The koan of an open-source LLM

<p>A proposal for a new definition of an "open source" LLM and why no definition will ever just work.<br />This is AI generated audio with Python and 11Labs. Music generated by Meta's MusicGen.<br />Source code: <a class="linkified" href="https://github.com/natolambert/interconnects-tools" target="_blank">https://github.com/natolambert/interconnects-tools</a><br />Original post: <a class="linkified" href="https://www.interconnects.ai/p/an-open-source-llm" target="_blank">https://www.interconnects.ai/p/an-open-source-llm</a></p><p>00:00 The koan of an open-source LLM<br />03:22 A new naming scheme for open LLMs<br />07:09 Pivot points and politics<br />08:16 Claude 3, arms race, commoditization, and national security<br />10:01 Doomers debunking bio risks of LLMs themselves<br />11:21 Mistral's perceived reversal and the EU<br />13:22 Messy points: Transparency, safety, and copyright<br />13:32 The muddling of transparency<br />15:22 The muddling of "safety"<br />16:30 The muddling of licenses and copyright<br />20:12 Vibes points and next steps</p><p>Figure 1: <a class="linkified" href="https://huggingface.co/datasets/natolambert/interconnects-figures/resolve/main/open-source/img_046.png" target="_blank">https://huggingface.co/datasets/natolambert/interconnects-figures/resolve/main/open-source/img_046.png</a><br />Figure 2: <a class="linkified" href="https://huggingface.co/datasets/natolambert/interconnects-figures/resolve/main/open-source/img_064.png" target="_blank">https://huggingface.co/datasets/natolambert/interconnects-figures/resolve/main/open-source/img_064.png</a></p> <br /><br />This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit <a href="https://www.interconnects.ai/subscribe?utm_medium=podcast&#38;utm_campaign=CTA_2">www.interconnects.ai/subscribe</a>