: Agent Collaboration with Human Intervention. Source: AutoGen Multi-Agent Conversation Framework</span></div></div><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">Ever dreamed of a <strong>personal team of software engineers</strong> ready to bring your app ideas to life? Meet Microsoft’s AutoGen, your magic wand that conjures up <strong>conversational agents working in harmony</strong> to achieve a shared goal. It's like the Avengers, but for software development.</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">AutoGen's simplest setup stars two key players: the <code>AssistantAgent</code> and <code>UserProxyAgent</code>. The <code>AssistantAgent</code> is your AI sidekick, designed to <strong>execute tasks independently</strong>. Meanwhile, the <code>UserProxyAgent</code> is your <strong>digital stand-in</strong>, allowing you to step in when <strong>human intervention</strong> is needed but also being <strong>able to act on your behalf.</strong> It's like having your cake and eating it too!</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">The <code>AssistantAgent</code> <strong>performs tasks</strong>, while the <code>UserProxyAgent</code> <strong>provides feedback</strong> to the <code>AssistantAgent</code> on whether the task is succesfully completed. Once the task is set, these agents spring into action, <strong>working autonomously</strong> until they either require human oversight or they've successfully completed the task.</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">This framework shines in use cases like <strong>code generation</strong>. The <code>AssistantAgent</code> crafts the code, while the <code>UserProxyAgent</code> tests it. If bugs appear, the <code>UserProxyAgent</code> provides feedback, and the cycle repeats. It's like owning a <strong>personal code factory</strong>.</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">The beauty of AutoGen lies in its scalability. You can <strong>introduce additional agents to fulfill various roles</strong>, akin to building your software development team. Need a <strong>tester agent</strong> or a <strong>product manager</strong>? AutoGen has got you covered. For inspiration, check out <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/OpenBMB/ChatDev" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">ChatDev</a>, a project that built an entire software team using a similar concept.</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">AutoGen also boasts several <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://microsoft.github.io/autogen/docs/Use-Cases/enhanced_inference#error-handling" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">Enhanced Inference</a> features that empower your workflow. It <strong>unifies the API</strong> for several models, enabling you to link multiple LLMs behind a single agent: if one model fails the next in line takes the mantle, <strong>ensuring robustness</strong>. Furthermore, it provides a <strong>prompt results cache</strong> for faster, cost-effective inference, <strong>chat logging for debugging</strong> and <strong>prompt templating</strong> so that common instructions (such as explain your reasoning step by step) can be reused across multiple tasks.</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">But that's not all. Do you want to <strong>enhance your agents with <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://research.ibm.com/blog/retrieval-augmented-generation-RAG" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">RAG</a></strong>? Just add <code>Retrieve</code> to their class names and <strong>provide a docs path</strong> et voila! Your agents can now tap into custom databases providing context-specific information. Want an <strong>agent that learns</strong> on the go? Try explaining via prompts some stuff to the <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://microsoft.github.io/autogen/docs/reference/agentchat/contrib/teachable_agent" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">TeachableAgent</a> and it will store the information you provide in a <strong>MemGPT-style</strong> embeddings DB, allowing it to <strong>recall and reason over this context</strong> in future interactions. An extremely powerful usage for this is having your agents <strong>reason about what worked</strong> in their interactions and <strong>produce a summary recipe</strong> of the steps that did work, then teach the TeachableAgent that recipe so it knows how to do it next time.</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">To cap it all off, AutoGen offers robust <strong>security measures</strong>. The AutoGen team strongly advises running your agents' code <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://microsoft.github.io/autogen/docs/FAQ#code-execution" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">inside a Docker container</a>. You know, just in case you stumble upon the agent combo that unlocks AGI. 😉😉. With AutoGen, the sky's indeed the limit.</p><h4 class="MuiTypography-root MuiTypography-h4 MuiTypography-alignLeft css-18lh8g2" style="margin:0.8em 0 0.3em 0" id="9-guardrails--babysit-llms-so-they-behave-as-intended">9. <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/guardrails-ai/guardrails" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">Guardrails</a> — babysit LLMs so they behave as intended</h4><div class="css-1ximumr e1883nak1"><div><span style="box-sizing:border-box;display:inline-block;overflow:hidden;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;position:relative;max-width:100%"><span style="box-sizing:border-box;display:block;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;max-width:100%"><img style="display:block;max-width:100%;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0" alt="" aria-hidden="true" src="data:image/svg+xml,%3csvg%20xmlns=%27http://www.w3.org/2000/svg%27%20version=%271.1%27%20width=%272400%27%20height=%27682%27/%3e"/></span><img alt="Open-source Python package for specifying structure and type, validating and correcting the outputs of large language models (LLMs)" src="data:image/gif;base64,R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAAALAAAAAABAAEAAAIBRAA7" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%;filter:blur(20px);background-size:cover;background-image:url(&quot;data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAAQAAAABCAIAAAB2XpiaAAAACXBIWXMAABYlAAAWJQFJUiTwAAAAFUlEQVQImWP49P+fnrERIwPDqlWrACjcBYI1ylm7AAAAAElFTkSuQmCC&quot;);background-position:0% 0%"/><noscript><img alt="Open-source Python package for specifying structure and type, validating and correcting the outputs of large language models (LLMs)" srcset="https://tryolabs.imgix.net/assets/blog/top-python-libraries-2023/guardrails-15538bd6ac.png?auto=format&amp;fit=max&amp;w=3840 1x" src="https://tryolabs.imgix.net/assets/blog/top-python-libraries-2023/guardrails-15538bd6ac.png?auto=format&amp;fit=max&amp;w=3840" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%" loading="lazy"/></noscript></span></div></div><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">Have you ever had a fantastic prompt, carefully crafted through a meticulous <strong>prompt-engineering</strong> process only to find your Large Language Model breaking its chains and responding with a <strong>non-structured answer</strong>? Or an LLM not following your instructions and mentioning your competitor?</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">The flexibility of LLMs provides an immensely vast range of applications but also makes its <strong>integration into a greater pipeline</strong> challenging. Enter Guardrails: a library designed to <strong>specify structure and type, as well as validating and correcting</strong> the outputs of large language models.</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">Guardrails works by defining a <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://docs.guardrailsai.com/defining_guards/rail/#what-is-rail" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">.rail</a> specification, a flavor of XML, that’s intended to be human readable and allows for <strong>defining the structure</strong> of the requested output, <strong>validators</strong> to ensure the values meet a given quality criteria, and <strong>corrective</strong> actions such as re-asking the LLM or filtering out the invalid output if the validations are not met.</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">Here’s an example that specifies the quality criteria (generated length, URL reachability) in the <code>format</code> fields of the <code>RAIL</code> spec, re-asks for <code>explanation</code> and filters the <code>follow_up_url</code> if not valid.</p><div/><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">Guards such as these can be initialized from a <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://docs.guardrailsai.com/defining_guards/rail/" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">RailSpec</a> as well as from <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://docs.guardrailsai.com/defining_guards/pydantic/" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">PyDantic</a> type definitions, or even <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://docs.guardrailsai.com/defining_guards/strings/" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">strings</a> for less complex use cases. The structure can then be used to define the fields that the LLM must produce as well as their types, including <strong>nested objects defined within your logic</strong>. Several validators are provided out of the box ranging from simple <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://docs.guardrailsai.com/api_reference/validators/#guardrails.validators.ValidLength" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">string min/max length</a> or <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://docs.guardrailsai.com/api_reference/validators/#guardrails.validators.RegexMatch" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">regex</a> validations to <strong>syntactic bug-free</strong> <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://docs.guardrailsai.com/api_reference/validators/#guardrails.validators.BugFreePython" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">Python</a> or <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://docs.guardrailsai.com/api_reference/validators/#guardrails.validators.BugFreeSQL" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">SQL</a> code, <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://docs.guardrailsai.com/api_reference/validators/#guardrails.validators.IsProfanityFree" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">profanity-free</a> text and many [more] (<a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://docs.guardrailsai.com/api_reference/validators/" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">https://docs.guardrailsai.com/api_reference/validators/</a>).