Maximilian Mozes
@maximilianmozes.bsky.social
Senior Research Scientist at Cohere. PhD at UCL. He/him.
Reposted by Maximilian Mozes
Do LLMs need rationales for learning from mistakes? 🤔
When LLMs learn from previous incorrect answers, they typically observe corrective feedback in the form of rationales explaining each mistake. In our new preprint, we find these rationales do not help, in fact they hurt performance!
🧵
When LLMs learn from previous incorrect answers, they typically observe corrective feedback in the form of rationales explaining each mistake. In our new preprint, we find these rationales do not help, in fact they hurt performance!
🧵
February 13, 2025 at 3:38 PM
Do LLMs need rationales for learning from mistakes? 🤔
When LLMs learn from previous incorrect answers, they typically observe corrective feedback in the form of rationales explaining each mistake. In our new preprint, we find these rationales do not help, in fact they hurt performance!
🧵
When LLMs learn from previous incorrect answers, they typically observe corrective feedback in the form of rationales explaining each mistake. In our new preprint, we find these rationales do not help, in fact they hurt performance!
🧵
New preprint out! Thrilled to share our new work led by @lisaalaz.bsky.social
Do LLMs need rationales for learning from mistakes? 🤔
When LLMs learn from previous incorrect answers, they typically observe corrective feedback in the form of rationales explaining each mistake. In our new preprint, we find these rationales do not help, in fact they hurt performance!
🧵
When LLMs learn from previous incorrect answers, they typically observe corrective feedback in the form of rationales explaining each mistake. In our new preprint, we find these rationales do not help, in fact they hurt performance!
🧵
February 13, 2025 at 6:05 PM
New preprint out! Thrilled to share our new work led by @lisaalaz.bsky.social
Reposted by Maximilian Mozes
Check out @lisaalaz.bsky.social's internship work with us @cohere.com questioning the rationale behind rationales 🔥
February 13, 2025 at 4:18 PM
Check out @lisaalaz.bsky.social's internship work with us @cohere.com questioning the rationale behind rationales 🔥