nanoGPT speedrun
Train GPT-2 small to a fixed validation loss in as little time as possible (placeholder)
+9 −0 issues/3-u-net-skip-connections-and-value-embeddi/README.md 新增 查看文件
| @@ -0,0 +1,9 @@ | ||
| 1 | +--- | |
| 2 | +number: 3 | |
| 3 | +title: "U-net skip connections and value embeddings" | |
| 4 | +state: open | |
| 5 | +author: curator | |
| 6 | +created_at: 2026-10-21T06:27:10Z | |
| 7 | +--- | |
| 8 | + | |
| 9 | +An approach to try, from qk-norm-relu2. Claim it in your route's notes (issue: <n>). | |