Claudio Fanconi
@fanconic.bsky.social
Please insert pretentious things about myself here:
PhD student in Machine Learning @ University of Cambridge | Previously MSc & Bsc @ ETH Zürich
fanconic.github.io
PhD student in Machine Learning @ University of Cambridge | Previously MSc & Bsc @ ETH Zürich
fanconic.github.io
This work was done in collaboration with researchers at the University of Oxford (FLAIR Lab) and Sakana AI.
Sakana AI wrote a detailed blog post about our paper here:
sakana.ai/llm-squared/
Sakana AI wrote a detailed blog post about our paper here:
sakana.ai/llm-squared/
Sakana AI
Can LLMs invent better ways to train LLMs?
sakana.ai
November 25, 2024 at 2:05 PM
This work was done in collaboration with researchers at the University of Oxford (FLAIR Lab) and Sakana AI.
Sakana AI wrote a detailed blog post about our paper here:
sakana.ai/llm-squared/
Sakana AI wrote a detailed blog post about our paper here:
sakana.ai/llm-squared/
This seems to be in line with the Superficial Alignment Hypothesis from LIMA (arxiv.org/pdf/2305.11206) and URIAL (arxiv.org/pdf/2312.01552).
On the other hand, I found this technical report interesting, challenging the hypothesis:
arxiv.org/pdf/2410.03717
Might be interesting other point of view
On the other hand, I found this technical report interesting, challenging the hypothesis:
arxiv.org/pdf/2410.03717
Might be interesting other point of view
arxiv.org
November 22, 2024 at 1:28 AM
This seems to be in line with the Superficial Alignment Hypothesis from LIMA (arxiv.org/pdf/2305.11206) and URIAL (arxiv.org/pdf/2312.01552).
On the other hand, I found this technical report interesting, challenging the hypothesis:
arxiv.org/pdf/2410.03717
Might be interesting other point of view
On the other hand, I found this technical report interesting, challenging the hypothesis:
arxiv.org/pdf/2410.03717
Might be interesting other point of view