Upload bcpt-sft_config.yaml with huggingface_hub
Browse files- bcpt-sft_config.yaml +51 -0
bcpt-sft_config.yaml
ADDED
|
@@ -0,0 +1,51 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
model_config_path: /home/athuser/luxi-files/kotodama/configs/model.yaml
|
| 2 |
+
model_config_section: proxy
|
| 3 |
+
attn_res: true
|
| 4 |
+
attn_res_boundaries:
|
| 5 |
+
- 0
|
| 6 |
+
- 3
|
| 7 |
+
- 7
|
| 8 |
+
- 12
|
| 9 |
+
- 21
|
| 10 |
+
- 25
|
| 11 |
+
dataset: pipeline/formatted/tokenized/train
|
| 12 |
+
eval_dataset: pipeline/formatted/tokenized/eval
|
| 13 |
+
packed: true
|
| 14 |
+
max_seq_len: 4096
|
| 15 |
+
batch_size: 4
|
| 16 |
+
gradient_accumulation: 1
|
| 17 |
+
max_steps: -1
|
| 18 |
+
bf16: true
|
| 19 |
+
max_grad_norm: 1.0
|
| 20 |
+
muon_momentum: 0.95
|
| 21 |
+
muon_weight_decay: 0.01
|
| 22 |
+
muon_ns_iterations: 5
|
| 23 |
+
muon_ns_coefficients: gram_ns
|
| 24 |
+
adamw_betas:
|
| 25 |
+
- 0.9
|
| 26 |
+
- 0.95
|
| 27 |
+
adamw_weight_decay: 0.1
|
| 28 |
+
warmup_ratio: 0.05
|
| 29 |
+
wsd_decay_start: 1.0
|
| 30 |
+
wsd_decay_type: sqrt
|
| 31 |
+
logging_steps: 10
|
| 32 |
+
grad_analysis_every: 10
|
| 33 |
+
weight_drift_every: 10
|
| 34 |
+
geo_tier1_every: 10
|
| 35 |
+
sample_every: 50
|
| 36 |
+
save_every: 25
|
| 37 |
+
eval_steps: 25
|
| 38 |
+
checkpoint_keep: 3
|
| 39 |
+
async_save: true
|
| 40 |
+
checkpoint_compress: true
|
| 41 |
+
checkpoint_shm_dir: /dev/shm/luxia-sft-ckpts
|
| 42 |
+
wandb_project: kotodama-sft-sweep
|
| 43 |
+
wandb_entity: aethera
|
| 44 |
+
num_workers: 4
|
| 45 |
+
pretrained_checkpoint: /home/athuser/luxi-files/kotodama/checkpoints/books-cpt/step_00017336.pt.zst
|
| 46 |
+
muon_lr: 0.003
|
| 47 |
+
adamw_lr: 0.00030000000000000003
|
| 48 |
+
num_epochs: 2
|
| 49 |
+
wandb_run_name: sweep-books-cpt-lr3e-03-ep2
|
| 50 |
+
output_dir: outputs/sweep/sweep-books-cpt-lr3e-03-ep2
|
| 51 |
+
checkpoint_dir: outputs/sweep/sweep-books-cpt-lr3e-03-ep2/checkpoints
|