</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">With Guardrails in your stack, your LLM applications are bound to be <strong>more reliable and robust</strong>, making it easier to <strong>integrate them into your flows</strong>.</p><h4 class="MuiTypography-root MuiTypography-h4 MuiTypography-alignLeft css-18lh8g2" style="margin:0.8em 0 0.3em 0" id="10-temporian--the-pandas-built-for-preprocessing-temporal-data">10. <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/google/temporian" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">Temporian</a> — the “Pandas” built for preprocessing temporal data</h4><div class="css-1ximumr e1883nak1"><div><span style="box-sizing:border-box;display:inline-block;overflow:hidden;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;position:relative;max-width:100%"><span style="box-sizing:border-box;display:block;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;max-width:100%"><img style="display:block;max-width:100%;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0" alt="" aria-hidden="true" src="data:image/svg+xml,%3csvg%20xmlns=%27http://www.w3.org/2000/svg%27%20version=%271.1%27%20width=%272400%27%20height=%27682%27/%3e"/></span><img alt="Tryolabs &amp; Google library for safe, simple and efficient preprocessing and feature engineering of temporal data in Python logo" src="data:image/gif;base64,R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAAALAAAAAABAAEAAAIBRAA7" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%;filter:blur(20px);background-size:cover;background-image:url(&quot;data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAAQAAAABCAIAAAB2XpiaAAAACXBIWXMAABYlAAAWJQFJUiTwAAAAFUlEQVQImWP4//9/dHAoAwPr79+/AS8ABvddHr78AAAAAElFTkSuQmCC&quot;);background-position:0% 0%"/><noscript><img alt="Tryolabs &amp; Google library for safe, simple and efficient preprocessing and feature engineering of temporal data in Python logo" srcset="https://tryolabs.imgix.net/assets/blog/top-python-libraries-2023/temporian-e5626e3776.png?auto=format&amp;fit=max&amp;w=3840 1x" src="https://tryolabs.imgix.net/assets/blog/top-python-libraries-2023/temporian-e5626e3776.png?auto=format&amp;fit=max&amp;w=3840" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%" loading="lazy"/></noscript></span></div></div><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">Time is of the essence. Time is money. Time is running out… You’ve heard it before right? <strong><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://temporian.readthedocs.io/en/stable/user_guide/#what-is-temporal-data" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">Temporal data</a> is ubiquitous in most human activity</strong>, that's why many popular libraries allow representing and manipulating time-series data: a series of values with equal time intervals between them.</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">However, the classical time series approach falls short for representing many kinds of real-world temporal <strong>data that is not equally spaced</strong>, such as user clicks on a website or sales happening in several of a retailer’s stores.</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">The go-to approach in this scenario has always been to use a general data preprocessing tool, such as pandas, and <strong>cram data into a tabular format</strong> by resampling and aggregating it — <strong>losing both flexibility and granularity</strong> in the whole feature engineering process.</p><div class="css-1ximumr e1883nak1"><div><span style="box-sizing:border-box;display:inline-block;overflow:hidden;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;position:relative;max-width:100%"><span style="box-sizing:border-box;display:block;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;max-width:100%"><img style="display:block;max-width:100%;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0" alt="" aria-hidden="true" src="data:image/svg+xml,%3csvg%20xmlns=%27http://www.w3.org/2000/svg%27%20version=%271.1%27%20width=%271201%27%20height=%27651%27/%3e"/></span><img alt="Graphic showing the different kinds of temporal data logo" src="data:image/gif;base64,R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAAALAAAAAABAAEAAAIBRAA7" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%;filter:blur(20px);background-size:cover;background-image:url(&quot;data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAAQAAAACCAIAAADwyuo0AAAACXBIWXMAAAsTAAALEwEAmpwYAAAAI0lEQVQImWP4////xYsXQ0PDamvrGGbPnh0aGsbAwOjp5QUAwP4LmOGb7W4AAAAASUVORK5CYII=&quot;);background-position:0% 0%"/><noscript><img alt="Graphic showing the different kinds of temporal data logo" srcset="https://tryolabs.imgix.net/assets/blog/top-python-libraries-2023/temporal-data-154d546cda.png?auto=format&amp;fit=max&amp;w=1920 1x, https://tryolabs.imgix.net/assets/blog/top-python-libraries-2023/temporal-data-154d546cda.png?auto=format&amp;fit=max&amp;w=3840 2x" src="https://tryolabs.imgix.net/assets/blog/top-python-libraries-2023/temporal-data-154d546cda.png?auto=format&amp;fit=max&amp;w=3840" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%" loading="lazy"/></noscript></span></div></div><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">This is why Tryolabs has <strong>partnered with Google</strong> to develop a state-of-the-art library for temporal data manipulation. <strong>Temporian</strong> allows for <strong>safe</strong>, <strong>simple</strong> and <strong>efficient</strong> preprocessing and feature engineering of temporal data through a Python API backed by a C++ implementation for <strong>blazing-fast speeds</strong>. Temporian supports multivariate time-series, multivariate time-sequences, event logs, and cross-source event streams.</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">At the core of Temporian is the <code>EventSet</code>, which contains timestamped records called Events that hold feature values. <code>EventSets</code> can contain multivariate time series with uniform or non-uniform timestamps, as well as independent series belonging to different entities. Handy <strong>operators like lag, resample, and window functions</strong> help fill in gaps and align timestamps for arithmetic operations.</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">To get a sense of the API, here’s a simple Temporian snippet that computes the daily revenue for each store in a sales dataset, sampled at 10pm every workday:</p><div/><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">Temporian proposes a new paradigm through which to work with <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://temporian.readthedocs.io/en/stable/user_guide/#what-is-temporal-data" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">temporal data</a>. And being tailor-made for it, plus its core computations running as <strong>highly-optimized C++ code</strong>, enables it to make commonplace temporal operations safer, easier to write, and much, much faster to execute — with <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/tryolabs/temporian-examples/blob/main/examples/grouped_simple_moving_average.ipynb" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">benchmarks</a> showing <strong>speedups of upwards of 100x</strong> against equivalent pandas code.</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">If you are curious, check out the <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://temporian.readthedocs.io/en/stable/tutorials/" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">Tutorials</a> in the documentation, which will walk you through real examples using the library. And of course, all feedback is more than welcome :)</p><h3 class="MuiTypography-root MuiTypography-h3 MuiTypography-alignLeft css-1a86icr" style="margin:0.5em 0 0.3em 0" id="runner-ups">Runner-ups</h3><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">The following libraries almost made it to our top picks and we feel they deserve special recognition:</p><ul style="margin:0 0 0.5em 0"><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/Deci-AI/data-gradients" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>DataGradients</strong></a> — say hello to your personal assistant for <strong>computer vision dataset analysis</strong>. DataGradients, a brainchild of <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://deci.ai/" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">Deci AI</a>, is an open-source Python library that comes packed with solutions for common tasks in <strong>vision dataset management</strong>. Whether it's identifying corrupt images, wrong labels, biases, extracting summaries for a deeper understanding of available data, or providing guidance for hyperparameter selection, DataGradients has got you covered.