[CLOSE — superseded by KakeyaLattice] PR-R1d-β (research, archived): retrieval-aux loss + attention-localization instrumentation#67

FluffyAIcode wants to merge 3 commits into

AgentMemory/v04-pr-r1c-cross-attn-toy-cap-train-8e7fFluffyAIcode/Kakeya-LLM-Inference-engine:AgentMemory/v04-pr-r1c-cross-attn-toy-cap-train-8e7ffrom

AgentMemory/v04-pr-r1d-cross-attn-retrieval-aux-8e7fFluffyAIcode/Kakeya-LLM-Inference-engine:AgentMemory/v04-pr-r1d-cross-attn-retrieval-aux-8e7fCopy head branch name to clipboard