VLite7Mini41m Backup checkpoint-930 at global step 930
Browse files- checkpoint-930/config.json +16 -0
- checkpoint-930/generation_config.json +6 -0
- checkpoint-930/model.safetensors +3 -0
- checkpoint-930/optimizer.pt +3 -0
- checkpoint-930/rng_state.pth +3 -0
- checkpoint-930/scaler.pt +3 -0
- checkpoint-930/scheduler.pt +3 -0
- checkpoint-930/token_loss_weights_v40.pt +3 -0
- checkpoint-930/tokenizer.json +0 -0
- checkpoint-930/tokenizer_config.json +13 -0
- checkpoint-930/trainer_state.json +778 -0
- checkpoint-930/training_args.bin +3 -0
checkpoint-930/config.json
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"VLite7Mini41mForCausalLM"
|
| 4 |
+
],
|
| 5 |
+
"dtype": "float32",
|
| 6 |
+
"hidden_size": 512,
|
| 7 |
+
"initializer_range": 0.02,
|
| 8 |
+
"intermediate_size": 1280,
|
| 9 |
+
"max_position_embeddings": 512,
|
| 10 |
+
"model_type": "vlite7_mini_41m",
|
| 11 |
+
"num_attention_heads": 16,
|
| 12 |
+
"num_hidden_layers": 12,
|
| 13 |
+
"transformers_version": "5.10.2",
|
| 14 |
+
"use_cache": false,
|
| 15 |
+
"vocab_size": 32736
|
| 16 |
+
}
|
checkpoint-930/generation_config.json
ADDED
|
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_from_model_config": true,
|
| 3 |
+
"output_attentions": false,
|
| 4 |
+
"output_hidden_states": false,
|
| 5 |
+
"transformers_version": "5.10.2"
|
| 6 |
+
}
|
checkpoint-930/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2f744fc9dbb78d3b424f71c8ba3bfc2fd82c78fabaac99ce9ab63d43c67c43b1
|
| 3 |
+
size 259079288
|
checkpoint-930/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c89f80acd5d383a99687209b689eca6d25b8c145c93438fb8ab1927917bbdcaf
|
| 3 |
+
size 98163083
|
checkpoint-930/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:61c19bab1174704a4a4441475683bf1270277af15d2e2c95e964789128e482c4
|
| 3 |
+
size 14645
|
checkpoint-930/scaler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bb5a7413a574a3eefea9bdb74daf7569d45057766e0c5d15656bad4391508a41
|
| 3 |
+
size 1383
|
checkpoint-930/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:575c542669bc9cfb6772b3cd2c090f4f58d4d118403fe0107a1504eb7890bda4
|
| 3 |
+
size 5817
|
checkpoint-930/token_loss_weights_v40.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ea572531bd71b4185b73bd3a0a98d90779b75212c034ed508702971bb0d3fcc8
|
| 3 |
+
size 132690
|
checkpoint-930/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
checkpoint-930/tokenizer_config.json
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"backend": "tokenizers",
|
| 3 |
+
"bos_token": "<s>",
|
| 4 |
+
"eos_token": "</s>",
|
| 5 |
+
"extra_special_tokens": [
|
| 6 |
+
"<|user|>",
|
| 7 |
+
"<|assistant|>"
|
| 8 |
+
],
|
| 9 |
+
"model_max_length": 1000000000000000019884624838656,
|
| 10 |
+
"pad_token": "<pad>",
|
| 11 |
+
"tokenizer_class": "TokenizersBackend",
|
| 12 |
+
"unk_token": "<unk>"
|
| 13 |
+
}
|
checkpoint-930/trainer_state.json
ADDED
|
@@ -0,0 +1,778 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.36908423454707806,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 930,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.003968647683301915,
|
| 14 |
+
"grad_norm": 2101.331787109375,
|
| 15 |
+
"learning_rate": 2.844835142723442e-07,
|
| 16 |
+
"loss": 466.4673828125,
|
| 17 |
+
"original_ce_loss": 10.5001,
|
| 18 |
+
"step": 10
|
| 19 |
+
},
|
| 20 |
+
{
|
| 21 |
+
"epoch": 0.00793729536660383,
|
| 22 |
+
"grad_norm": 2084.678466796875,
|
| 23 |
+
"learning_rate": 8.534505428170324e-07,
|
| 24 |
+
"loss": 466.596875,
|
| 25 |
+
"original_ce_loss": 10.4967,
|
| 26 |
+
"step": 20
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.011905943049905745,