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/descendant-ai/functime" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>functime</strong></a> — a comprehensive library for time-series forecasting and feature extraction on large panel datasets. Offering unique preprocessing options and cross-validation splitters. It stands out for its incredible speed and efficiency, <strong>capable of processing 100,000 time-series in seconds</strong>, thanks to its smart use of <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://pola.rs/" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">Polars</a> for parallel feature engineering. But functime is more than just speed; it's a powerful ally equipped with proven machine learning algorithms and supports exogenous features across all forecasters. It's also automated, managing lags and hyperparameter tuning with FLAML. With over 100 time-series feature extractors available in its custom <code>ts</code> namespace, functime offers an efficient, user-friendly platform that is transforming the way we approach time-series data.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/predibase/lorax" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>LoRaX</strong></a> — an inference server that scales to 1000s of fine-tuned LLMs, by <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://predibase.com/" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">Predibase</a>. LoRaX (”LoRA eXchange”) is designed to serve a <strong>multitude of fine-tuned models on a single GPU</strong>, significantly cutting down the costs associated with serving while maintaining impressive throughput and latency. Key features include Dynamic Adapter Loading, allowing just-in-time loading of any fine-tuned LoRA adapter without affecting concurrent requests; Heterogeneous Continuous Batching, which combines different adapter requests to maintain stable latency and throughput; and Adapter Exchange Scheduling, which efficiently manages adapter prefetching and offloading between GPU and CPU. It supports a range of Large Language Models like Llama and Mistral as base models, and is compatible with LoRA adapters trained using PEFT and Ludwig libraries.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/normal-computing/outlines" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>outlines</strong></a> — <strong>guided text generation</strong> with LLMs, especially focusing on constrained text generation. Allows for model guidance such as, multiple choice, type constraints and dynamic stopping by using a very efficient regex-guided generation, leading to faster inference when compared to other guidance approaches. Generate valid JSON 100% of the time! Check out their <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://news.ycombinator.com/item?id=37125118" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">HN Launch Blog post</a> for more insights.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/facebookresearch/Pearl" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>Pearl</strong></a> — a gift from Meta's Applied Reinforcement Learning (RL) team. It's a <strong>production-ready</strong> AI Agent Library that <strong>prioritizes cumulative long-term rewards over immediate feedback</strong>. These agents can adapt to environments with limited observability, sparse feedback, and high stochasticity, making them a fitting choice for a wide range of real-world problems. Pearl's modular implementation of state-of-the-art RL algorithms is a testament to its versatility and forward-thinking approach.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/huggingface/peft" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>PEFT</strong></a> — brought to you by <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://huggingface.co/" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">Hugging Face</a>, PEFT or Parameter-Efficient Fine-Tuning methods, is your ticket to <strong>fine-tuning pre-trained language models</strong> for a wide array of downstream applications without having to change all the model's parameters, saving on computational resources while also retaining the good properties of more generalistic models and improving performance on your specific task.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/vllm-project/vllm" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>vLLM</strong></a> — A high-throughput and memory-efficient <strong>inference and serving engine</strong> for LLMs. It improves LLM serving performance by 3.5x over previous SOTA, without changing model architecture. Since output size isn’t known beforehand LLMs typically reserve very large and wasteful long blocks of contiguous memory as large as the model output size. The secret (actually not secret, it’s open sourced) sauce is an algorithm called PagedAttention where they physically split the key-value caches that LLMs use into fixed length, short and dynamically allocated blocks. The dynamic assignment prevents memory waste, allowing more prompts to fit into VRAM and hence enabling <strong>faster throughput</strong>. At the time of writing, vLLM has been adopted as the serving solution by <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://youtu.be/5ZlavKF_98U?si=nzy8YNL5yZWkj_lj&amp;t=1260" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">many of the big players</a>, becoming a de-facto standard serving solution.</li></ul><h3 class="MuiTypography-root MuiTypography-h3 MuiTypography-alignLeft css-1a86icr" style="margin:0.5em 0 0.3em 0" id="the-hidden-gems-unveiling-the-long-tail-of-2023-python-libraries">The Hidden Gems: Unveiling the Long Tail of 2023 Python Libraries</h3><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">While our top picks have certainly captured the limelight, there's a whole constellation of Python libraries that didn't make it to the main stage but are equally deserving of your attention. With more than 120 innovative Python libraries analyzed, each with its own unique capabilities and cutting-edge features, if we were to do justice to each one of them, we'd be here until the next Winter Olympics!</p><p class="MuiTypography-root MuiTypography-body1 MuiTypography-alignLeft MuiTypography-paragraph css-1texoht" style="word-break:break-word">To help you navigate this treasure trove, we've neatly <strong>categorized</strong> these libraries and distilled their key features into succinct, one-liner summaries.</p><h4 class="MuiTypography-root MuiTypography-h4 MuiTypography-alignLeft css-18lh8g2" style="margin:0.8em 0 0.3em 0" id="causal-inference">Causal inference</h4><ul style="margin:0 0 0.5em 0"><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/py-why/causaltune" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>CausalTune</strong></a> — a library for automated tuning and selection for causal estimators.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/pymc-labs/CausalPy" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>CausalPy</strong></a> — A Python package for causal inference in quasi-experimental settings.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/py-why/pywhy-llm" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>PyWhy-LLM</strong></a> — experimental library integrating LLM capabilities to support causal analyses.</li></ul><h4 class="MuiTypography-root MuiTypography-h4 MuiTypography-alignLeft css-18lh8g2" style="margin:0.8em 0 0.3em 0" id="cli-llm-tools">CLI LLM Tools</h4><ul style="margin:0 0 0.5em 0"><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/npiv/chatblade" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>Chatblade</strong></a> — ChatGPT on the command line, providing utility methods to extract JSON or Markdown from ChatGPT responses.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/darrenburns/elia" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>Elia</strong></a> — A terminal ChatGPT client built with Textual.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/gorilla-llm/gorilla-cli" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>Gorilla CLI</strong></a> — powers your command-line interactions with a user-centric tool. Simply state your objective, and Gorilla CLI will generate potential commands for execution.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/simonw/llm" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>LLM</strong></a> — A CLI utility and Python library for interacting with Large Language Models, both via remote APIs and models that can be installed and run on your own machine. By the author of <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://datasette.io/" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">Datasette</a>.</li></ul><h4 class="MuiTypography-root MuiTypography-h4 MuiTypography-alignLeft css-18lh8g2" style="margin:0.8em 0 0.3em 0" id="code-tools">Code Tools</h4><ul style="margin:0 0 0.5em 0"><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/Chainlit/chainlit" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>Chainlit</strong></a> — “the Streamlit for ChatGPT”, create ChatGPT-like UIs on top of any Python code in minutes!</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/jameslamb/pydistcheck" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>pydistcheck</strong></a> — Linter that finds portability issues in Python package distributions (wheels and sdists).</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/xetdata/pyxet" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>pyxet</strong></a> — lightweight interface for the XetHub platform, a blob-store with a filesystem like interface and git capabilities.</li></ul><h4 class="MuiTypography-root MuiTypography-h4 MuiTypography-alignLeft css-18lh8g2" style="margin:0.8em 0 0.3em 0" id="computer-vision">Computer vision</h4><ul style="margin:0 0 0.5em 0"><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/deepdoctection/deepdoctection" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>deepdoctection</strong></a> — orchestrates document extraction and document layout analysis tasks using deep learning models.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/facefusion/facefusion" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>FaceFusion</strong></a> — Next generation face swapper and enhancer.