| # h4_fineweb_edu_22x320 — H4: deep-narrow 32M (22 layers × 320) trained from scratch on FineWeb-Edu |
| |
| **Why it exists:** tests whether a deeper, narrower shape at the same parameter count improves grammatical knowledge (BLiMP) without losing ARC-Easy. It uses the same data pool, recipe and token budget as Z1 arm Z (`z1_Z_fineweb_edu/`, 15 layers × 384); only the model shape differs. |
|
|
| ## Setup |
|
|
| - **Model:** 31.0M parameters, 22 layers, d_model 320, 5 heads (head dim 64), RoPE, SwiGLU, RMSNorm, QK-norm, value residual; BPE tokenizer with 12,288 tokens (`tokenizer.json` in the repository root). |
| - **Recipe:** identical to `z1_Z_fineweb_edu/`: Muon (hidden 2-D weights) + AdamW, learning rate 6e-4 with 2,000 warmup steps and cosine decay to 6e-5, batch 32 × 1024 tokens, 150,000 steps (4.9B tokens), seed 1337. Training windows are drawn without replacement (each window at most once), less than one epoch. |
| - **Data:** the same 5.0B-token FineWeb-Edu pool as `z1_Z_fineweb_edu/` (ODC-BY; three sources deduplicated by document hash), scanned against WikiText-2 and ARC test/validation with normalized 13-gram and short-question matching; matching documents were removed. |
| - **Checkpoints:** every 30,000 steps (about 1B tokens); the last one is step 150,000. |
| - **Status:** research checkpoints, **not a leaderboard submission**. The comparison is with Z1 arm Z (seeds 1337 and 1338); the architecture and trainer differ from our published 32M, so their numbers are not directly comparable. |
| - **Format:** PyTorch checkpoint dict with `model`, `opt`, `step`, `config`; `train_gpt_ref.py` in the repository root rebuilds the model from `config`. |
|
|
