Upload gpthinker
Browse files- config.json +39 -0
- generation_config.json +6 -0
- model.safetensors +3 -0
- optimizer.pt +3 -0
- rng_state.pth +3 -0
- scheduler.pt +3 -0
- trainer_state.json +2749 -0
- training_args.bin +3 -0
config.json
ADDED
|
@@ -0,0 +1,39 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "gpt2",
|
| 3 |
+
"activation_function": "gelu_new",
|
| 4 |
+
"architectures": [
|
| 5 |
+
"GPT2LMHeadModel"
|
| 6 |
+
],
|
| 7 |
+
"attn_pdrop": 0.1,
|
| 8 |
+
"bos_token_id": 50256,
|
| 9 |
+
"embd_pdrop": 0.1,
|
| 10 |
+
"eos_token_id": 50256,
|
| 11 |
+
"initializer_range": 0.02,
|
| 12 |
+
"layer_norm_epsilon": 1e-05,
|
| 13 |
+
"model_type": "gpt2",
|
| 14 |
+
"n_ctx": 1024,
|
| 15 |
+
"n_embd": 768,
|
| 16 |
+
"n_head": 12,
|
| 17 |
+
"n_inner": null,
|
| 18 |
+
"n_layer": 12,
|
| 19 |
+
"n_positions": 1024,
|
| 20 |
+
"reorder_and_upcast_attn": false,
|
| 21 |
+
"resid_pdrop": 0.1,
|
| 22 |
+
"scale_attn_by_inverse_layer_idx": false,
|
| 23 |
+
"scale_attn_weights": true,
|
| 24 |
+
"summary_activation": null,
|
| 25 |
+
"summary_first_dropout": 0.1,
|
| 26 |
+
"summary_proj_to_labels": true,
|
| 27 |
+
"summary_type": "cls_index",
|
| 28 |
+
"summary_use_proj": true,
|
| 29 |
+
"task_specific_params": {
|
| 30 |
+
"text-generation": {
|
| 31 |
+
"do_sample": true,
|
| 32 |
+
"max_length": 50
|
| 33 |
+
}
|
| 34 |
+
},
|
| 35 |
+
"torch_dtype": "float32",
|
| 36 |
+
"transformers_version": "4.48.2",
|
| 37 |
+
"use_cache": true,
|
| 38 |
+
"vocab_size": 50258
|
| 39 |
+
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_from_model_config": true,
|
| 3 |
+
"bos_token_id": 50256,
|
| 4 |
+
"eos_token_id": 50256,
|
| 5 |
+
"transformers_version": "4.48.2"
|
| 6 |
+
}
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:829361bd714b9a08e9360745d357ce2b435a798442754ba244403be591286ab0
|
| 3 |
+
size 497777280
|
optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cb4522dfbdb311c7c7d64af7ac16bc368a865bfc0b06e0ccce4e1694e77b3159
|
| 3 |
+
size 995648442
|
rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:629903c4105fe72e7c7d3206916a13adb502449f5f94f4167b4c71e314e0b989
|
| 3 |
+
size 14244
|
scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:83b8e20fbd46e8be147f4fa8213d0e1954c561c901c282254e331132610448c2
|
| 3 |
+
size 1064
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2749 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 3.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 31635,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.00948316737790422,
|
| 13 |
+
"grad_norm": 3.985076904296875,
|
| 14 |
+
"learning_rate": 9.7e-06,
|
| 15 |
+
"loss": 2.4371,
|
| 16 |
+
"step": 100
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.01896633475580844,
|
| 20 |
+