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/kadirnar/segment-anything-video" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>MetaSeg</strong></a> — packaged version of the <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/facebookresearch/segment-anything" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">Segment Anything Model (SAM)</a>.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/visioncortex/vtracer" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>VTracer</strong></a> — open source software to convert raster images (like jpg &amp; png) into vector graphics (svg)</li></ul><h4 class="MuiTypography-root MuiTypography-h4 MuiTypography-alignLeft css-18lh8g2" style="margin:0.8em 0 0.3em 0" id="data-and-features">Data and Features</h4><ul style="margin:0 0 0.5em 0"><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/HumanSignal/Adala" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>Adala</strong></a> — Adala - Autonomous DAta (Labeling) Agent framework.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/refuel-ai/autolabel" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>Autolabel</strong></a> — Label, clean and enrich text datasets with LLMs.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/facebookresearch/balance" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>balance</strong></a> — simple workflow and methods for dealing with biased data samples when looking to infer from them to some target population of interest. <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://import-balance.org/blog/2023/01/09/bringing-balance-to-your-data/" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">See launch blog post</a>. By META.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/bytewax/bytewax" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>Bytewax</strong></a> — Python framework that simplifies event and stream processing. Because Bytewax couples the stream and event processing capabilities of Flink, Spark, and Kafka Streams with the friendly and familiar interface of Python, you can re-use the Python libraries you already know and love.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/featureform/featureform" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>Featureform</strong></a> — feature store. Turn your existing data infrastructure into a feature store.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/taylorai/galactic" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>Galactic</strong></a> cleaning and curation tools for massive unstructured text datasets. <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://x.com/andersonbcdefg/status/1703093978982400057?t=9O2R93Ok8Mp6KqFegZVHJQ&amp;s=09" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">Ben (48/100) on X</a></li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/great-expectations/great_expectations" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>Great Expectations</strong></a> — helps data teams build a shared understanding of their data through quality testing, documentation, and profiling.</li></ul><h4 class="MuiTypography-root MuiTypography-h4 MuiTypography-alignLeft css-18lh8g2" style="margin:0.8em 0 0.3em 0" id="data-visualization">Data Visualization</h4><ul style="margin:0 0 0.5em 0"><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/Kanaries/pygwalker" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>PyGWalker</strong></a> — turn your pandas DataFrame into a Tableau-style User Interface for visual analysis.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/mckinsey/vizro" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>Vizro</strong></a> — a toolkit for creating modular data visualization applications. By McKinsey.</li></ul><h4 class="MuiTypography-root MuiTypography-h4 MuiTypography-alignLeft css-18lh8g2" style="margin:0.8em 0 0.3em 0" id="embeddings-and-vector-dbs">Embeddings and Vector DBs</h4><ul style="margin:0 0 0.5em 0"><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/epsilla-cloud/vectordb" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>Epsilla</strong></a> — a high performance Vector Database Management System, focused on scalability, high performance, and cost-effectiveness of vector search.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/lancedb/lancedb" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>LanceDB</strong></a> — open-source database for vector-search built with persistent storage, which greatly simplifies retrieval, filtering and management of embeddings.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/kantord/SeaGOAT" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>SeaGOAT</strong></a> — local search tool that leverages vector embeddings to enable to search your codebase semantically.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/huggingface/text-embeddings-inference" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>Text Embeddings Inference</strong></a> — A blazing fast inference solution for text embeddings models.</li></ul><h4 class="MuiTypography-root MuiTypography-h4 MuiTypography-alignLeft css-18lh8g2" style="margin:0.8em 0 0.3em 0" id="federated-learning">Federated learning</h4><ul style="margin:0 0 0.5em 0"><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/adap/flower" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>Flower</strong></a> — A Friendly Federated Learning Framework.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/NevronAI/metisfl" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>MetisFL</strong></a> — federated learning framework that allows developers to easily federate their machine learning workflows and train their models across distributed data silos without ever collecting the data in a centralized location.</li></ul><h4 class="MuiTypography-root MuiTypography-h4 MuiTypography-alignLeft css-18lh8g2" style="margin:0.8em 0 0.3em 0" id="generative-ai">Generative AI</h4><ul style="margin:0 0 0.5em 0"><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/facebookresearch/audiocraft" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>AudioCraft</strong></a> — library for audio processing and generation with deep learning. By Meta.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/Storia-AI/image-eval" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>Image Eval</strong></a> — A toolkit for evaluating your favorite image generation models. <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://www.linkedin.com/posts/mihaileric_github-storia-aiimage-eval-activity-7108125553634799616-Q21k?utm_source=share&amp;utm_medium=" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">LinkedIn Launch Post</a>.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/brycedrennan/imaginAIry" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>imaginAIry</strong></a> — Pythonic generation of stable diffusion images.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/cabralpinto/modular-diffusion" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>Modular Diffusion</strong></a> — Python library for designing and training your own Diffusion Models with PyTorch.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/sapientml/sapientml" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>SapientML</strong></a> — Generative AutoML for Tabular Data.</li></ul><h4 class="MuiTypography-root MuiTypography-h4 MuiTypography-alignLeft css-18lh8g2" style="margin:0.8em 0 0.3em 0" id="llm-accuracy-enhancements">LLM Accuracy Enhancements</h4><ul style="margin:0 0 0.5em 0"><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/Forethought-Technologies/AutoChain" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>AutoChain</strong></a> — AutoChain: Build lightweight, extensible, and testable LLM Agents</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/Significant-Gravitas/Auto-GPT" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>Auto-GPT</strong></a> — An experimental open-source attempt to make GPT-4 fully autonomous.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/huggingface/autotrain-advanced" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>Autotrain-Advanced</strong></a> — faster and easier training and deployments of state-of-the-art machine learning models.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/stanfordnlp/dspy" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>DSPy</strong></a> — framework for solving advanced tasks with language models (LMs) and retrieval models (RMs). DSPy unifies techniques for prompting and fine-tuning LMs — and approaches for reasoning and tool/retrieval augmentation. By Stanford NLP.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/zilliztech/GPTCache" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>GPTCache</strong></a> — GPTCache is a library for creating semantic cache to store responses from LLM queries.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/raphaelsty/neural-cherche" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>Neural-Cherche</strong></a> — fine-tune neural search models such as Splade, ColBERT, and SparseEmbed on a specific dataset, and run efficient inference on a fine-tuned retriever or ranker.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/cpacker/MemGPT" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>MemGPT</strong></a> — Teaching LLMs memory management for unbounded context 📚🦙.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/karpathy/nanoGPT" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>nanoGPT</strong></a> — The simplest, fastest repository for training/finetuning medium-sized GPTs.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/promptslab/Promptify" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>Promptify</strong></a> — common prompts that work well to leverage LLMs for a variety of scenarios.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/Xpitfire/symbolicai" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>SymbolicAI</strong></a> — Compositional Differentiable Programming Library.