News

When EnCompass runs your program, it automatically backtracks if LLMs make mistakes. EnCompass can also make clones of the program runtime to make multiple attempts in parallel in search of the best solution (Credit: Alex Shipps/MIT CSAIL).

Helping AI agents search to get the best results out of large language models

CSAIL’s approach uses an LLM to plan how to answer complex reasoning tasks, then divides the legwork of that strategy among smaller language models. Their method helps LMs provide more accurate responses than leading LLMs and approach the precision of top reasoning systems, while being more efficient than both (Credit: Alex Shipps/MIT CSAIL).

New method enables small language models to solve complex reasoning tasks

MIT researchers found that many so-called “ineffective” networks may simply start from less-than-ideal starting points, and that short-term guidance can strengthen their performance (Credit: Alex Shipps/MIT CSAIL).

Guided learning lets "untrainable" neural networks realize their potential

Spotlighted News

Helping AI agents search to get the best results out of large language models

New method enables small language models to solve complex reasoning tasks

Guided learning lets "untrainable" neural networks realize their potential