> For the complete documentation index, see [llms.txt](https://tarka-air.gitbook.io/home/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://tarka-air.gitbook.io/home/home.md).

# Home

<p align="center">Latest works</p>

<table data-card-size="large" data-view="cards"><thead><tr><th></th><th></th><th></th><th></th><th data-hidden data-card-target data-type="content-ref"></th><th data-hidden data-card-cover data-type="image">Cover image</th><th data-hidden data-type="image">Cover image (dark)</th><th data-hidden data-type="image">Cover image (dark)</th></tr></thead><tbody><tr><td><h3>Tarka Embedding 30M V1</h3></td><td><mark style="color:$info;">Release  |  7-1-2026</mark></td><td><mark style="color:$info;">We achieved 20× compression and recovered  ~86% MTEB performance. The model supports elastic upscaling at inference, avoiding full-weight loading and reducing both memory footprint and compute cost.</mark><br></td><td><a href="https://huggingface.co/Tarka-AIR/Tarka-Embedding-30M-V1"><kbd>Open - Source</kbd></a></td><td><a href="/pages/ABnCtumAP0cLOkzUkBNZ">/pages/ABnCtumAP0cLOkzUkBNZ</a></td><td><a href="/files/5JJNF8Sim67m21lsW3SZ">/files/5JJNF8Sim67m21lsW3SZ</a></td><td></td><td></td></tr><tr><td><h3>Reduce and Refine</h3></td><td><mark style="color:$info;">Release  |  1-12-2025</mark></td><td><mark style="color:$info;">This work explores model compression by progressively reducing a 28-layer model to a lean 6-layer model without major performance loss. Reduce and Refine demonstrates a practical path to faster, lighter, more efficient LLMs.</mark></td><td><a href="https://huggingface.co/collections/Tarka-AIR/tarka-embed-v1"><kbd>Open - Source</kbd></a></td><td><a href="/pages/gDCUWXEBdH0thvnOVOOP">/pages/gDCUWXEBdH0thvnOVOOP</a></td><td><a href="/files/x3SCsA1vqQegreFi3ffw">/files/x3SCsA1vqQegreFi3ffw</a></td><td></td><td></td></tr><tr><td><h3>Tarka Embedding V1</h3></td><td><mark style="color:$info;">Release  |  9-11-2025</mark></td><td><mark style="color:$info;">The Tarka Embedding V1 series is a compact and efficient text embedding model family developed to explore the capabilities of knowledge distillation, coreset selection, and model compression techniques</mark></td><td><a href="https://huggingface.co/collections/Tarka-AIR/tarka-embed-v1"><kbd>Open - Source</kbd></a></td><td><a href="/pages/VFUK5S6OyQxbSUyuLKdH">/pages/VFUK5S6OyQxbSUyuLKdH</a></td><td><a href="/files/3jaBPbkGB7OK7rkvTMPA">/files/3jaBPbkGB7OK7rkvTMPA</a></td><td><a href="/files/IFJo7Xc1q4CSdhasY8NL">/files/IFJo7Xc1q4CSdhasY8NL</a></td><td><a href="/files/ylqIqcnwvETTtkTyLQ8x">/files/ylqIqcnwvETTtkTyLQ8x</a></td></tr></tbody></table>