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/getzep/zep" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>zep</strong></a> — a long-term memory store for LLM / Chatbot applications. Easily add relevant documents, chat history memory &amp; rich user data to your LLM app's prompts.</li></ul><h4 class="MuiTypography-root MuiTypography-h4 MuiTypography-alignLeft css-18lh8g2" style="margin:0.8em 0 0.3em 0" id="llm-app-building">LLM App Building</h4><ul style="margin:0 0 0.5em 0"><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/safevideo/autollm" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>autollm</strong></a> — Ship RAG based LLM web apps in seconds.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/ThousandBirdsInc/chidori" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>Chidoriv</strong></a> — reactive runtime for building AI agents. It provides a framework for building AI agents that are reactive, observable, and robust. It supports building agents with Node.js, Python, and Rust.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/lm-sys/FastChat" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>FastChat</strong></a> — open platform for training, serving, and evaluating large language model based chatbots.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/Writesonic/GPTRouter" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>GPTRouter</strong></a> — smoothly manage multiple LLMs and image models, speed up responses, and ensure non-stop reliability. Similar to <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/BerriAI/litellm" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">LiteLLM</a>, our top pick!</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/microsoft/guidance" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>guidance</strong></a> — a guidance language for controlling large language models.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/deepset-ai/haystack" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>haystack</strong></a> — end-to-end NLP framework that enables you to build NLP applications powered by LLMs, Transformer models, vector search and more.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/jxnl/instructor" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>Instructor</strong></a> — interact with OpenAI’s function call API from Python code, with Python structs / objects.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/1rgs/jsonformer" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>Jsonformer</strong></a> — A Bulletproof Way to Generate Structured JSON from Language Models</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/langroid/langroid" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>Langroid</strong></a> — easily build LLM-powered applications. Set up Agents, equip them with optional components (LLM, vector-store and methods), assign them tasks, and have them collaboratively solve a problem by exchanging messages.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/pathwaycom/llm-app" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>LLM App</strong></a> — build innovative AI applications by providing real-time human-like responses to user queries based on the most up-to-date knowledge available in your data sources.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/bsilverthorn/maccarone" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>maccarone</strong></a> — AI-managed code blocks in Python, lets you delegate sections of your Python program to AI ownership.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/jackmpcollins/magentic" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>magentic</strong></a> — prompt LLMs as simple Python functions using decorators.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/microsoft/semantic-kernel" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>Semantic Kernel</strong></a> — integrate cutting-edge LLM technology quickly and easily into your apps. Microsoft’s “version” of <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/hwchase17/langchain" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">LangChain</a>.</li></ul><h4 class="MuiTypography-root MuiTypography-h4 MuiTypography-alignLeft css-18lh8g2" style="margin:0.8em 0 0.3em 0" id="llm-code-tools">LLM Code Tools</h4><ul style="margin:0 0 0.5em 0"><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/paul-gauthier/aider" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>aider</strong></a> — command line tool that lets you pair program with GPT-3.5/GPT-4, to edit code stored in your local git repository.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/pgosar/ChatGDB" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>ChatGDB</strong></a> — Harness the power of ChatGPT inside the GDB debugger!</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/Dataherald/dataherald" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>Dataherald</strong></a> — natural language-to-SQL engine built for enterprise-level question answering over structured data. <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://news.ycombinator.com/item?id=37240363" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">HN launch post</a>.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/fauxpilot/fauxpilot" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>FauxPilot</strong></a> — open-source GitHub Copilot server.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/AntonOsika/gpt-engineer" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>GPT Engineer</strong></a> — Specify what you want it to build, the AI asks for clarification, and then builds it.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/mpoon/gpt-repository-loader" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>gpt-repository-loader</strong></a> — command-line tool that converts the contents of a Git repository into a text format that can be interpreted by LLMs.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/santiagobasulto/ipython-gpt" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>ipython-gpt</strong></a> — extension that allows you to use ChatGPT directly from your Jupyter Notebook or IPython Shell.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/jupyterlab/jupyter-ai" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>Jupyter AI</strong></a> — generative AI extension for JupyterLab.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/mljar/plotai" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>PlotAI</strong></a> — use ChatGPT to create plots in Python and Matplotlib directly in your Python script or notebook.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/approximatelabs/sketch" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>sketch</strong></a> — AI code-writing assistant for pandas users that understands the context of your data, greatly improving the relevance of suggestions.</li></ul><h4 class="MuiTypography-root MuiTypography-h4 MuiTypography-alignLeft css-18lh8g2" style="margin:0.8em 0 0.3em 0" id="llm-development">LLM Development</h4><ul style="margin:0 0 0.5em 0"><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/argilla-io/distilabel" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>distilabel</strong></a> — AI Feedback framework for scalable LLM alignment.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/eth-sri/language-model-arithmetic" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>language-model-arithmetic</strong></a> — controlled text generation via language model arithmetic.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/Lightning-AI/lit-gpt" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>Lit-GPT</strong></a> — Hackable implementation of state-of-the-art open-source LLMs based on nanoGPT. Supports flash attention, 4-bit and 8-bit quantization, LoRA and LLaMA-Adapter fine-tuning, pre-training.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/Lightning-AI/lit-llama" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>Lit-LLaMA</strong></a> — Implementation of the LLaMA language model based on nanoGPT. Supports flash attention, Int8 and GPTQ 4bit quantization, LoRA and LLaMA-Adapter fine-tuning, pre-training.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/eth-sri/lmql" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>LMQL</strong></a> — a query language for programming (large) language models.</li></ul><h4 class="MuiTypography-root MuiTypography-h4 MuiTypography-alignLeft css-18lh8g2" style="margin:0.8em 0 0.3em 0" id="llm-experimentation">LLM Experimentation</h4><ul style="margin:0 0 0.5em 0"><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/ianarawjo/ChainForge" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>ChainForge</strong></a> — open-source visual programming environment for battle-testing prompts to LLMs.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/logspace-ai/langflow" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>Langflow</strong></a> — UI for <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/hwchase17/langchain" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">LangChain</a>, designed with react-flow to provide an effortless way to experiment and prototype flows.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/hegelai/prompttools" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>PromptTools</strong></a> — a set of open-source, self-hostable tools for experimenting with, testing, and evaluating LLMs, vector databases, and prompts. <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://news.ycombinator.com/item?id=36958175" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">HN launch post</a>.</li></ul><h4 class="MuiTypography-root MuiTypography-h4 MuiTypography-alignLeft css-18lh8g2" style="margin:0.8em 0 0.3em 0" id="llm-metrics">LLM Metrics</h4><ul style="margin:0 0 0.5em 0"><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/confident-ai/deepeval" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>DeepEval</strong></a> — a simple-to-use, open-source evaluation framework for LLM applications.