|
| 30 |
+
"grad_norm": 2071.28759765625,
|
| 31 |
+
"learning_rate": 1.4224175713617208e-06,
|
| 32 |
+
"loss": 466.864599609375,
|
| 33 |
+
"original_ce_loss": 10.4927,
|
| 34 |
+
"step": 30
|
| 35 |
+
},
|
| 36 |
+
{
|
| 37 |
+
"epoch": 0.01587459073320766,
|
| 38 |
+
"grad_norm": 2066.62744140625,
|
| 39 |
+
"learning_rate": 1.9913845999064094e-06,
|
| 40 |
+
"loss": 466.76552734375,
|
| 41 |
+
"original_ce_loss": 10.4905,
|
| 42 |
+
"step": 40
|
| 43 |
+
},
|
| 44 |
+
{
|
| 45 |
+
"epoch": 0.019843238416509576,
|
| 46 |
+
"grad_norm": 1969.30712890625,
|
| 47 |
+
"learning_rate": 2.5603516284510975e-06,
|
| 48 |
+
"loss": 466.3068359375,
|
| 49 |
+
"original_ce_loss": 10.4853,
|
| 50 |
+
"step": 50
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.02381188609981149,
|
| 54 |
+
"grad_norm": 2112.412353515625,
|
| 55 |
+
"learning_rate": 3.1293186569957857e-06,
|
| 56 |
+
"loss": 465.471630859375,
|
| 57 |
+
"original_ce_loss": 10.4831,
|
| 58 |
+
"step": 60
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.027780533783113404,
|
| 62 |
+
"grad_norm": 2076.740478515625,
|
| 63 |
+
"learning_rate": 3.6982856855404742e-06,
|
| 64 |
+
"loss": 466.027197265625,
|
| 65 |
+
"original_ce_loss": 10.4791,
|
| 66 |
+
"step": 70
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.03174918146641532,
|
| 70 |
+
"grad_norm": 1999.7998046875,
|
| 71 |
+
"learning_rate": 4.267252714085162e-06,
|
| 72 |
+
"loss": 465.8962890625,
|
| 73 |
+
"original_ce_loss": 10.4756,
|
| 74 |
+
"step": 80
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.03571782914971723,
|
| 78 |
+
"grad_norm": 2193.890625,
|
| 79 |
+
"learning_rate": 4.836219742629851e-06,
|
| 80 |
+
"loss": 465.14560546875,
|
| 81 |
+
"original_ce_loss": 10.4709,
|
| 82 |
+
"step": 90
|
| 83 |
+
},
|
| 84 |
+
{
|
| 85 |
+
"epoch": 0.03968647683301915,
|
| 86 |
+
"grad_norm": 2095.904296875,
|
| 87 |
+
"learning_rate": 5.405186771174539e-06,
|
| 88 |
+
"loss": 465.246826171875,
|
| 89 |
+
"original_ce_loss": 10.4638,
|
| 90 |
+
"step": 100
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"epoch": 0.043655124516321066,
|
| 94 |
+
"grad_norm": 2066.883056640625,
|
| 95 |
+
"learning_rate": 5.974153799719227e-06,
|
| 96 |
+
"loss": 465.16279296875,
|
| 97 |
+
"original_ce_loss": 10.4587,
|
| 98 |
+
"step": 110
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.04762377219962298,
|
| 102 |
+
"grad_norm": 2181.439697265625,
|
| 103 |
+
"learning_rate": 6.543120828263916e-06,
|
| 104 |
+
"loss": 464.945849609375,
|
| 105 |
+
"original_ce_loss": 10.4472,
|
| 106 |
+
"step": 120
|
| 107 |
+
},
|
| 108 |
+
{
|
| 109 |
+
"epoch": 0.05159241988292489,
|
| 110 |
+
"grad_norm": 2059.518310546875,
|
| 111 |
+
"learning_rate": 7.112087856808604e-06,
|
| 112 |
+
"loss": 464.43115234375,
|
| 113 |
+
"original_ce_loss": 10.4344,
|
| 114 |
+
"step": 130
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.05556106756622681,
|
| 118 |
+
"grad_norm": 2026.2479248046875,
|
| 119 |
+
"learning_rate": 7.681054885353293e-06,
|
| 120 |
+
"loss": 463.418798828125,
|
| 121 |
+
"original_ce_loss": 10.4118,
|
| 122 |
+
"step": 140
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.05952971524952872,
|
| 126 |
+
"grad_norm": 2144.916748046875,
|
| 127 |
+
"learning_rate": 8.25002191389798e-06,
|
| 128 |
+
"loss": 462.48349609375,
|
| 129 |
+
"original_ce_loss": 10.3735,
|
| 130 |
+
"step": 150
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 0.06349836293283063,
|
| 134 |
+
"grad_norm": 2034.1129150390625,
|
| 135 |
+
"learning_rate": 8.81898894244267e-06,
|
| 136 |
+
"loss": 459.26025390625,
|
| 137 |
+
"original_ce_loss": 10.2947,