"grad_norm": 3.8551318645477295,
|
| 21 |
+
"learning_rate": 1.97e-05,
|
| 22 |
+
"loss": 2.1056,
|
| 23 |
+
"step": 200
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.02844950213371266,
|
| 27 |
+
"grad_norm": 4.302079200744629,
|
| 28 |
+
"learning_rate": 2.97e-05,
|
| 29 |
+
"loss": 1.9608,
|
| 30 |
+
"step": 300
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.03793266951161688,
|
| 34 |
+
"grad_norm": 3.31756329536438,
|
| 35 |
+
"learning_rate": 3.97e-05,
|
| 36 |
+
"loss": 1.8338,
|
| 37 |
+
"step": 400
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.0474158368895211,
|
| 41 |
+
"grad_norm": 2.4619405269622803,
|
| 42 |
+
"learning_rate": 4.97e-05,
|
| 43 |
+
"loss": 1.7855,
|
| 44 |
+
"step": 500
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.0474158368895211,
|
| 48 |
+
"eval_loss": 1.6501274108886719,
|
| 49 |
+
"eval_runtime": 72.2019,
|
| 50 |
+
"eval_samples_per_second": 129.83,
|
| 51 |
+
"eval_steps_per_second": 16.232,
|
| 52 |
+
"step": 500
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.05689900426742532,
|
| 56 |
+
"grad_norm": 2.553483724594116,
|
| 57 |
+
"learning_rate": 4.9844226754456404e-05,
|
| 58 |
+
"loss": 1.7277,
|
| 59 |
+
"step": 600
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.06638217164532954,
|
| 63 |
+
"grad_norm": 2.0428194999694824,
|
| 64 |
+
"learning_rate": 4.9683635779669185e-05,
|
| 65 |
+
"loss": 1.6971,
|
| 66 |
+
"step": 700
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.07586533902323377,
|
| 70 |
+
"grad_norm": 1.9449608325958252,
|
| 71 |
+
"learning_rate": 4.9523044804881966e-05,
|
| 72 |
+
"loss": 1.6537,
|
| 73 |
+
"step": 800
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.08534850640113797,
|
| 77 |
+
"grad_norm": 2.5439252853393555,
|
| 78 |
+
"learning_rate": 4.9362453830094753e-05,
|
| 79 |
+
"loss": 1.6464,
|
| 80 |
+
"step": 900
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.0948316737790422,
|
| 84 |
+
"grad_norm": 2.118544578552246,
|
| 85 |
+
"learning_rate": 4.9201862855307534e-05,
|
| 86 |
+
"loss": 1.5804,
|
| 87 |
+
"step": 1000
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.0948316737790422,
|
| 91 |
+
"eval_loss": 1.5088456869125366,
|
| 92 |
+
"eval_runtime": 72.0739,
|
| 93 |
+
"eval_samples_per_second": 130.061,
|
| 94 |
+
"eval_steps_per_second": 16.261,
|
| 95 |
+
"step": 1000
|
| 96 |
+
},
|
| 97 |
+
{
|
| 98 |
+
"epoch": 0.10431484115694642,
|
| 99 |
+
"grad_norm": 1.8551363945007324,
|
| 100 |
+
"learning_rate": 4.9041271880520315e-05,
|
| 101 |
+
"loss": 1.6341,
|
| 102 |
+
"step": 1100
|
| 103 |
+
},
|
| 104 |
+
{
|
| 105 |
+
"epoch": 0.11379800853485064,
|
| 106 |
+
"grad_norm": 1.9903297424316406,
|
| 107 |
+
"learning_rate": 4.88806809057331e-05,
|
| 108 |
+
"loss": 1.5718,
