refactor: keep muon_adamw as default optimizer and drop nora docs

- revert CLI/create_optimizer/display defaults to muon_adamw
- revert README, README-zh-CN, params.md to pre-merge state
This commit is contained in:
2026-08-01 07:51:51 +08:00
parent ffbd9b57c9
commit 25c9e81b2b
4 changed files with 24 additions and 29 deletions
+18 -19
View File
@@ -25,35 +25,36 @@
| Parameter | Description | Default |
|-----------|-------------|---------|
| `--warmup_ratio` | Fraction of total steps used for LR warmup | 0.05 |
| `--max_lr` | NAdamW learning rate; schedulers scale every optimizer group proportionally | 3e-4 |
| `--max_lr` | Maximum learning rate (cosine decay after warmup) | 3e-4 |
| `--max_grad_norm` | Maximum gradient norm for clipping (None disables) | 1.0 |
### Optimizer
The default `nora_nadamw` optimizer sends internal `Linear.weight` matrices to
**Nora** and embeddings, the LM head, norms, biases, LoRA factors, and fallback
parameters to **NAdamW**. Parameters are classified by module role and identity,
so tied embedding/head weights occur in exactly one group. Nora requires complete
rows under DTensor sharding and rejects layouts sharded along the last dimension.
| Parameter | Description | Default |
|-----------|-------------|---------|
| `--optimizer` | Built-in optimizer (`nora_nadamw`, `muon_adamw`) | `nora_nadamw` |
| `--weight_decay` | NAdamW decay for eligible fallback parameters; known embeddings, heads, norms, biases, and LoRA factors use 0 | 0.1 |
| `--nora_lr` | Nora learning rate | 5e-3 |
| `--nora_beta` | Nora momentum-buffer EMA factor | 0.95 |
| `--nora_momentum` | Nora Nesterov interpolation factor | 0.95 |
| `--nora_weight_decay` | Nora matrix weight decay | 0.0 |
`muon_adamw` preserves the previous MuonMix behavior and the following options:
The default `muon_adamw` optimizer sends matrix parameters through **Muon** and
non-matrix parameters through **AdamW** (`fused=True`).
| Parameter | Description | Default |
|-----------|-------------|---------|
| `--optimizer` | Built-in optimizer (`muon_adamw`, `nora_nadamw`) | `muon_adamw` |
| `--weight_decay` | Weight decay (applied to Muon matrix params; non-matrix use 0) | 0.1 |
| `--muon_momentum` | Muon momentum factor | 0.95 |
| `--muon_nesterov` | Enable Nesterov momentum for Muon | True |
| `--muon_ns_steps` | Newton-Schulz iteration steps for Muon | 5 |
| `--muon_adjust_lr` | Muon LR adjustment strategy (`original`, `match_rms_adamw`) | `match_rms_adamw` |
`nora_nadamw` routes internal `Linear.weight` matrices to **Nora** and
embeddings, the LM head, norms, biases, LoRA factors, and fallback parameters to
**NAdamW**. Parameters are classified by module role and identity, so tied
embedding/head weights occur in exactly one group. Nora requires complete rows
under DTensor sharding and rejects layouts sharded along the last dimension.
| Parameter | Description | Default |
|-----------|-------------|---------|
| `--nora_lr` | Nora learning rate | 5e-3 |
| `--nora_beta` | Nora momentum-buffer EMA factor | 0.95 |
| `--nora_momentum` | Nora Nesterov interpolation factor | 0.95 |
| `--nora_weight_decay` | Nora matrix weight decay | 0.0 |
Optimizer identity and hyperparameters are saved in checkpoint metadata. Optimizer
states are intentionally not interchangeable: resume older MuonMix checkpoints
with `--optimizer=muon_adamw`.
@@ -159,9 +160,7 @@ nohup python scripts/tools/train.py \
--batch_per_device=4 \
--grad_accum_steps=8 \
--warmup_ratio=0.05 \
--optimizer=nora_nadamw \
--max_lr=1e-4 \
--nora_lr=5e-3 \
--max_grad_norm=1.0 \
--weight_decay=0.1 \
--window_size=2048 \