|
| 138 |
+
"step": 160
|
| 139 |
+
},
|
| 140 |
+
{
|
| 141 |
+
"epoch": 0.06746701061613256,
|
| 142 |
+
"grad_norm": Infinity,
|
| 143 |
+
"learning_rate": 9.387955970987358e-06,
|
| 144 |
+
"loss": 453.1400390625,
|
| 145 |
+
"original_ce_loss": 10.1282,
|
| 146 |
+
"step": 170
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.07143565829943446,
|
| 150 |
+
"grad_norm": 1609.81005859375,
|
| 151 |
+
"learning_rate": 9.900026296677575e-06,
|
| 152 |
+
"loss": 441.692138671875,
|
| 153 |
+
"original_ce_loss": 9.8347,
|
| 154 |
+
"step": 180
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"epoch": 0.07540430598273638,
|
| 158 |
+
"grad_norm": 1214.3907470703125,
|
| 159 |
+
"learning_rate": 1.0468993325222264e-05,
|
| 160 |
+
"loss": 426.199365234375,
|
| 161 |
+
"original_ce_loss": 9.439,
|
| 162 |
+
"step": 190
|
| 163 |
+
},
|
| 164 |
+
{
|
| 165 |
+
"epoch": 0.0793729536660383,
|
| 166 |
+
"grad_norm": 774.9044189453125,
|
| 167 |
+
"learning_rate": 1.0981063650912485e-05,
|
| 168 |
+
"loss": 411.568994140625,
|
| 169 |
+
"original_ce_loss": 9.0833,
|
| 170 |
+
"step": 200
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.08334160134934021,
|
| 174 |
+
"grad_norm": 444.1908264160156,
|
| 175 |
+
"learning_rate": 1.1550030679457174e-05,
|
| 176 |
+
"loss": 398.152783203125,
|
| 177 |
+
"original_ce_loss": 8.7625,
|
| 178 |
+
"step": 210
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.08731024903264213,
|
| 182 |
+
"grad_norm": 351.087158203125,
|
| 183 |
+
"learning_rate": 1.2118997708001861e-05,
|
| 184 |
+
"loss": 384.9672607421875,
|
| 185 |
+
"original_ce_loss": 8.449,
|
| 186 |
+
"step": 220
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 0.09127889671594404,
|
| 190 |
+
"grad_norm": 419.7826843261719,
|
| 191 |
+
"learning_rate": 1.2631068033692081e-05,
|
| 192 |
+
"loss": 373.1683349609375,
|
| 193 |
+
"original_ce_loss": 8.1801,
|
| 194 |
+
"step": 230
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 0.09524754439924596,
|
| 198 |
+
"grad_norm": 538.2559814453125,
|
| 199 |
+
"learning_rate": 1.320003506223677e-05,
|
| 200 |
+
"loss": 364.2321533203125,
|
| 201 |
+
"original_ce_loss": 7.9687,
|
| 202 |
+
"step": 240
|
| 203 |
+
},
|
| 204 |
+
{
|
| 205 |
+
"epoch": 0.09921619208254787,
|
| 206 |
+
"grad_norm": 128.8115234375,
|
| 207 |
+
"learning_rate": 1.3769002090781456e-05,
|
| 208 |
+
"loss": 358.3741455078125,
|
| 209 |
+
"original_ce_loss": 7.835,
|
| 210 |
+
"step": 250
|
| 211 |
+
},
|
| 212 |
+
{
|
| 213 |
+
"epoch": 0.10318483976584979,
|
| 214 |
+
"grad_norm": 64.08294677734375,
|
| 215 |
+
"learning_rate": 1.4337969119326145e-05,
|
| 216 |
+
"loss": 354.5633544921875,
|
| 217 |
+
"original_ce_loss": 7.7481,
|
| 218 |
+
"step": 260
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 0.10715348744915171,
|
| 222 |
+
"grad_norm": 808.9030151367188,
|
| 223 |
+
"learning_rate": 1.4850039445016365e-05,
|
| 224 |
+
"loss": 351.1748779296875,
|
| 225 |
+
"original_ce_loss": 7.6716,
|
| 226 |
+
"step": 270
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.11112213513245361,
|
| 230 |
+
"grad_norm": 75.48285675048828,
|
| 231 |
+
"learning_rate": 1.5419006473561053e-05,
|
| 232 |
+
"loss": 350.7310546875,
|
| 233 |
+
"original_ce_loss": 7.667,
|
| 234 |
+
"step": 280
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.11509078281575554,
|
| 238 |
+
"grad_norm": 73.40753173828125,
|
| 239 |
+
"learning_rate": 1.5931076799251275e-05,
|
| 240 |
+
"loss": 348.7738037109375,
|
| 241 |
+
"original_ce_loss": 7.6291,
|
| 242 |
+
"step": 290
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 0.11905943049905744,
|
| 246 |