|
| 109 |
+
"step": 1200
|
| 110 |
+
},
|
| 111 |
+
{
|
| 112 |
+
"epoch": 0.12328117591275486,
|
| 113 |
+
"grad_norm": 2.2142210006713867,
|
| 114 |
+
"learning_rate": 4.8720089930945884e-05,
|
| 115 |
+
"loss": 1.5718,
|
| 116 |
+
"step": 1300
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 0.1327643432906591,
|
| 120 |
+
"grad_norm": 2.2737417221069336,
|
| 121 |
+
"learning_rate": 4.8559498956158664e-05,
|
| 122 |
+
"loss": 1.5137,
|
| 123 |
+
"step": 1400
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"epoch": 0.1422475106685633,
|
| 127 |
+
"grad_norm": 2.3361587524414062,
|
| 128 |
+
"learning_rate": 4.839890798137145e-05,
|
| 129 |
+
"loss": 1.5332,
|
| 130 |
+
"step": 1500
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 0.1422475106685633,
|
| 134 |
+
"eval_loss": 1.4451285600662231,
|
| 135 |
+
"eval_runtime": 72.138,
|
| 136 |
+
"eval_samples_per_second": 129.945,
|
| 137 |
+
"eval_steps_per_second": 16.247,
|
| 138 |
+
"step": 1500
|
| 139 |
+
},
|
| 140 |
+
{
|
| 141 |
+
"epoch": 0.15173067804646753,
|
| 142 |
+
"grad_norm": 2.335610866546631,
|
| 143 |
+
"learning_rate": 4.823831700658423e-05,
|
| 144 |
+
"loss": 1.5669,
|
| 145 |
+
"step": 1600
|
| 146 |
+
},
|
| 147 |
+
{
|
| 148 |
+
"epoch": 0.16121384542437173,
|
| 149 |
+
"grad_norm": 1.811543583869934,
|
| 150 |
+
"learning_rate": 4.8077726031797014e-05,
|
| 151 |
+
"loss": 1.4985,
|
| 152 |
+
"step": 1700
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 0.17069701280227595,
|
| 156 |
+
"grad_norm": 2.1588528156280518,
|
| 157 |
+
"learning_rate": 4.79171350570098e-05,
|
| 158 |
+
"loss": 1.4979,
|
| 159 |
+
"step": 1800
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.18018018018018017,
|
| 163 |
+
"grad_norm": 1.7643985748291016,
|
| 164 |
+
"learning_rate": 4.775654408222258e-05,
|
| 165 |
+
"loss": 1.5246,
|
| 166 |
+
"step": 1900
|
| 167 |
+
},
|
| 168 |
+
{
|
| 169 |
+
"epoch": 0.1896633475580844,
|
| 170 |
+
"grad_norm": 1.9193495512008667,
|
| 171 |
+
"learning_rate": 4.759595310743536e-05,
|
| 172 |
+
"loss": 1.4915,
|
| 173 |
+
"step": 2000
|
| 174 |
+
},
|
| 175 |
+
{
|
| 176 |
+
"epoch": 0.1896633475580844,
|
| 177 |
+
"eval_loss": 1.403477430343628,
|
| 178 |
+
"eval_runtime": 71.9579,
|
| 179 |
+
"eval_samples_per_second": 130.271,
|
| 180 |
+
"eval_steps_per_second": 16.287,
|
| 181 |
+
"step": 2000
|
| 182 |
+
},
|
| 183 |
+
{
|
| 184 |
+
"epoch": 0.19914651493598862,
|
| 185 |
+
"grad_norm": 1.8307377099990845,
|
| 186 |
+
"learning_rate": 4.743536213264815e-05,
|
| 187 |
+
"loss": 1.5009,
|
| 188 |
+
"step": 2100
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 0.20862968231389284,
|
| 192 |
+
"grad_norm": 1.7923104763031006,
|
| 193 |
+
"learning_rate": 4.727477115786093e-05,
|
| 194 |
