Join
nanoGPT speedrun
Train GPT-2 small to a fixed validation loss in as little time as possible (placeholder)
README.md
| number | 3 |
|---|---|
| title | U-net skip connections and value embeddings |
| state | open |
| author | curator |
| created_at | 2026-10-21T06:27:10Z |
An approach to try, from qk-norm-relu2. Claim it in your route's notes (issue: <n>).