Delete meta_000100.json
Browse files- meta_000100.json +0 -43
meta_000100.json
DELETED
|
@@ -1,43 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"step": 100,
|
| 3 |
-
"val_bpb": 0.3895867845485602,
|
| 4 |
-
"model_config": {
|
| 5 |
-
"sequence_len": 8196,
|
| 6 |
-
"vocab_size": 32768,
|
| 7 |
-
"n_layer": 12,
|
| 8 |
-
"n_head": 6,
|
| 9 |
-
"n_kv_head": 6,
|
| 10 |
-
"n_embd": 768,
|
| 11 |
-
"window_pattern": "SSSL"
|
| 12 |
-
},
|
| 13 |
-
"user_config": {
|
| 14 |
-
"run": "dummy",
|
| 15 |
-
"device_type": "",
|
| 16 |
-
"model_tag": "simple_20261101_toolmix",
|
| 17 |
-
"model_step": null,
|
| 18 |
-
"load_source": "sft",
|
| 19 |
-
"output_tag": "simple_20261101_smoltalk_refine",
|
| 20 |
-
"load_optimizer": 1,
|
| 21 |
-
"num_iterations": 100,
|
| 22 |
-
"max_seq_len": 8196,
|
| 23 |
-
"device_batch_size": 1,
|
| 24 |
-
"total_batch_size": 524544,
|
| 25 |
-
"embedding_lr": 0.3,
|
| 26 |
-
"unembedding_lr": 0.008,
|
| 27 |
-
"matrix_lr": 0.02,
|
| 28 |
-
"init_lr_frac": 0.8,
|
| 29 |
-
"warmup_ratio": 0.0,
|
| 30 |
-
"warmdown_ratio": 0.5,
|
| 31 |
-
"final_lr_frac": 0.0,
|
| 32 |
-
"eval_every": 200,
|
| 33 |
-
"eval_tokens": 81960,
|
| 34 |
-
"chatcore_every": -1,
|
| 35 |
-
"chatcore_max_cat": -1,
|
| 36 |
-
"chatcore_max_sample": 24,
|
| 37 |
-
"mmlu_epochs": 3,
|
| 38 |
-
"gsm8k_epochs": 4,
|
| 39 |
-
"chat_only": true,
|
| 40 |
-
"tool_mix": false,
|
| 41 |
-
"smoltalk_partition": "unused_after_tool_mix"
|
| 42 |
-
}
|
| 43 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|