+
"loss": 1.4968,
|
| 195 |
+
"step": 2200
|
| 196 |
+
},
|
| 197 |
+
{
|
| 198 |
+
"epoch": 0.21811284969179706,
|
| 199 |
+
"grad_norm": 1.925938367843628,
|
| 200 |
+
"learning_rate": 4.711418018307371e-05,
|
| 201 |
+
"loss": 1.4696,
|
| 202 |
+
"step": 2300
|
| 203 |
+
},
|
| 204 |
+
{
|
| 205 |
+
"epoch": 0.22759601706970128,
|
| 206 |
+
"grad_norm": 2.106110095977783,
|
| 207 |
+
"learning_rate": 4.69535892082865e-05,
|
| 208 |
+
"loss": 1.4853,
|
| 209 |
+
"step": 2400
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 0.2370791844476055,
|
| 213 |
+
"grad_norm": 2.345017433166504,
|
| 214 |
+
"learning_rate": 4.679299823349928e-05,
|
| 215 |
+
"loss": 1.4868,
|
| 216 |
+
"step": 2500
|
| 217 |
+
},
|
| 218 |
+
{
|
| 219 |
+
"epoch": 0.2370791844476055,
|
| 220 |
+
"eval_loss": 1.3772392272949219,
|
| 221 |
+
"eval_runtime": 72.0321,
|
| 222 |
+
"eval_samples_per_second": 130.136,
|
| 223 |
+
"eval_steps_per_second": 16.271,
|
| 224 |
+
"step": 2500
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 0.24656235182550973,
|
| 228 |
+
"grad_norm": 1.5003846883773804,
|
| 229 |
+
"learning_rate": 4.663240725871206e-05,
|
| 230 |
+
"loss": 1.4641,
|
| 231 |
+
"step": 2600
|
| 232 |
+
},
|
| 233 |
+
{
|
| 234 |
+
"epoch": 0.25604551920341395,
|
| 235 |
+
"grad_norm": 1.8472124338150024,
|
| 236 |
+
"learning_rate": 4.647181628392485e-05,
|
| 237 |
+
"loss": 1.4594,
|
| 238 |
+
"step": 2700
|
| 239 |
+
},
|
| 240 |
+
{
|
| 241 |
+
"epoch": 0.2655286865813182,
|
| 242 |
+
"grad_norm": 1.8818256855010986,
|
| 243 |
+
"learning_rate": 4.631122530913763e-05,
|
| 244 |
+
"loss": 1.4547,
|
| 245 |
+
"step": 2800
|
| 246 |
+
},
|
| 247 |
+
{
|
| 248 |
+
"epoch": 0.2750118539592224,
|
| 249 |
+
"grad_norm": 1.5926233530044556,
|
| 250 |
+
"learning_rate": 4.615063433435041e-05,
|
| 251 |
+
"loss": 1.4414,
|
| 252 |
+
"step": 2900
|
| 253 |
+
},
|
| 254 |
+
{
|
| 255 |
+
"epoch": 0.2844950213371266,
|
| 256 |
+
"grad_norm": 1.505327820777893,
|
| 257 |
+
"learning_rate": 4.59900433595632e-05,
|
| 258 |
+
"loss": 1.4165,
|
| 259 |
+
"step": 3000
|
| 260 |
+
},
|
| 261 |
+
{
|
| 262 |
+
"epoch": 0.2844950213371266,
|
| 263 |
+
"eval_loss": 1.3518378734588623,
|
| 264 |
+
"eval_runtime": 71.9886,
|
| 265 |
+
"eval_samples_per_second": 130.215,
|
| 266 |
+
"eval_steps_per_second": 16.28,
|
| 267 |
+
"step": 3000
|
| 268 |
+
},
|
| 269 |
+
{
|
| 270 |
+
"epoch": 0.29397818871503084,
|
| 271 |
+
"grad_norm": 1.77092707157135,
|
| 272 |
+
"learning_rate": 4.582945238477598e-05,
|
| 273 |
+
"loss": 1.4222,
|
| 274 |
+
"step": 3100
|
| 275 |
+
},
|
| 276 |
+
{
|
| 277 |
+
"epoch": 0.30346135609293506,
|
| 278 |
+
"grad_norm": 2.265411376953125,
|
| 279 |
+
"learning_rate": 4.566886140998876e-05,
|
| 280 |
+
"loss": 1.3973,
