Upload folder using huggingface_hub
Browse files- meta_021400.json +34 -0
- model_021400.pt +3 -0
- optim_021400.pt +3 -0
meta_021400.json
ADDED
|
@@ -0,0 +1,34 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"val_bpb": 0.8119751250726281,
|
| 3 |
+
"model_config": {
|
| 4 |
+
"sequence_len": 2048,
|
| 5 |
+
"vocab_size": 65536,
|
| 6 |
+
"n_layer": 20,
|
| 7 |
+
"n_head": 10,
|
| 8 |
+
"n_kv_head": 10,
|
| 9 |
+
"n_embd": 1280
|
| 10 |
+
},
|
| 11 |
+
"user_config": {
|
| 12 |
+
"run": "dummy",
|
| 13 |
+
"depth": 20,
|
| 14 |
+
"max_seq_len": 2048,
|
| 15 |
+
"num_iterations": -1,
|
| 16 |
+
"target_flops": -1.0,
|
| 17 |
+
"target_param_data_ratio": 20,
|
| 18 |
+
"device_batch_size": 32,
|
| 19 |
+
"total_batch_size": 524288,
|
| 20 |
+
"embedding_lr": 0.2,
|
| 21 |
+
"unembedding_lr": 0.004,
|
| 22 |
+
"weight_decay": 0.0,
|
| 23 |
+
"matrix_lr": 0.02,
|
| 24 |
+
"grad_clip": 1.0,
|
| 25 |
+
"eval_every": 250,
|
| 26 |
+
"eval_tokens": 10485760,
|
| 27 |
+
"core_metric_every": 2000,
|
| 28 |
+
"core_metric_max_per_task": 500,
|
| 29 |
+
"sample_every": 2000,
|
| 30 |
+
"model_tag": ""
|
| 31 |
+
},
|
| 32 |
+
"device_batch_size": 32,
|
| 33 |
+
"max_seq_len": 2048
|
| 34 |
+
}
|
model_021400.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:574670de45c667a98fe72c2d145f24007ecc1778721e2481e41500842dd3ba13
|
| 3 |
+
size 2076230219
|
optim_021400.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:17e3bf898baa87fd10ed053e1f4fed14b5366c33f08119d2b973afba0a488897
|
| 3 |
+
size 407642867
|