code1phoenix commited on
Commit
6abf3d9
·
verified ·
1 Parent(s): 805637d

Upload folder using huggingface_hub

Browse files
midtraining_v3/meta_000811.json ADDED
@@ -0,0 +1,29 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "step": 811,
3
+ "val_bpb": 0.4384961627580738,
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": "dummy",
14
+ "device_type": "",
15
+ "dtype": "bfloat16",
16
+ "num_iterations": -1,
17
+ "max_seq_len": 2048,
18
+ "device_batch_size": 32,
19
+ "unembedding_lr": 0.004,
20
+ "embedding_lr": 0.2,
21
+ "matrix_lr": 0.02,
22
+ "init_lr_frac": 1.0,
23
+ "weight_decay": 0.0,
24
+ "eval_every": 150,
25
+ "eval_tokens": 10485760,
26
+ "total_batch_size": 524288,
27
+ "dry_run": 0
28
+ }
29
+ }
midtraining_v3/model_000811.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6bb04e1ec9d05aaf1b42dc52a721c281ea636310e4a48f8a95f234808eb9959d
3
+ size 2076230219
midtraining_v3/optim_000811_rank0.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:572561d87cadfea55dfea5e67721f023407c2930f1d7a21b65609044c5616a70
3
+ size 1348753485