|
| 281 |
+
"step": 3200
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 0.3129445234708393,
|
| 285 |
+
"grad_norm": 1.4207345247268677,
|
| 286 |
+
"learning_rate": 4.550827043520154e-05,
|
| 287 |
+
"loss": 1.4423,
|
| 288 |
+
"step": 3300
|
| 289 |
+
},
|
| 290 |
+
{
|
| 291 |
+
"epoch": 0.32242769084874345,
|
| 292 |
+
"grad_norm": 1.72047758102417,
|
| 293 |
+
"learning_rate": 4.534767946041433e-05,
|
| 294 |
+
"loss": 1.3939,
|
| 295 |
+
"step": 3400
|
| 296 |
+
},
|
| 297 |
+
{
|
| 298 |
+
"epoch": 0.3319108582266477,
|
| 299 |
+
"grad_norm": 1.7695670127868652,
|
| 300 |
+
"learning_rate": 4.518708848562711e-05,
|
| 301 |
+
"loss": 1.3911,
|
| 302 |
+
"step": 3500
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 0.3319108582266477,
|
| 306 |
+
"eval_loss": 1.3347505331039429,
|
| 307 |
+
"eval_runtime": 72.0526,
|
| 308 |
+
"eval_samples_per_second": 130.099,
|
| 309 |
+
"eval_steps_per_second": 16.266,
|
| 310 |
+
"step": 3500
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.3413940256045519,
|
| 314 |
+
"grad_norm": 1.93614661693573,
|
| 315 |
+
"learning_rate": 4.502649751083989e-05,
|
| 316 |
+
"loss": 1.405,
|
| 317 |
+
"step": 3600
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.3508771929824561,
|
| 321 |
+
"grad_norm": 1.4412301778793335,
|
| 322 |
+
"learning_rate": 4.486590653605268e-05,
|
| 323 |
+
"loss": 1.421,
|
| 324 |
+
"step": 3700
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.36036036036036034,
|
| 328 |
+
"grad_norm": 1.5761134624481201,
|
| 329 |
+
"learning_rate": 4.470531556126546e-05,
|
| 330 |
+
"loss": 1.3758,
|
| 331 |
+
"step": 3800
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.36984352773826457,
|
| 335 |
+
"grad_norm": 1.7923239469528198,
|
| 336 |
+
"learning_rate": 4.454472458647824e-05,
|
| 337 |
+
"loss": 1.4087,
|
| 338 |
+
"step": 3900
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.3793266951161688,
|
| 342 |
+
"grad_norm": 2.2492587566375732,
|
| 343 |
+
"learning_rate": 4.438413361169103e-05,
|
| 344 |
+
"loss": 1.3797,
|
| 345 |
+
"step": 4000
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.3793266951161688,
|
| 349 |
+
"eval_loss": 1.3214360475540161,
|
| 350 |
+
"eval_runtime": 72.0741,
|
| 351 |
+
"eval_samples_per_second": 130.061,
|
| 352 |
+
"eval_steps_per_second": 16.261,
|
| 353 |
+
"step": 4000
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 0.388809862494073,
|
| 357 |
+
"grad_norm": 1.978060245513916,
|
| 358 |
+
"learning_rate": 4.422354263690381e-05,
|
| 359 |
+
"loss": 1.4024,
|
| 360 |
+
"step": 4100
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 0.39829302987197723,
|
| 364 |
+
"grad_norm": 1.7838459014892578,
|
| 365 |
+
"learning_rate": 4.406295166211659e-05,
|
| 366 |
+
"loss": 1.4047,
