Upload 5 files
Browse files- meta_001679.json +59 -0
- model_001679.pt +3 -0
- optim_001679_rank0.pt +3 -0
- token_bytes.pt +3 -0
- tokenizer.pkl +3 -0
meta_001679.json
ADDED
|
@@ -0,0 +1,59 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"step": 1679,
|
| 3 |
+
"val_bpb": 0.33559814257945075,
|
| 4 |
+
"model_config": {
|
| 5 |
+
"sequence_len": 8196,
|
| 6 |
+
"vocab_size": 32768,
|
| 7 |
+
"n_layer": 12,
|
| 8 |
+
"n_head": 6,
|
| 9 |
+
"n_kv_head": 6,
|
| 10 |
+
"n_embd": 768,
|
| 11 |
+
"window_pattern": "SSSL"
|
| 12 |
+
},
|
| 13 |
+
"user_config": {
|
| 14 |
+
"run": "dummy",
|
| 15 |
+
"device_type": "",
|
| 16 |
+
"fp8": false,
|
| 17 |
+
"fp8_recipe": "tensorwise",
|
| 18 |
+
"depth": 12,
|
| 19 |
+
"aspect_ratio": 64,
|
| 20 |
+
"head_dim": 128,
|
| 21 |
+
"max_seq_len": 8196,
|
| 22 |
+
"window_pattern": "SSSL",
|
| 23 |
+
"ordered_data": true,
|
| 24 |
+
"num_iterations": -1,
|
| 25 |
+
"target_flops": -1.0,
|
| 26 |
+
"target_param_data_ratio": 8.0,
|
| 27 |
+
"device_batch_size": 1,
|
| 28 |
+
"total_batch_size": 524544,
|
| 29 |
+
"embedding_lr": 0.3,
|
| 30 |
+
"unembedding_lr": 0.008,
|
| 31 |
+
"weight_decay": 0.28,
|
| 32 |
+
"matrix_lr": 0.02,
|
| 33 |
+
"scalar_lr": 0.5,
|
| 34 |
+
"warmup_steps": 40,
|
| 35 |
+
"warmdown_ratio": 0.65,
|
| 36 |
+
"final_lr_frac": 0.05,
|
| 37 |
+
"resume_from_step": -1,
|
| 38 |
+
"eval_every": 250,
|
| 39 |
+
"eval_tokens": 81960,
|
| 40 |
+
"core_metric_every": -1,
|
| 41 |
+
"core_metric_max_per_task": 500,
|
| 42 |
+
"sample_every": -1,
|
| 43 |
+
"save_every": -1,
|
| 44 |
+
"model_tag": "simple_20261101_curriculum"
|
| 45 |
+
},
|
| 46 |
+
"device_batch_size": 1,
|
| 47 |
+
"max_seq_len": 8196,
|
| 48 |
+
"total_batch_size": 524544,
|
| 49 |
+
"dataloader_state_dict": {
|
| 50 |
+
"pq_idx": 1,
|
| 51 |
+
"rg_idx": 179,
|
| 52 |
+
"epoch": 1
|
| 53 |
+
},
|
| 54 |
+
"loop_state": {
|
| 55 |
+
"min_val_bpb": 0.33149468985271474,
|
| 56 |
+
"smooth_train_loss": 0.6637563544034941,
|
| 57 |
+
"total_training_time": 27801.5051074028
|
| 58 |
+
}
|
| 59 |
+
}
|
model_001679.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:36f82f6baf494f82d913018e209d0f685d058a176ae26e9b263e7b31d762c23e
|
| 3 |
+
size 792761399
|
optim_001679_rank0.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4859ad7f0ff4f1edafacc6c74f6a7b284373a7e5b085841e492967a1c37241f4
|
| 3 |
+
size 1246165237
|
token_bytes.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a6cc922eaf6548da806759457b823c758ae418a078d7a7a2cf2836c595288db2
|
| 3 |
+
size 132649
|
tokenizer.pkl
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0c87a69830b1a3e1699872ff6392fef68ab5eae802b1311376dbb59e5d3c1134
|
| 3 |
+
size 408048
|