+
"grad_norm": 62.33892059326172,
|
| 247 |
+
"learning_rate": 1.650004382779596e-05,
|
| 248 |
+
"loss": 348.23212890625,
|
| 249 |
+
"original_ce_loss": 7.6153,
|
| 250 |
+
"step": 300
|
| 251 |
+
},
|
| 252 |
+
{
|
| 253 |
+
"epoch": 0.12302807818235936,
|
| 254 |
+
"grad_norm": 71.73210144042969,
|
| 255 |
+
"learning_rate": 1.706901085634065e-05,
|
| 256 |
+
"loss": 348.2334228515625,
|
| 257 |
+
"original_ce_loss": 7.6091,
|
| 258 |
+
"step": 310
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"epoch": 0.12699672586566127,
|
| 262 |
+
"grad_norm": 135.77700805664062,
|
| 263 |
+
"learning_rate": 1.763797788488534e-05,
|
| 264 |
+
"loss": 347.606640625,
|
| 265 |
+
"original_ce_loss": 7.5999,
|
| 266 |
+
"step": 320
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 0.1309653735489632,
|
| 270 |
+
"grad_norm": 104.31532287597656,
|
| 271 |
+
"learning_rate": 1.8206944913430027e-05,
|
| 272 |
+
"loss": 346.425048828125,
|
| 273 |
+
"original_ce_loss": 7.5693,
|
| 274 |
+
"step": 330
|
| 275 |
+
},
|
| 276 |
+
{
|
| 277 |
+
"epoch": 0.1349340212322651,
|
| 278 |
+
"grad_norm": 103.98027801513672,
|
| 279 |
+
"learning_rate": 1.8775911941974716e-05,
|
| 280 |
+
"loss": 346.0474853515625,
|
| 281 |
+
"original_ce_loss": 7.5694,
|
| 282 |
+
"step": 340
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.13890266891556702,
|
| 286 |
+
"grad_norm": 91.162841796875,
|
| 287 |
+
"learning_rate": 1.9344878970519405e-05,
|
| 288 |
+
"loss": 347.2004150390625,
|
| 289 |
+
"original_ce_loss": 7.5861,
|
| 290 |
+
"step": 350
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.14287131659886892,
|
| 294 |
+
"grad_norm": 99.40938568115234,
|
| 295 |
+
"learning_rate": 1.991384599906409e-05,
|
| 296 |
+
"loss": 346.3804931640625,
|
| 297 |
+
"original_ce_loss": 7.574,
|
| 298 |
+
"step": 360
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 0.14683996428217086,
|
| 302 |
+
"grad_norm": 104.16796875,
|
| 303 |
+
"learning_rate": 2.048281302760878e-05,
|
| 304 |
+
"loss": 345.12392578125,
|
| 305 |
+
"original_ce_loss": 7.546,
|
| 306 |
+
"step": 370
|
| 307 |
+
},
|
| 308 |
+
{
|
| 309 |
+
"epoch": 0.15080861196547277,
|
| 310 |
+
"grad_norm": 99.03206634521484,
|
| 311 |
+
"learning_rate": 2.105178005615347e-05,
|
| 312 |
+
"loss": 344.866552734375,
|
| 313 |
+
"original_ce_loss": 7.5415,
|
| 314 |
+
"step": 380
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 0.15477725964877467,
|
| 318 |
+
"grad_norm": 103.81749725341797,
|
| 319 |
+
"learning_rate": 2.1620747084698155e-05,
|
| 320 |
+
"loss": 346.1248046875,
|
| 321 |
+
"original_ce_loss": 7.5586,
|
| 322 |
+
"step": 390
|
| 323 |
+
},
|
| 324 |
+
{
|
| 325 |
+
"epoch": 0.1587459073320766,
|
| 326 |
+
"grad_norm": 128.07980346679688,
|
| 327 |
+
"learning_rate": 2.2189714113242844e-05,
|
| 328 |
+
"loss": 345.1707275390625,
|
| 329 |
+
"original_ce_loss": 7.5483,
|
| 330 |
+
"step": 400
|
| 331 |
+
},
|
| 332 |
+
{
|
| 333 |
+
"epoch": 0.16271455501537851,
|
| 334 |
+
"grad_norm": 148.4123992919922,
|
| 335 |
+
"learning_rate": 2.2758681141787533e-05,
|
| 336 |
+
"loss": 346.3683349609375,
|
| 337 |
+
"original_ce_loss": 7.5609,
|
| 338 |
+
"step": 410
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.16668320269868042,
|
| 342 |
+
"grad_norm": 255.14109802246094,
|
| 343 |
+
"learning_rate": 2.3327648170332218e-05,
|
| 344 |
+
"loss": 342.63876953125,
|
| 345 |
+
"original_ce_loss": 7.4901,
|
| 346 |
+
"step": 420
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 0.17065185038198233,
|
| 350 |
+
"grad_norm": 324.200927734375,
|
| 351 |
+
"learning_rate": 2.3896615198876907e-05,
|
| 352 |
+