|
| 367 |
+
"step": 4200
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 0.40777619724988146,
|
| 371 |
+
"grad_norm": 1.682637333869934,
|
| 372 |
+
"learning_rate": 4.3902360687329377e-05,
|
| 373 |
+
"loss": 1.3709,
|
| 374 |
+
"step": 4300
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 0.4172593646277857,
|
| 378 |
+
"grad_norm": 1.5510674715042114,
|
| 379 |
+
"learning_rate": 4.374176971254216e-05,
|
| 380 |
+
"loss": 1.4175,
|
| 381 |
+
"step": 4400
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 0.4267425320056899,
|
| 385 |
+
"grad_norm": 1.7401492595672607,
|
| 386 |
+
"learning_rate": 4.358117873775494e-05,
|
| 387 |
+
"loss": 1.3801,
|
| 388 |
+
"step": 4500
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 0.4267425320056899,
|
| 392 |
+
"eval_loss": 1.3049076795578003,
|
| 393 |
+
"eval_runtime": 72.1294,
|
| 394 |
+
"eval_samples_per_second": 129.961,
|
| 395 |
+
"eval_steps_per_second": 16.249,
|
| 396 |
+
"step": 4500
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"epoch": 0.4362256993835941,
|
| 400 |
+
"grad_norm": 1.6590989828109741,
|
| 401 |
+
"learning_rate": 4.3420587762967726e-05,
|
| 402 |
+
"loss": 1.3827,
|
| 403 |
+
"step": 4600
|
| 404 |
+
},
|
| 405 |
+
{
|
| 406 |
+
"epoch": 0.44570886676149835,
|
| 407 |
+
"grad_norm": 1.5440171957015991,
|
| 408 |
+
"learning_rate": 4.325999678818051e-05,
|
| 409 |
+
"loss": 1.3617,
|
| 410 |
+
"step": 4700
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 0.45519203413940257,
|
| 414 |
+
"grad_norm": 1.716539978981018,
|
| 415 |
+
"learning_rate": 4.309940581339329e-05,
|
| 416 |
+
"loss": 1.3463,
|
| 417 |
+
"step": 4800
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 0.4646752015173068,
|
| 421 |
+
"grad_norm": 1.3042521476745605,
|
| 422 |
+
"learning_rate": 4.2938814838606075e-05,
|
| 423 |
+
"loss": 1.3456,
|
| 424 |
+
"step": 4900
|
| 425 |
+
},
|
| 426 |
+
{
|
| 427 |
+
"epoch": 0.474158368895211,
|
| 428 |
+
"grad_norm": 1.3467687368392944,
|
| 429 |
+
"learning_rate": 4.2778223863818856e-05,
|
| 430 |
+
"loss": 1.3559,
|
| 431 |
+
"step": 5000
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 0.474158368895211,
|
| 435 |
+
"eval_loss": 1.2918757200241089,
|
| 436 |
+
"eval_runtime": 72.0072,
|
| 437 |
+
"eval_samples_per_second": 130.181,
|
| 438 |
+
"eval_steps_per_second": 16.276,
|
| 439 |
+
"step": 5000
|
| 440 |
+
},
|
| 441 |
+
{
|
| 442 |
+
"epoch": 0.48364153627311524,
|
| 443 |
+
"grad_norm": 1.3807010650634766,
|
| 444 |
+
"learning_rate": 4.261763288903164e-05,
|
| 445 |
+
"loss": 1.3507,
|
| 446 |
+
"step": 5100
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 0.49312470365101946,
|
| 450 |
+
"grad_norm": 1.3885177373886108,
|
| 451 |
+
"learning_rate": 4.2457041914244425e-05,
