Research papers improving performance of LLMs [2/3]

Research papers published from January 16th to February 15th, 2025 proposing context length and architectural changes in LLMs

In partnership with

What’s in it today?

• ReLearn makes LLMs forget unwanted knowledge and remember how to speak good

• Coupled Adam fixes Adam so language model embeddings aren't too "extra"

• TransMLA converts GQA models to MLA ones for better LLM expression, because apparently size does matter

• LASP-2 makes linear attention training zoom by decluttering communic…