"loss": 342.1252197265625,
|
| 353 |
+
"original_ce_loss": 7.4664,
|
| 354 |
+
"step": 430
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 0.17462049806528426,
|
| 358 |
+
"grad_norm": 507.1748352050781,
|
| 359 |
+
"learning_rate": 2.4465582227421596e-05,
|
| 360 |
+
"loss": 338.0018798828125,
|
| 361 |
+
"original_ce_loss": 7.3921,
|
| 362 |
+
"step": 440
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 0.17858914574858617,
|
| 366 |
+
"grad_norm": 505.0448303222656,
|
| 367 |
+
"learning_rate": 2.5034549255966285e-05,
|
| 368 |
+
"loss": 337.4599853515625,
|
| 369 |
+
"original_ce_loss": 7.3639,
|
| 370 |
+
"step": 450
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 0.18255779343188808,
|
| 374 |
+
"grad_norm": 647.8860473632812,
|
| 375 |
+
"learning_rate": 2.5603516284510974e-05,
|
| 376 |
+
"loss": 332.9173828125,
|
| 377 |
+
"original_ce_loss": 7.264,
|
| 378 |
+
"step": 460
|
| 379 |
+
},
|
| 380 |
+
{
|
| 381 |
+
"epoch": 0.18652644111519,
|
| 382 |
+
"grad_norm": 613.0924682617188,
|
| 383 |
+
"learning_rate": 2.6172483313055663e-05,
|
| 384 |
+
"loss": 332.1484130859375,
|
| 385 |
+
"original_ce_loss": 7.2431,
|
| 386 |
+
"step": 470
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 0.19049508879849192,
|
| 390 |
+
"grad_norm": 570.2612915039062,
|
| 391 |
+
"learning_rate": 2.674145034160035e-05,
|
| 392 |
+
"loss": 326.793359375,
|
| 393 |
+
"original_ce_loss": 7.1306,
|
| 394 |
+
"step": 480
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 0.19446373648179383,
|
| 398 |
+
"grad_norm": 500.37518310546875,
|
| 399 |
+
"learning_rate": 2.7310417370145038e-05,
|
| 400 |
+
"loss": 323.844970703125,
|
| 401 |
+
"original_ce_loss": 7.0655,
|
| 402 |
+
"step": 490
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 0.19843238416509573,
|
| 406 |
+
"grad_norm": 560.87255859375,
|
| 407 |
+
"learning_rate": 2.7879384398689727e-05,
|
| 408 |
+
"loss": 322.6688232421875,
|
| 409 |
+
"original_ce_loss": 7.042,
|
| 410 |
+
"step": 500
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 0.20240103184839767,
|
| 414 |
+
"grad_norm": 499.6172180175781,
|
| 415 |
+
"learning_rate": 2.8448351427234416e-05,
|
| 416 |
+
"loss": 318.80859375,
|
| 417 |
+
"original_ce_loss": 6.9507,
|
| 418 |
+
"step": 510
|
| 419 |
+
},
|
| 420 |
+
{
|
| 421 |
+
"epoch": 0.20636967953169957,
|
| 422 |
+
"grad_norm": 471.60455322265625,
|
| 423 |
+
"learning_rate": 2.844801087524661e-05,
|
| 424 |
+
"loss": 316.4650390625,
|
| 425 |
+
"original_ce_loss": 6.9075,
|
| 426 |
+
"step": 520
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 0.21033832721500148,
|
| 430 |
+
"grad_norm": 526.83154296875,
|
| 431 |
+
"learning_rate": 2.844698923559002e-05,
|
| 432 |
+
"loss": 313.0283935546875,
|
| 433 |
+
"original_ce_loss": 6.8281,
|
| 434 |
+
"step": 530
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 0.21430697489830342,
|
| 438 |
+
"grad_norm": 478.6401062011719,
|
| 439 |
+
"learning_rate": 2.8445286557184383e-05,
|
| 440 |
+
"loss": 309.740966796875,
|
| 441 |
+
"original_ce_loss": 6.7505,
|
| 442 |
+
"step": 540
|
| 443 |
+
},
|
| 444 |
+
{
|
| 445 |
+
"epoch": 0.21827562258160532,
|
| 446 |
+
"grad_norm": 687.4654541015625,
|
| 447 |
+
"learning_rate": 2.844290292155998e-05,
|
| 448 |
+
"loss": 306.3443115234375,
|
| 449 |
+
"original_ce_loss": 6.675,
|
| 450 |
+
"step": 550
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 0.22224427026490723,
|
| 454 |
+
"grad_norm": 488.8817138671875,
|
| 455 |
+
"learning_rate": 2.8439838442853727e-05,
|
| 456 |
+
"loss": 303.6910400390625,
|
| 457 |
+
"original_ce_loss": 6.6184,
|
| 458 |
+
"step": 560