|
| 452 |
+
"loss": 1.3552,
|
| 453 |
+
"step": 5200
|
| 454 |
+
},
|
| 455 |
+
{
|
| 456 |
+
"epoch": 0.5026078710289237,
|
| 457 |
+
"grad_norm": 1.2807698249816895,
|
| 458 |
+
"learning_rate": 4.2296450939457205e-05,
|
| 459 |
+
"loss": 1.3642,
|
| 460 |
+
"step": 5300
|
| 461 |
+
},
|
| 462 |
+
{
|
| 463 |
+
"epoch": 0.5120910384068279,
|
| 464 |
+
"grad_norm": 1.4009428024291992,
|
| 465 |
+
"learning_rate": 4.2135859964669986e-05,
|
| 466 |
+
"loss": 1.3781,
|
| 467 |
+
"step": 5400
|
| 468 |
+
},
|
| 469 |
+
{
|
| 470 |
+
"epoch": 0.5215742057847321,
|
| 471 |
+
"grad_norm": 1.3763035535812378,
|
| 472 |
+
"learning_rate": 4.1975268989882774e-05,
|
| 473 |
+
"loss": 1.3717,
|
| 474 |
+
"step": 5500
|
| 475 |
+
},
|
| 476 |
+
{
|
| 477 |
+
"epoch": 0.5215742057847321,
|
| 478 |
+
"eval_loss": 1.280537724494934,
|
| 479 |
+
"eval_runtime": 72.1115,
|
| 480 |
+
"eval_samples_per_second": 129.993,
|
| 481 |
+
"eval_steps_per_second": 16.253,
|
| 482 |
+
"step": 5500
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 0.5310573731626363,
|
| 486 |
+
"grad_norm": 1.5511786937713623,
|
| 487 |
+
"learning_rate": 4.1814678015095555e-05,
|
| 488 |
+
"loss": 1.3502,
|
| 489 |
+
"step": 5600
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 0.5405405405405406,
|
| 493 |
+
"grad_norm": 1.4995437860488892,
|
| 494 |
+
"learning_rate": 4.1654087040308336e-05,
|
| 495 |
+
"loss": 1.3599,
|
| 496 |
+
"step": 5700
|
| 497 |
+
},
|
| 498 |
+
{
|
| 499 |
+
"epoch": 0.5500237079184448,
|
| 500 |
+
"grad_norm": 1.3496274948120117,
|
| 501 |
+
"learning_rate": 4.149349606552112e-05,
|
| 502 |
+
"loss": 1.3421,
|
| 503 |
+
"step": 5800
|
| 504 |
+
},
|
| 505 |
+
{
|
| 506 |
+
"epoch": 0.559506875296349,
|
| 507 |
+
"grad_norm": 1.3634631633758545,
|
| 508 |
+
"learning_rate": 4.1332905090733904e-05,
|
| 509 |
+
"loss": 1.3617,
|
| 510 |
+
"step": 5900
|
| 511 |
+
},
|
| 512 |
+
{
|
| 513 |
+
"epoch": 0.5689900426742532,
|
| 514 |
+
"grad_norm": 1.5579423904418945,
|
| 515 |
+
"learning_rate": 4.1172314115946685e-05,
|
| 516 |
+
"loss": 1.3604,
|
| 517 |
+
"step": 6000
|
| 518 |
+
},
|
| 519 |
+
{
|
| 520 |
+
"epoch": 0.5689900426742532,
|
| 521 |
+
"eval_loss": 1.2698478698730469,
|
| 522 |
+
"eval_runtime": 72.1231,
|
| 523 |
+
"eval_samples_per_second": 129.972,
|
| 524 |
+
"eval_steps_per_second": 16.25,
|
| 525 |
+
"step": 6000
|
| 526 |
+
},
|
| 527 |
+
{
|
| 528 |
+
"epoch": 0.5784732100521575,
|
| 529 |
+
"grad_norm": 1.380241870880127,
|
| 530 |
+
"learning_rate": 4.101332905090734e-05,
|
| 531 |
+
"loss": 1.3379,
|
| 532 |
+
"step": 6100
|
| 533 |
+
},
|
| 534 |
+
{
|
| 535 |
+
"epoch": 0.5879563774300617,
|
| 536 |
+
"grad_norm": 1.764551043510437,
|
| 537 |
+
"learning_rate": 4.085273807612012e-05,
|