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/fiddler-labs/fiddler-auditor" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>Fiddler Auditor</strong></a> — a tool to evaluate the robustness of language models.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/explodinggradients/ragas" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>ragas</strong></a> — Evaluation framework for your Retrieval Augmented Generation (RAG) pipelines.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/TonicAI/tvalmetrics" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>tvalmetrics</strong></a> — Metrics to evaluate the quality of responses of your Retrieval Augmented Generation (RAG) applications.</li></ul><h4 class="MuiTypography-root MuiTypography-h4 MuiTypography-alignLeft css-18lh8g2" style="margin:0.8em 0 0.3em 0" id="llm-serving">LLM Serving</h4><ul style="margin:0 0 0.5em 0"><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/ray-project/aviary" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>Aviary</strong></a> — an LLM serving solution that makes it easy to deploy and manage a variety of open source LLMs. By the authors of <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/ray-project/ray" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">Ray</a>.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/nomic-ai/gpt4all" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>GPT4All</strong></a> — an ecosystem to run powerful and customized large language models that work locally on consumer grade CPUs and any GPU (ex pygpt4all/pyllamacpp) with python bindings.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/scaleapi/llm-engine" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>LLM Engine</strong></a> — engine for fine-tuning and serving large language models. By <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://scale.com/" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">Scale AI</a>.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/wealthsimple/llm-gateway" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>LLM Gateway</strong></a> — gateway for secure &amp; reliable communications with OpenAI and other LLM providers.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/punica-ai/punica" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>punica</strong></a> — Serving multiple LoRA finetuned LLM as one.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/jmorganca/ollama" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>Ollama</strong></a> — Get up and running with Llama 2 and other large language models locally.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/amaiya/onprem" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>OnPrem.LLM</strong></a> — tool for running on-premises large language models with non-public data.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/bentoml/OpenLLM" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>OpenLLM</strong></a> — An open platform for operating large language models (LLMs) in production. Fine-tune, serve, deploy, and monitor any LLMs with ease. By <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://bentoml.com/" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">BentoML</a>.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/traceloop/openllmetry" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>OpenLLMetry</strong></a> — Open-source observability for your LLM application, based on OpenTelemetry.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/imartinez/privateGPT" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>privateGPT</strong></a> — Interact privately with your documents using the power of GPT, 100% privately, no data leaks.</li></ul><h4 class="MuiTypography-root MuiTypography-h4 MuiTypography-alignLeft css-18lh8g2" style="margin:0.8em 0 0.3em 0" id="llm-tools">LLM Tools</h4><ul style="margin:0 0 0.5em 0"><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/junruxiong/IncarnaMind" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>IncarnaMind</strong></a> — Connect and chat with your multiple documents (pdf and txt) through GPT and Claude LLMs in a minute.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/ARPSyndicate/puncia" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>Puncia</strong></a> — leveraging AI and other tools, it will tell you everything about a web domain or subdomain, like finding hidden subdomains.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/jamesturk/scrapeghost/" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>scrapeghost</strong></a> — experimental library for scraping websites using OpenAI's GPT API.</li></ul><h4 class="MuiTypography-root MuiTypography-h4 MuiTypography-alignLeft css-18lh8g2" style="margin:0.8em 0 0.3em 0" id="mlops-llmops-devops">MLOps, LLMOps, DevOps</h4><ul style="margin:0 0 0.5em 0"><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/Arize-ai/phoenix" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>phoenix</strong></a> — ML Observability in a Notebook - Uncover Insights, Surface Problems, Monitor, and Fine Tune your Generative LLM, CV and Tabular Models.</li></ul><h4 class="MuiTypography-root MuiTypography-h4 MuiTypography-alignLeft css-18lh8g2" style="margin:0.8em 0 0.3em 0" id="multimodal-ai-tools">Multimodal AI Tools</h4><ul style="margin:0 0 0.5em 0"><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/haotian-liu/LLaVA" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>LLaVAv</strong></a> — Visual Instruction Tuning - Large Language-and-Vision Assistant built towards multimodal GPT-4 level capabilities.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/roboflow/multimodal-maestro" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>Multimodal-Maestro</strong></a> — effective prompting for Large Multimodal Models like GPT-4 Vision, LLaVA or CogVLM.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/facebookresearch/nougat" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>Nougat</strong></a> — the academic document PDF parser that understands LaTeX math and tables.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/unum-cloud/uform" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>UForm</strong></a> — Pocket-Sized Multi-Modal AI For Semantic Search &amp; Recommendation Systems.</li></ul><h4 class="MuiTypography-root MuiTypography-h4 MuiTypography-alignLeft css-18lh8g2" style="margin:0.8em 0 0.3em 0" id="python-ml">Python ML</h4><ul style="margin:0 0 0.5em 0"><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/Felix-Petersen/difflogic" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>difflogic</strong></a> — A Library for Differentiable Logic Gate Networks by <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://twitter.com/FHKPetersen" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">Felix Petersen</a>.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/pytorch-labs/tensordict" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>TensorDict</strong></a> — a dictionary-like class that inherits properties from tensors, such as indexing, shape operations, casting to device etc. The main purpose of TensorDict is to make code-bases more readable and modular by abstracting away tailored operations.</li></ul><h4 class="MuiTypography-root MuiTypography-h4 MuiTypography-alignLeft css-18lh8g2" style="margin:0.8em 0 0.3em 0" id="performance-and-scalability">Performance and scalability</h4><ul style="margin:0 0 0.5em 0"><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/facebookincubator/AITemplate" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>AITemplate</strong></a> — Python framework which renders neural network into high performance CUDA/HIP C++ code. Specialized for FP16 TensorCore (NVIDIA GPU) and MatrixCore (AMD GPU) inference.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/PanQiWei/AutoGPTQ" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>AutoGPTQ</strong></a> — easy-to-use LLMs quantization package with user-friendly APIs, based on GPTQ algorithm.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/mosaicml/composer" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>composer</strong></a> — PyTorch library that enables you to train neural networks faster, at lower cost, and to higher accuracy. Implements more than two dozen speedup methods that can be applied to your training loop in just a few lines of code.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/PotatoSpudowski/fastLLaMa" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>fastLLaMa</strong></a> — Python wrapper to run Inference of LLaMA models using C++.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/hidet-org/hidet" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>hidet</strong></a> — open-source deep learning compiler, written in Python. It supports end-to-end compilation of DNN models from PyTorch and ONNX to efficient cuda kernels.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/lcompilers/lpython" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>LPython</strong></a> — compiler that aggressively optimizes type-annotated Python code. It has several backends, including LLVM, C, C++, and WASM. LPython’s primary tenet is speed. <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://lpython.org/blog/2023/07/lpython-novel-fast-retargetable-python-compiler/" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">Launch blog post</a>.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/bigscience-workshop/petals" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>Petals</strong></a> — Run 100B+ language models at home, BitTorrent-style. Fine-tuning and inference up to 10x faster than offloading.