Upload checkpoint (depth=12, 1000 steps)
Browse files- checkpoint_d12/config.json +31 -0
- checkpoint_d12/params.pkl +3 -0
checkpoint_d12/config.json
ADDED
|
@@ -0,0 +1,31 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"num_shards": 50,
|
| 3 |
+
"hf_repo_id": "vorushin/tpuchat",
|
| 4 |
+
"depth": 12,
|
| 5 |
+
"aspect_ratio": 64,
|
| 6 |
+
"head_dim": 128,
|
| 7 |
+
"vocab_size": 32768,
|
| 8 |
+
"seq_len": 2048,
|
| 9 |
+
"window_pattern": "SSSL",
|
| 10 |
+
"softcap": 15.0,
|
| 11 |
+
"num_iterations": 1000,
|
| 12 |
+
"target_param_data_ratio": 10.5,
|
| 13 |
+
"device_batch_size": 2,
|
| 14 |
+
"grad_accum_steps": 8,
|
| 15 |
+
"total_batch_size": -1,
|
| 16 |
+
"max_chars_per_doc": 10000,
|
| 17 |
+
"learning_rate": 0.0003,
|
| 18 |
+
"weight_decay": 0.1,
|
| 19 |
+
"beta1": 0.9,
|
| 20 |
+
"beta2": 0.95,
|
| 21 |
+
"eps": 1e-08,
|
| 22 |
+
"warmup_ratio": 0.02,
|
| 23 |
+
"warmdown_ratio": 0.5,
|
| 24 |
+
"final_lr_frac": 0.0,
|
| 25 |
+
"eval_every": 100,
|
| 26 |
+
"eval_steps": 10,
|
| 27 |
+
"log_every": 10,
|
| 28 |
+
"save_every": -1,
|
| 29 |
+
"sample_every": 250,
|
| 30 |
+
"param_seed": 42
|
| 31 |
+
}
|
checkpoint_d12/params.pkl
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:537134a36b83ae6fd9e80932e69de632e5d391070eb372a98346a209fad4444d
|
| 3 |
+
size 1145578631
|