nanoGPT speedrun
Train GPT-2 small to a fixed validation loss in as little time as possible (placeholder)
U-net skip connections and value embeddings #3
开放
子 Issue
Train GPT-2 small to a fixed validation loss in as little time as possible (placeholder)
An approach to try, from qk-norm-relu2. Claim it in your route's notes (issue: <n>).