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/alasdairforsythe/tokenmonster" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>TokenMonster</strong></a> — Determine the tokens that optimally represents a dataset at any specific vocabulary size</li></ul><h4 class="MuiTypography-root MuiTypography-h4 MuiTypography-alignLeft css-18lh8g2" style="margin:0.8em 0 0.3em 0" id="python-programming">Python Programming</h4><ul style="margin:0 0 0.5em 0"><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/hbakri/django-ninja-crud" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>Django Ninja CRUD</strong></a> — declarative CRUD Endpoints &amp; Tests with Django Ninja.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/nick-at-dave/DotDict" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>DotDict</strong></a> — A simple Python library to make chained attributes possible.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/grai-io/grai-core" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>grai-core</strong></a> — Data lineage made simple. Grai makes it easy to understand and test how your data relates across databases, warehouses, APIs and dashboards. <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://news.ycombinator.com/item?id=36758122" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">HN launch blog post</a>.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/bugen/pypipe" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>pypipe</strong></a> — Python pipe command line tool.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/reactive-python/reactpy" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>ReactPy</strong></a> — library for building user interfaces in Python without Javascript, made from <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://reactpy.dev/docs/guides/creating-interfaces/your-first-components/index.html#your-first-components" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">components</a> which look and behave similarly to those found in <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://reactjs.org/" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">ReactJS</a>.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/pynecone-io/pynecone" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>Reflex</strong></a> — open source framework to build web apps in pure Python. <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://news.ycombinator.com/item?id=35136827" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">Launch announcement</a>.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/javiber/scrat" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>scrat</strong></a> — caching of expensive function results, like <code>lru_cache</code> but with persistency to disk.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/hynek/svcs/" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>svcs</strong></a> — a dependency container for Python <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://svcs.hynek.me/en/latest/index.html" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">SVCS</a></li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/ZeroIntensity/view.py" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>view.py</strong></a> — lightning-fast, modern web framework. Currently in a very high alpha stage of development. <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://news.ycombinator.com/item?id=37448914" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">HN launch post</a>.</li></ul><h4 class="MuiTypography-root MuiTypography-h4 MuiTypography-alignLeft css-18lh8g2" style="margin:0.8em 0 0.3em 0" id="optimization--math">Optimization / Math</h4><ul style="margin:0 0 0.5em 0"><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/google/lineax" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>Lineax</strong></a> — a JAX library for linear solves and linear least squares. <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://twitter.com/PatrickKidger/status/1666451005901357062" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">Launch Tweet</a>.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/icaros-usc/pyribs/" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>pyribs</strong></a> — a bare-bones Python library for quality diversity optimization.</li></ul><h4 class="MuiTypography-root MuiTypography-h4 MuiTypography-alignLeft css-18lh8g2" style="margin:0.8em 0 0.3em 0" id="reinforcement-learning">Reinforcement Learning</h4><ul style="margin:0 0 0.5em 0"><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/carperai/cheese" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>cheese</strong></a> — adaptive human in the loop evaluation of language and embedding models.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/HumanCompatibleAI/imitation" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>imitation</strong></a> — Clean PyTorch implementations of imitation and reward learning algorithms.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/allenai/RL4LMs" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>RL4LMs</strong></a> — modular RL library to fine-tune language models to human preferences. By AI2.</li><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/carperai/trlx" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>trlX</strong></a> — distributed training of language models with Reinforcement Learning via Human Feedback (RLHF).</li></ul><h4 class="MuiTypography-root MuiTypography-h4 MuiTypography-alignLeft css-18lh8g2" style="margin:0.8em 0 0.3em 0" id="time-series">Time Series</h4><ul style="margin:0 0 0.5em 0"><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/aeon-toolkit/aeon" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>aeon</strong></a> — A unified framework for machine learning with time series.</li></ul><h4 class="MuiTypography-root MuiTypography-h4 MuiTypography-alignLeft css-18lh8g2" style="margin:0.8em 0 0.3em 0" id="video-processing">Video Processing</h4><ul style="margin:0 0 0.5em 0"><li style="margin:0.8em 0"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://github.com/vapoursynth/vapoursynth" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer"><strong>VapourSynth</strong></a> — video processing framework with simplicity in mind. <a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-15nqva1" href="https://www.vapoursynth.com/doc/pythonreference.html" style="text-decoration:none;color:#3C5BFF" target="_blank" rel="nofollow noopener noreferrer">Python docs</a>.</li></ul><div class="css-qkomnt eu5qz581"/><div class="css-1bs84lh e11jq86a5"><section class="MuiPaper-root MuiPaper-transparent section e11jq86a3 css-j3282q" backgroundrounded="plainWhite" color="plainWhite" backgroundalpha="1"><div class="MuiContainer-root MuiContainer-maxWidthLg e11jq86a0 css-169ahgk" style="z-index:10"><div class="css-1wfs28s" style="opacity:0"><div style="opacity:0;transform:translateY(50px) translateZ(0)"><div class="css-1xdhyk6 e1fqmuyn0"><div class="MuiTypography-root MuiTypography-body1 MuiTypography-alignCenter css-1clwwok"><h2 class="MuiTypography-root MuiTypography-h2 MuiTypography-gutterBottom css-ewrohi">Wondering how <mark class="eucalyptus">AI</mark> can help you?</h2><button class="MuiButtonBase-root MuiButton-root MuiButton-contained MuiButton-containedEucalyptus MuiButton-sizeMedium MuiButton-containedSizeMedium MuiButton-disableElevation MuiButton-root MuiButton-contained MuiButton-containedEucalyptus MuiButton-sizeMedium MuiButton-containedSizeMedium MuiButton-disableElevation css-1mez2mw" tabindex="0" type="button">Let's talk</button></div></div></div></div></div></section></div></div></div></div></div></section></div><div class="css-1bs84lh e11jq86a5"><section class="MuiPaper-root MuiPaper-transparent section e11jq86a3 css-1ast0va" backgroundrounded="frenchMiddle" color="frenchMiddle" backgroundalpha="1"><div class="MuiContainer-root MuiContainer-maxWidthLg e11jq86a0 css-169ahgk" style="z-index:10"><div initial="hidden" variants="[object Object]" viewport="[object Object]" class="css-1emghr6 e11jq86a4"><div><div class="css-145b9pr e1pm0t65"><div class="css-6pmffo e1pm0t64"><div class="css-1y673bg e17615560"><a class="" href="/"><span style="box-sizing:border-box;display:inline-block;overflow:hidden;width:150px;height:60px;background:none;opacity:1;border:0;margin:0;padding:0;position:relative"><img alt="Tryolabs logo" src="data:image/gif;base64,R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAAALAAAAAABAAEAAAIBRAA7" decoding="async" data-nimg="fixed" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%"/><noscript><img alt="Tryolabs logo" srcset="https://tryolabs.imgix.net/logos/tryolabs-white.svg?auto=format&amp;fit=max&amp;w=256 1x, https://tryolabs.imgix.net/logos/tryolabs-white.svg?auto=format&amp;fit=max&amp;w=384 2x" src="https://tryolabs.imgix.net/logos/tryolabs-white.svg?auto=format&amp;fit=max&amp;w=384" decoding="async" data-nimg="fixed" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%" loading="lazy"/></noscript></span></a></div><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-105x4vy" href="mailto:hello@tryolabs.com" style="text-decoration:underline;color:inherit" target="_blank">hello@tryolabs.com</a></div><div class="css-aflfjq e1pm0t63"><div class="css-131fx2n e1pm0t62"><div class="MuiTypography-root MuiTypography-h6 MuiTypography-gutterBottom css-fwf0an">Services</div><p class="MuiTypography-root MuiTypography-body1 css-e6mi5y"><a class="" style="text-decoration:none;color:inherit" href="/services/custom-ai">Custom AI</a></p><p class="MuiTypography-root MuiTypography-body1 css-e6mi5y"><a class="" style="text-decoration:none;color:inherit" href="/services/video-analytics">Video Analytics</a></p><p class="MuiTypography-root MuiTypography-body1 css-e6mi5y"><a class="" style="text-decoration:none;color:inherit" href="/services/demand-forecasting">Demand Forecasting</a></p><p class="MuiTypography-root MuiTypography-body1 css-e6mi5y"><a class="" style="text-decoration:none;color:inherit" href="/services/price-optimization">Price Optimization</a></p><p class="MuiTypography-root MuiTypography-body1 css-e6mi5y"><a class="" style="text-decoration:none;color:inherit" href="/services/product-matching">Product Matching</a></p><p