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 0.22621291794820914,
|
| 462 |
+
"grad_norm": 405.0874328613281,
|
| 463 |
+
"learning_rate": 2.8436093267803745e-05,
|
| 464 |
+
"loss": 300.372998046875,
|
| 465 |
+
"original_ce_loss": 6.5505,
|
| 466 |
+
"step": 570
|
| 467 |
+
},
|
| 468 |
+
{
|
| 469 |
+
"epoch": 0.23018156563151107,
|
| 470 |
+
"grad_norm": 401.0887451171875,
|
| 471 |
+
"learning_rate": 2.843166757574229e-05,
|
| 472 |
+
"loss": 298.1054931640625,
|
| 473 |
+
"original_ce_loss": 6.4938,
|
| 474 |
+
"step": 580
|
| 475 |
+
},
|
| 476 |
+
{
|
| 477 |
+
"epoch": 0.23415021331481298,
|
| 478 |
+
"grad_norm": 386.3600769042969,
|
| 479 |
+
"learning_rate": 2.8426561578587193e-05,
|
| 480 |
+
"loss": 293.0273193359375,
|
| 481 |
+
"original_ce_loss": 6.3834,
|
| 482 |
+
"step": 590
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 0.23811886099811488,
|
| 486 |
+
"grad_norm": 406.5262451171875,
|
| 487 |
+
"learning_rate": 2.842077552083171e-05,
|
| 488 |
+
"loss": 290.217822265625,
|
| 489 |
+
"original_ce_loss": 6.3263,
|
| 490 |
+
"step": 600
|
| 491 |
+
},
|
| 492 |
+
{
|
| 493 |
+
"epoch": 0.24208750868141682,
|
| 494 |
+
"grad_norm": 297.536376953125,
|
| 495 |
+
"learning_rate": 2.841430967953279e-05,
|
| 496 |
+
"loss": 286.4669189453125,
|
| 497 |
+
"original_ce_loss": 6.2343,
|
| 498 |
+
"step": 610
|
| 499 |
+
},
|
| 500 |
+
{
|
| 501 |
+
"epoch": 0.24605615636471873,
|
| 502 |
+
"grad_norm": 291.78033447265625,
|
| 503 |
+
"learning_rate": 2.8407164364297865e-05,
|
| 504 |
+
"loss": 283.9787109375,
|
| 505 |
+
"original_ce_loss": 6.1821,
|
| 506 |
+
"step": 620
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 0.25002480404802063,
|
| 510 |
+
"grad_norm": 232.47671508789062,
|
| 511 |
+
"learning_rate": 2.839933991726995e-05,
|
| 512 |
+
"loss": 279.2822265625,
|
| 513 |
+
"original_ce_loss": 6.0786,
|
| 514 |
+
"step": 630
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 0.25399345173132254,
|
| 518 |
+
"grad_norm": 280.51068115234375,
|
| 519 |
+
"learning_rate": 2.8390836713111327e-05,
|
| 520 |
+
"loss": 278.6858154296875,
|
| 521 |
+
"original_ce_loss": 6.0634,
|
| 522 |
+
"step": 640
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 0.25796209941462445,
|
| 526 |
+
"grad_norm": 242.15980529785156,
|
| 527 |
+
"learning_rate": 2.838165515898556e-05,
|
| 528 |
+
"loss": 275.26123046875,
|
| 529 |
+
"original_ce_loss": 5.9858,
|
| 530 |
+
"step": 650
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 0.2619307470979264,
|
| 534 |
+
"grad_norm": 235.6015625,
|
| 535 |
+
"learning_rate": 2.837281213374654e-05,
|
| 536 |
+
"loss": 275.1755859375,
|
| 537 |
+
"original_ce_loss": 5.9812,
|
| 538 |
+
"step": 660
|
| 539 |
+
},
|
| 540 |
+
{
|
| 541 |
+
"epoch": 0.2658993947812283,
|
| 542 |
+
"grad_norm": 360.1375732421875,
|
| 543 |
+
"learning_rate": 2.836234295273547e-05,
|
| 544 |
+
"loss": 272.129296875,
|
| 545 |
+
"original_ce_loss": 5.9163,
|
| 546 |
+
"step": 670
|
| 547 |
+
},
|
| 548 |
+
{
|
| 549 |
+
"epoch": 0.2698680424645302,
|
| 550 |
+
"grad_norm": 211.124267578125,
|
| 551 |
+
"learning_rate": 2.8351196786139552e-05,
|
| 552 |
+
"loss": 269.84140625,
|
| 553 |
+
"original_ce_loss": 5.8635,
|
| 554 |
+
"step": 680
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 0.27383669014783213,
|
| 558 |
+
"grad_norm": 227.95819091796875,
|
| 559 |
+
"learning_rate": 2.8339374167676766e-05,
|
| 560 |
+
"loss": 266.558349609375,
|
| 561 |
+
"original_ce_loss": 5.7916,
|
| 562 |
+
"step": 690
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 0.27780533783113404,
|
| 566 |
+
"grad_norm": 209.81101989746094,
|
| 567 |
+
