Upload hyperparameters.yaml with huggingface_hub
Browse files- hyperparameters.yaml +37 -0
hyperparameters.yaml
ADDED
|
@@ -0,0 +1,37 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
model_name: pythia-1b
|
| 2 |
+
out_dir: /data/users/zichunyu/out/pythia-1b/fineweb/sample-100BT
|
| 3 |
+
resume: false
|
| 4 |
+
data:
|
| 5 |
+
class_path: litgpt.data.FineWeb
|
| 6 |
+
init_args:
|
| 7 |
+
data_path: /data/users/zichunyu/data/fineweb/sample-100BT
|
| 8 |
+
val_split_fraction: 0.0005
|
| 9 |
+
seed: 42
|
| 10 |
+
num_workers: 8
|
| 11 |
+
train:
|
| 12 |
+
save_interval: 5000
|
| 13 |
+
log_interval: 50
|
| 14 |
+
global_batch_size: 512
|
| 15 |
+
micro_batch_size: 16
|
| 16 |
+
lr_warmup_steps: 2000
|
| 17 |
+
max_tokens: 50000000000
|
| 18 |
+
tie_embeddings: false
|
| 19 |
+
max_norm: 1.0
|
| 20 |
+
min_lr: 4.0e-05
|
| 21 |
+
eval:
|
| 22 |
+
interval: 200000
|
| 23 |
+
max_iters: 100
|
| 24 |
+
initial_validation: false
|
| 25 |
+
optimizer:
|
| 26 |
+
class_path: torch.optim.AdamW
|
| 27 |
+
init_args:
|
| 28 |
+
lr: 0.0004
|
| 29 |
+
weight_decay: 0.1
|
| 30 |
+
betas:
|
| 31 |
+
- 0.9
|
| 32 |
+
- 0.95
|
| 33 |
+
devices: auto
|
| 34 |
+
tokenizer_dir: checkpoints/EleutherAI/pythia-1b
|
| 35 |
+
logger_name: wandb
|
| 36 |
+
exp_name: pythia-1b_fineweb_sample-100BT
|
| 37 |
+
seed: 42
|