Upload 4 files
Browse files
base_checkpoints_jp/d20/meta_021400.json
ADDED
|
@@ -0,0 +1,32 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"step": 21400,
|
| 3 |
+
"val_bpb": 0.6682459571973981,
|
| 4 |
+
"model_config": {
|
| 5 |
+
"sequence_len": 2048,
|
| 6 |
+
"vocab_size": 65536,
|
| 7 |
+
"n_layer": 20,
|
| 8 |
+
"n_head": 10,
|
| 9 |
+
"n_kv_head": 10,
|
| 10 |
+
"n_embd": 1280
|
| 11 |
+
},
|
| 12 |
+
"user_config": {
|
| 13 |
+
"run": "d20-jp-1760620493",
|
| 14 |
+
"depth": 20,
|
| 15 |
+
"max_seq_len": 2048,
|
| 16 |
+
"target_param_data_ratio": 20,
|
| 17 |
+
"num_iterations": -1,
|
| 18 |
+
"device_batch_size": 32,
|
| 19 |
+
"total_batch_size": 524288,
|
| 20 |
+
"embedding_lr": 0.2,
|
| 21 |
+
"unembedding_lr": 0.004,
|
| 22 |
+
"matrix_lr": 0.02,
|
| 23 |
+
"weight_decay": 0.0,
|
| 24 |
+
"eval_every": 250,
|
| 25 |
+
"eval_tokens": 10485760,
|
| 26 |
+
"DATASET_REPO_ID": "kajuma/ABEJA-CC-JA-edu",
|
| 27 |
+
"CONFIG_NAME": "10%",
|
| 28 |
+
"SPLIT": "train",
|
| 29 |
+
"TOTAL_SHARDS": 378,
|
| 30 |
+
"DOWNLOAD_CACHE_DIR": "download_cache_jp"
|
| 31 |
+
}
|
| 32 |
+
}
|
base_checkpoints_jp/d20/model_021400.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:15d7a38c13b2002334a00947d32efda8c494e0332c43e6139ef67de9cb9db1f5
|
| 3 |
+
size 2076230219
|
tokenizer/token_bytes.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:58f4caab8676cbcb4897476c8849412dadccaea1b650b987d49be2ff946ab3c9
|
| 3 |
+
size 263721
|
tokenizer/tokenizer.pkl
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:26d1385fb4408ba4ba4e53f957d1723e5ebd3b7a60f9d049039d4cfed0047f01
|
| 3 |
+
size 1051423
|