Guillaume Lajoie

Context compaction in agentic RL can run up to 5x faster simply by streaming the KV cache instead of flushing it—accidentally turning standard LLMs into recurrent agents that preserve evicted context purely through RL.

Foundation model agents can achieve stable cooperation in social dilemmas by inferring behavioral similarities, defying classical game theory's expectations of mutual defection.

MOJO outperforms traditional supervised learning models by effectively utilizing unlabelled data, achieving state-of-the-art results in neural decoding tasks with minimal labelled input.

Control interventions are often detected by LLMs, with awareness levels varying significantly across models and tasks, revealing vulnerabilities in AI safety protocols.

Offline policy optimization with a world model allows for affective music recommendation that improves user valence and arousal, even when ethical constraints preclude online experimentation.