[Article Voiceover] Reverse engineering OpenAI's o1
September 17, 2024
18 min
49 of 156

September 17, 2024
[Article Voiceover] Reverse engineering OpenAI's o1
0:00 / 18:51
Excerpt playback isn’t available for this episode — use the full-episode link.
Search the Record
Moment search isn’t open for this episode yet
When it opens, you’ll be able to type any phrase and jump straight to the second it was said.
Nathan Lambert Episodes Around September 17, 2024
Episode 49 of 156 — walk the feed in the order it was published.
18:51This episode[Article Voiceover] Reverse engineering OpenAI's o1
Summary
<p>What productionizing test-time compute shows us about the future of AI. Exploration has landed in language model training.<br />This is AI generated audio with Python and 11Labs.<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/reverse-engineering-openai-o1" target="_blank">https://www.interconnects.ai/p/reverse-engineering-openai-o1</a></p><p>00:00 Reverse engineering OpenAI's o1<br />01:52 From Q-star to Strawberry to o1<br />05:13 Training o1 with reinforcement learning<br />09:24 What is o1 doing when given a prompt?<br />11:49 Questions to consider to understand o1's structure<br />11:56 1. How does an RL-trained language model act?<br />12:38 2. Is it an online / test-time search?<br />14:20 3. Is it one model at inference?<br />15:29 Open-source o1, the future of o1, and the future of AI</p><p>Fig 1: <a class="linkified" href="https://huggingface.co/datasets/natolambert/interconnects-figures/resolve/main/o1/img_014.png" target="_blank">https://huggingface.co/datasets/natolambert/interconnects-figures/resolve/main/o1/img_014.png</a><br />Fig 2: <a class="linkified" href="https://huggingface.co/datasets/natolambert/interconnects-figures/resolve/main/o1/img_016.png" target="_blank">https://huggingface.co/datasets/natolambert/interconnects-figures/resolve/main/o1/img_016.png</a><br />Fig 3: <a class="linkified" href="https://huggingface.co/datasets/natolambert/interconnects-figures/resolve/main/o1/img_018.png" target="_blank">https://huggingface.co/datasets/natolambert/interconnects-figures/resolve/main/o1/img_018.png</a><br />Fig 4: <a class="linkified" href="https://huggingface.co/datasets/natolambert/interconnects-figures/resolve/main/o1/img_020.png" target="_blank">https://huggingface.co/datasets/natolambert/interconnects-figures/resolve/main/o1/img_020.png</a><br />Fig 5: <a class="linkified" href="https://huggingface.co/datasets/natolambert/interconnects-figures/resolve/main/o1/img_024.png" target="_blank">https://huggingface.co/datasets/natolambert/interconnects-figures/resolve/main/o1/img_024.png</a><br />Fig 6: <a class="linkified" href="https://huggingface.co/datasets/natolambert/interconnects-figures/resolve/main/o1/img_026.png" target="_blank">https://huggingface.co/datasets/natolambert/interconnects-figures/resolve/main/o1/img_026.png</a><br />Fig 7: <a class="linkified" href="https://huggingface.co/datasets/natolambert/interconnects-figures/resolve/main/o1/img_034.png" target="_blank">https://huggingface.co/datasets/natolambert/interconnects-figures/resolve/main/o1/img_034.png</a><br />Fig 8: <a class="linkified" href="https://huggingface.co/datasets/natolambert/interconnects-figures/resolve/main/o1/img_048.png" target="_blank">https://huggingface.co/datasets/natolambert/interconnects-figures/resolve/main/o1/img_048.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&utm_campaign=CTA_2">www.interconnects.ai/subscribe</a>





