"learning_rate": 2.8326875663456015e-05,
|
| 568 |
+
"loss": 266.0473876953125,
|
| 569 |
+
"original_ce_loss": 5.7767,
|
| 570 |
+
"step": 700
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 0.28177398551443594,
|
| 574 |
+
"grad_norm": 210.2840576171875,
|
| 575 |
+
"learning_rate": 2.8313701871949996e-05,
|
| 576 |
+
"loss": 266.055859375,
|
| 577 |
+
"original_ce_loss": 5.7716,
|
| 578 |
+
"step": 710
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 0.28574263319773785,
|
| 582 |
+
"grad_norm": 172.32827758789062,
|
| 583 |
+
"learning_rate": 2.8299853423966575e-05,
|
| 584 |
+
"loss": 264.1585693359375,
|
| 585 |
+
"original_ce_loss": 5.7324,
|
| 586 |
+
"step": 720
|
| 587 |
+
},
|
| 588 |
+
{
|
| 589 |
+
"epoch": 0.2897112808810398,
|
| 590 |
+
"grad_norm": 208.0069122314453,
|
| 591 |
+
"learning_rate": 2.828533098261854e-05,
|
| 592 |
+
"loss": 262.2663330078125,
|
| 593 |
+
"original_ce_loss": 5.6872,
|
| 594 |
+
"step": 730
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 0.2936799285643417,
|
| 598 |
+
"grad_norm": 216.8614044189453,
|
| 599 |
+
"learning_rate": 2.8270135243291892e-05,
|
| 600 |
+
"loss": 258.9171875,
|
| 601 |
+
"original_ce_loss": 5.6204,
|
| 602 |
+
"step": 740
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 0.2976485762476436,
|
| 606 |
+
"grad_norm": 278.64862060546875,
|
| 607 |
+
"learning_rate": 2.8254266933612513e-05,
|
| 608 |
+
"loss": 258.2635009765625,
|
| 609 |
+
"original_ce_loss": 5.5968,
|
| 610 |
+
"step": 750
|
| 611 |
+
},
|
| 612 |
+
{
|
| 613 |
+
"epoch": 0.30161722393094553,
|
| 614 |
+
"grad_norm": 186.52618408203125,
|
| 615 |
+
"learning_rate": 2.8237726813411343e-05,
|
| 616 |
+
"loss": 257.0150390625,
|
| 617 |
+
"original_ce_loss": 5.5646,
|
| 618 |
+
"step": 760
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 0.30558587161424744,
|
| 622 |
+
"grad_norm": 167.430419921875,
|
| 623 |
+
"learning_rate": 2.8220515674687993e-05,
|
| 624 |
+
"loss": 253.3641357421875,
|
| 625 |
+
"original_ce_loss": 5.4868,
|
| 626 |
+
"step": 770
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 0.30955451929754935,
|
| 630 |
+
"grad_norm": 193.23880004882812,
|
| 631 |
+
"learning_rate": 2.820263434157282e-05,
|
| 632 |
+
"loss": 250.8727783203125,
|
| 633 |
+
"original_ce_loss": 5.4357,
|
| 634 |
+
"step": 780
|
| 635 |
+
},
|
| 636 |
+
{
|
| 637 |
+
"epoch": 0.31352316698085125,
|
| 638 |
+
"grad_norm": 177.97837829589844,
|
| 639 |
+
"learning_rate": 2.8184083670287468e-05,
|
| 640 |
+
"loss": 251.076953125,
|
| 641 |
+
"original_ce_loss": 5.4343,
|
| 642 |
+
"step": 790
|
| 643 |
+
},
|
| 644 |
+
{
|
| 645 |
+
"epoch": 0.3174918146641532,
|
| 646 |
+
"grad_norm": 164.1091766357422,
|
| 647 |
+
"learning_rate": 2.8164864549103864e-05,
|
| 648 |
+
"loss": 249.2934814453125,
|
| 649 |
+
"original_ce_loss": 5.397,
|
| 650 |
+
"step": 800
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 0.3214604623474551,
|
| 654 |
+
"grad_norm": 155.91461181640625,
|
| 655 |
+
"learning_rate": 2.8144977898301676e-05,
|
| 656 |
+
"loss": 248.4820068359375,
|
| 657 |
+
"original_ce_loss": 5.3777,
|
| 658 |
+
"step": 810
|
| 659 |
+
},
|
| 660 |
+
{
|
| 661 |
+
"epoch": 0.32542911003075703,
|
| 662 |
+
"grad_norm": 172.32583618164062,
|
| 663 |
+
"learning_rate": 2.8124424670124282e-05,
|
| 664 |
+
"loss": 247.750146484375,
|
| 665 |
+
"original_ce_loss": 5.3599,
|
| 666 |
+
"step": 820
|
| 667 |
+
},
|
| 668 |
+
{
|
| 669 |
+
"epoch": 0.32939775771405894,
|
| 670 |
+
"grad_norm": 157.87686157226562,
|
| 671 |
+
"learning_rate": 2.810320584873312e-05,
|
| 672 |
+
"loss": 244.25830078125,
|
| 673 |
+
"original_ce_loss": 5.2861,