class="MuiTypography-root MuiTypography-body1 css-e6mi5y"><a class="" style="text-decoration:none;color:inherit" href="/services/predictive-maintenance">Predictive Maintenance</a></p><p class="MuiTypography-root MuiTypography-body1 css-e6mi5y"><a class="" style="text-decoration:none;color:inherit" href="/services/edge-computing">AI On The Edge</a></p><p class="MuiTypography-root MuiTypography-body1 css-e6mi5y"><a class="" style="text-decoration:none;color:inherit" href="/services/transformers">Transformers</a></p><p class="MuiTypography-root MuiTypography-body1 css-e6mi5y"><a class="" style="text-decoration:none;color:inherit" href="/services/data-engineering">Data Engineering</a></p></div><div class="css-131fx2n e1pm0t62"><div class="MuiTypography-root MuiTypography-h6 MuiTypography-gutterBottom css-fwf0an">Resources</div><p class="MuiTypography-root MuiTypography-body1 css-e6mi5y"><a class="" style="text-decoration:none;color:inherit" href="/blog">Blog</a></p><p class="MuiTypography-root MuiTypography-body1 css-e6mi5y"><a class="" style="text-decoration:none;color:inherit" href="/guides">Guides</a></p><p class="MuiTypography-root MuiTypography-body1 css-e6mi5y"><a class="" style="text-decoration:none;color:inherit" href="/blog/collections">Collections</a></p><p class="MuiTypography-root MuiTypography-body1 css-e6mi5y"><a class="MuiTypography-root MuiTypography-inherit MuiLink-root MuiLink-underlineAlways css-105x4vy" href="https://github.com/tryolabs/" style="text-decoration:none;color:inherit" target="_blank">Open Source</a></p></div><div class="css-131fx2n e1pm0t62"><div class="MuiTypography-root MuiTypography-h6 MuiTypography-gutterBottom css-fwf0an">Company</div><p class="MuiTypography-root MuiTypography-body1 css-e6mi5y"><a class="" style="text-decoration:none;color:inherit" href="/about">About</a></p><p class="MuiTypography-root MuiTypography-body1 css-e6mi5y"><a class="" style="text-decoration:none;color:inherit" href="/partnerships">Partnerships</a></p><p class="MuiTypography-root MuiTypography-body1 css-e6mi5y"><a class="" style="text-decoration:none;color:inherit" href="/team">Team</a></p><p class="MuiTypography-root MuiTypography-body1 css-e6mi5y"><a class="" style="text-decoration:none;color:inherit" href="/careers">Careers</a></p><p class="MuiTypography-root MuiTypography-body1 css-e6mi5y"><a class="" style="text-decoration:none;color:inherit" href="/customers">Customers</a></p><p class="MuiTypography-root MuiTypography-body1 css-e6mi5y"><a class="" style="text-decoration:none;color:inherit" href="/contact-offices">Contact</a></p><p class="MuiTypography-root MuiTypography-body1 css-e6mi5y"><a class="" style="text-decoration:none;color:inherit" href="/terms">Terms and Conditions</a></p></div></div><div class="css-91f7z2 e1pm0t61"><nav class="css-lrwzj5 e12pko990"><div><a href="https://twitter.com/tryolabs" target="_blank" rel="noreferrer"><span style="box-sizing:border-box;display:inline-block;overflow:hidden;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;position:relative;max-width:100%"><span style="box-sizing:border-box;display:block;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;max-width:100%"><img style="display:block;max-width:100%;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0" alt="" aria-hidden="true" src="data:image/svg+xml,%3csvg%20xmlns=%27http://www.w3.org/2000/svg%27%20version=%271.1%27%20width=%2732%27%20height=%2732%27/%3e"/></span><img alt="twitter" src="data:image/gif;base64,R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAAALAAAAAABAAEAAAIBRAA7" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%"/><noscript><img alt="twitter" srcset="https://tryolabs.imgix.net/logos/twitter.svg?auto=format&amp;fit=max&amp;w=32 1x, https://tryolabs.imgix.net/logos/twitter.svg?auto=format&amp;fit=max&amp;w=64 2x" src="https://tryolabs.imgix.net/logos/twitter.svg?auto=format&amp;fit=max&amp;w=64" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%" loading="lazy"/></noscript></span></a></div><div><a href="https://www.linkedin.com/company/tryolabs/" target="_blank" rel="noreferrer"><span style="box-sizing:border-box;display:inline-block;overflow:hidden;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;position:relative;max-width:100%"><span style="box-sizing:border-box;display:block;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;max-width:100%"><img style="display:block;max-width:100%;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0" alt="" aria-hidden="true" src="data:image/svg+xml,%3csvg%20xmlns=%27http://www.w3.org/2000/svg%27%20version=%271.1%27%20width=%2732%27%20height=%2732%27/%3e"/></span><img alt="linkedin" src="data:image/gif;base64,R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAAALAAAAAABAAEAAAIBRAA7" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%"/><noscript><img alt="linkedin" srcset="https://tryolabs.imgix.net/logos/linkedin.svg?auto=format&amp;fit=max&amp;w=32 1x, https://tryolabs.imgix.net/logos/linkedin.svg?auto=format&amp;fit=max&amp;w=64 2x" src="https://tryolabs.imgix.net/logos/linkedin.svg?auto=format&amp;fit=max&amp;w=64" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%" loading="lazy"/></noscript></span></a></div><div><a href="https://github.com/tryolabs" target="_blank" rel="noreferrer"><span style="box-sizing:border-box;display:inline-block;overflow:hidden;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;position:relative;max-width:100%"><span style="box-sizing:border-box;display:block;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;max-width:100%"><img style="display:block;max-width:100%;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0" alt="" aria-hidden="true" src="data:image/svg+xml,%3csvg%20xmlns=%27http://www.w3.org/2000/svg%27%20version=%271.1%27%20width=%2732%27%20height=%2732%27/%3e"/></span><img alt="github" src="data:image/gif;base64,R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAAALAAAAAABAAEAAAIBRAA7" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%"/><noscript><img alt="github" srcset="https://tryolabs.imgix.net/logos/github.svg?auto=format&amp;fit=max&amp;w=32 1x, https://tryolabs.imgix.net/logos/github.svg?auto=format&amp;fit=max&amp;w=64 2x" src="https://tryolabs.imgix.net/logos/github.svg?auto=format&amp;fit=max&amp;w=64" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%" loading="lazy"/></noscript></span></a></div><div><a href="https://www.instagram.com/tryolabsteam/" target="_blank" rel="noreferrer"><span style="box-sizing:border-box;display:inline-block;overflow:hidden;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;position:relative;max-width:100%"><span style="box-sizing:border-box;display:block;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;max-width:100%"><img style="display:block;max-width:100%;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0" alt="" aria-hidden="true" src="data:image/svg+xml,%3csvg%20xmlns=%27http://www.w3.org/2000/svg%27%20version=%271.1%27%20width=%2732%27%20height=%2732%27/%3e"/></span><img alt="instagram" src="data:image/gif;base64,R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAAALAAAAAABAAEAAAIBRAA7" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%"/><noscript><img alt="instagram" srcset="https://tryolabs.imgix.net/logos/instagram.svg?auto=format&amp;fit=max&amp;w=32 1x, https://tryolabs.imgix.net/logos/instagram.svg?auto=format&amp;fit=max&amp;w=64 2x" src="https://tryolabs.imgix.net/logos/instagram.svg?auto=format&amp;fit=max&amp;w=64" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%" loading="lazy"/></noscript></span></a></div><div><a href="https://www.youtube.com/channel/UCsGnAwYS4Ve9JKnaCvequUA" target="_blank" rel="noreferrer"><span style="box-sizing:border-box;display:inline-block;overflow:hidden;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;position:relative;max-width:100%"><span style="box-sizing:border-box;display:block;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0;max-width:100%"><img style="display:block;max-width:100%;width:initial;height:initial;background:none;opacity:1;border:0;margin:0;padding:0" alt="" aria-hidden="true" src="data:image/svg+xml,%3csvg%20xmlns=%27http://www.w3.org/2000/svg%27%20version=%271.1%27%20width=%2732%27%20height=%2732%27/%3e"/></span><img alt="youtube" src="data:image/gif;base64,R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAAALAAAAAABAAEAAAIBRAA7" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%"/><noscript><img alt="youtube" srcset="https://tryolabs.imgix.net/logos/youtube.svg?auto=format&amp;fit=max&amp;w=32 1x, https://tryolabs.imgix.net/logos/youtube.svg?auto=format&amp;fit=max&amp;w=64 2x" src="https://tryolabs.imgix.net/logos/youtube.svg?auto=format&amp;fit=max&amp;w=64" decoding="async" data-nimg="intrinsic" style="top:0;left:0;bottom:0;right:0;box-sizing:border-box;padding:0;border:none;margin:auto;display:block;width:0;height:0;min-width:100%;max-width:100%;min-height:100%;max-height:100%" loading="lazy"/></noscript></span></a></div></nav></div><div class="css-1du409g e1pm0t60"><p class="MuiTypography-root MuiTypography-body1 css-hd5z0k">© 2024. All rights reserved.</p></div></div></div></div></div></section></div><div class="css-y9g46x e3s69xv0">This website uses cookies to improve user experience. <a target="_blank" href="/terms">Read more</a><div><button class="MuiButtonBase-root MuiButton-root MuiButton-contained MuiButton-containedPrimary MuiButton-sizeSmall MuiButton-containedSizeSmall MuiButton-disableElevation MuiButton-root MuiButton-contained MuiButton-containedPrimary MuiButton-sizeSmall MuiButton-containedSizeSmall MuiButton-disableElevation css-1jukobd" tabindex="0" type="button">Accept</button><button class="MuiButtonBase-root MuiButton-root MuiButton-outlined MuiButton-outlinedPrimary MuiButton-sizeSmall MuiButton-outlinedSizeSmall MuiButton-disableElevation MuiButton-root MuiButton-outlined MuiButton-outlinedPrimary MuiButton-sizeSmall MuiButton-outlinedSizeSmall MuiButton-disableElevation css-u71hyd" tabindex="0" type="button">Decline</button></div></div></div></body>