ChrisMcCormick commited on
Commit
1a071dc
·
verified ·
1 Parent(s): 4f60492

Tool-free SFT of d24_decoderstack step 5568 (462 steps, val bpb 0.2732, ChatCORE 0.2527); model + meta + w1 optim shard

Browse files
chatsft_checkpoints/d24_decoderstack_w1/meta_000462.json ADDED
@@ -0,0 +1,38 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "step": 462,
3
+ "val_bpb": 0.2732188042985377,
4
+ "model_config": {
5
+ "sequence_len": 2048,
6
+ "vocab_size": 32768,
7
+ "n_layer": 24,
8
+ "n_head": 12,
9
+ "n_kv_head": 12,
10
+ "n_embd": 1536,
11
+ "window_pattern": "SSSL"
12
+ },
13
+ "user_config": {
14
+ "run": "sft-d24ds-notools-1xh100",
15
+ "device_type": "",
16
+ "model_tag": "d24_decoderstack_w1",
17
+ "model_step": 5568,
18
+ "load_optimizer": 1,
19
+ "num_iterations": -1,
20
+ "max_seq_len": null,
21
+ "device_batch_size": 16,
22
+ "total_batch_size": null,
23
+ "embedding_lr": null,
24
+ "unembedding_lr": null,
25
+ "matrix_lr": null,
26
+ "init_lr_frac": 0.8,
27
+ "warmup_ratio": 0.0,
28
+ "warmdown_ratio": 0.5,
29
+ "final_lr_frac": 0.0,
30
+ "eval_every": 50,
31
+ "eval_tokens": 20971520,
32
+ "chatcore_every": 100,
33
+ "chatcore_max_cat": -1,
34
+ "chatcore_max_sample": 24,
35
+ "mmlu_epochs": 3,
36
+ "gsm8k_epochs": 4
37
+ }
38
+ }
chatsft_checkpoints/d24_decoderstack_w1/model_000462.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8567df1f60c193469cab6077b0b037faaa2553a816abc27f866d9ae6a030a443
3
+ size 4227933321
chatsft_checkpoints/d24_decoderstack_w1/optim_000462_rank0.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:14037e5ab375ea9af752654cb09db2dc3225c315466fdb7adc3199844c7cf45c
3
+ size 5739601689