|
| 674 |
+
"step": 830
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 0.33336640539736084,
|
| 678 |
+
"grad_norm": 152.00811767578125,
|
| 679 |
+
"learning_rate": 2.8081322450160622e-05,
|
| 680 |
+
"loss": 243.0696044921875,
|
| 681 |
+
"original_ce_loss": 5.2528,
|
| 682 |
+
"step": 840
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 0.33733505308066275,
|
| 686 |
+
"grad_norm": 151.34193420410156,
|
| 687 |
+
"learning_rate": 2.80587755222615e-05,
|
| 688 |
+
"loss": 243.5411865234375,
|
| 689 |
+
"original_ce_loss": 5.2562,
|
| 690 |
+
"step": 850
|
| 691 |
+
},
|
| 692 |
+
{
|
| 693 |
+
"epoch": 0.34130370076396466,
|
| 694 |
+
"grad_norm": 158.08863830566406,
|
| 695 |
+
"learning_rate": 2.8035566144662623e-05,
|
| 696 |
+
"loss": 240.95771484375,
|
| 697 |
+
"original_ce_loss": 5.2061,
|
| 698 |
+
"step": 860
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 0.3452723484472666,
|
| 702 |
+
"grad_norm": 150.26466369628906,
|
| 703 |
+
"learning_rate": 2.80116954287113e-05,
|
| 704 |
+
"loss": 239.8141357421875,
|
| 705 |
+
"original_ce_loss": 5.1777,
|
| 706 |
+
"step": 870
|
| 707 |
+
},
|
| 708 |
+
{
|
| 709 |
+
"epoch": 0.3492409961305685,
|
| 710 |
+
"grad_norm": 151.70632934570312,
|
| 711 |
+
"learning_rate": 2.7987164517422063e-05,
|
| 712 |
+
"loss": 237.826611328125,
|
| 713 |
+
"original_ce_loss": 5.1408,
|
| 714 |
+
"step": 880
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 0.35320964381387043,
|
| 718 |
+
"grad_norm": 136.18653869628906,
|
| 719 |
+
"learning_rate": 2.7961974585421932e-05,
|
| 720 |
+
"loss": 238.3519775390625,
|
| 721 |
+
"original_ce_loss": 5.1429,
|
| 722 |
+
"step": 890
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 0.35717829149717234,
|
| 726 |
+
"grad_norm": 172.5818634033203,
|
| 727 |
+
"learning_rate": 2.7936126838894187e-05,
|
| 728 |
+
"loss": 238.17529296875,
|
| 729 |
+
"original_ce_loss": 5.1432,
|
| 730 |
+
"step": 900
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 0.36114693918047425,
|
| 734 |
+
"grad_norm": 139.51486206054688,
|
| 735 |
+
"learning_rate": 2.7909622515520584e-05,
|
| 736 |
+
"loss": 235.6244140625,
|
| 737 |
+
"original_ce_loss": 5.0856,
|
| 738 |
+
"step": 910
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"epoch": 0.36511558686377615,
|
| 742 |
+
"grad_norm": 143.4366455078125,
|
| 743 |
+
"learning_rate": 2.788246288442213e-05,
|
| 744 |
+
"loss": 234.6486328125,
|
| 745 |
+
"original_ce_loss": 5.0681,
|
| 746 |
+
"step": 920
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 0.36908423454707806,
|
| 750 |
+
"grad_norm": 139.58184814453125,
|
| 751 |
+
"learning_rate": 2.7854649246098267e-05,
|
| 752 |
+
"loss": 233.9451171875,
|
| 753 |
+
"original_ce_loss": 5.0506,
|
| 754 |
+
"step": 930
|
| 755 |
+
}
|
| 756 |
+
],
|
| 757 |
+
"logging_steps": 10,
|
| 758 |
+
"max_steps": 5040,
|
| 759 |
+
"num_input_tokens_seen": 0,
|
| 760 |
+
"num_train_epochs": 2,
|
| 761 |
+
"save_steps": 30,
|
| 762 |
+
"stateful_callbacks": {
|
| 763 |
+
"TrainerControl": {
|
| 764 |
+
"args": {
|
| 765 |
+
"should_epoch_stop": false,
|
| 766 |
+
"should_evaluate": false,
|
| 767 |
+
"should_log": false,
|
| 768 |
+
"should_save": true,
|
| 769 |
+
"should_training_stop": false
|
| 770 |
+
},
|
| 771 |
+
"attributes": {}
|
| 772 |
+
}
|
| 773 |
+
},
|
| 774 |
+
"total_flos": 3.069051681600307e+16,
|
| 775 |
+
"train_batch_size": 32,
|
| 776 |
+
"trial_name": null,
|
| 777 |
+
"trial_params": null
|
| 778 |
+
}
|
checkpoint-930/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a7b45389216e038e995655293423392b226d726a357436eb78f1f87c318fe803
|
| 3 |
+
size 5201
|