Reposted by Rasool FakoorTianwei Ni @twni2016.bsky.social · 23/04/2025Can we make LLMs reason effectively without a huge inference time cost? We show a powerful approach through learning and forgetting! Our recipe: ⬇️ 142
Reposted by Rasool FakoorTianwei Ni @twni2016.bsky.social · 23/04/2025Work completed during my internship at Amazon Science. Thank you to my co-authors @allenanie.bsky.social, Sapana Chaudhary, Yao Liu, Huzefa Rangwala, @rasoolfa.bsky.social! 021
Rasool Fakoor @rasoolfa.bsky.social · 31/01/2025arxiv.org/abs/2410.14655arxiv.orgBridging the Training-Inference Gap in LLMs by Leveraging Self-Generated TokensLanguage models are often trained to maximize the likelihood of the next token given past tokens in the training dataset. However, during inference time, they are utilized differently, generating text... 010
Rasool Fakoor @rasoolfa.bsky.social · 31/01/2025You can check out our paper that includes citations for teacher, professor, and student forcing. 110
