Tag
MIT open-sourced the RLM (Recursive Language Models) inference library, which handles ultra-long contexts by having the model recursively call itself programmatically, solving the limited context window problem of traditional models.