| 538 |
+
"loss": 1.3208,
|
| 539 |
+
"step": 6200
|
| 540 |
+
},
|
| 541 |
+
{
|
| 542 |
+
"epoch": 0.5974395448079659,
|
| 543 |
+
"grad_norm": 1.627012848854065,
|
| 544 |
+
"learning_rate": 4.069214710133291e-05,
|
| 545 |
+
"loss": 1.3448,
|
| 546 |
+
"step": 6300
|
| 547 |
+
},
|
| 548 |
+
{
|
| 549 |
+
"epoch": 0.6069227121858701,
|
| 550 |
+
"grad_norm": 1.539115071296692,
|
| 551 |
+
"learning_rate": 4.053155612654569e-05,
|
| 552 |
+
"loss": 1.3422,
|
| 553 |
+
"step": 6400
|
| 554 |
+
},
|
| 555 |
+
{
|
| 556 |
+
"epoch": 0.6164058795637744,
|
| 557 |
+
"grad_norm": 1.4698444604873657,
|
| 558 |
+
"learning_rate": 4.037257106150635e-05,
|
| 559 |
+
"loss": 1.3264,
|
| 560 |
+
"step": 6500
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 0.6164058795637744,
|
| 564 |
+
"eval_loss": 1.259299635887146,
|
| 565 |
+
"eval_runtime": 72.1176,
|
| 566 |
+
"eval_samples_per_second": 129.982,
|
| 567 |
+
"eval_steps_per_second": 16.251,
|
| 568 |
+
"step": 6500
|
| 569 |
+
},
|
| 570 |
+
{
|
| 571 |
+
"epoch": 0.6258890469416786,
|
| 572 |
+
"grad_norm": 1.8150815963745117,
|
| 573 |
+
"learning_rate": 4.021198008671913e-05,
|
| 574 |
+
"loss": 1.3262,
|
| 575 |
+
"step": 6600
|
| 576 |
+
},
|
| 577 |
+
{
|
| 578 |
+
"epoch": 0.6353722143195828,
|
| 579 |
+
"grad_norm": 1.4278889894485474,
|
| 580 |
+
"learning_rate": 4.005138911193191e-05,
|
| 581 |
+
"loss": 1.334,
|
| 582 |
+
"step": 6700
|
| 583 |
+
},
|
| 584 |
+
{
|
| 585 |
+
"epoch": 0.6448553816974869,
|
| 586 |
+
"grad_norm": 1.4713215827941895,
|
| 587 |
+
"learning_rate": 3.98907981371447e-05,
|
| 588 |
+
"loss": 1.2924,
|
| 589 |
+
"step": 6800
|
| 590 |
+
},
|
| 591 |
+
{
|
| 592 |
+
"epoch": 0.6543385490753911,
|
| 593 |
+
"grad_norm": 1.626541018486023,
|
| 594 |
+
"learning_rate": 3.9731813072105354e-05,
|
| 595 |
+
"loss": 1.3057,
|
| 596 |
+
"step": 6900
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 0.6638217164532954,
|
| 600 |
+
"grad_norm": 1.7835373878479004,
|
| 601 |
+
"learning_rate": 3.9571222097318134e-05,
|
| 602 |
+
"loss": 1.328,
|
| 603 |
+
"step": 7000
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 0.6638217164532954,
|
| 607 |
+
"eval_loss": 1.252388834953308,
|
| 608 |
+
"eval_runtime": 72.2427,
|
| 609 |
+
"eval_samples_per_second": 129.757,
|
| 610 |
+
"eval_steps_per_second": 16.223,
|
| 611 |
+
"step": 7000
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 0.6733048838311996,
|
| 615 |
+
"grad_norm": 1.8675563335418701,
|
| 616 |
+
"learning_rate": 3.9410631122530915e-05,
|
| 617 |
+
"loss": 1.322,
|
| 618 |
+
"step": 7100
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 0.6827880512091038,
|
| 622 |
+
"grad_norm": 1.5719430446624756,
|
| 623 |
+
"learning_rate": 3.92500401477437e-05,
|
| 624 |
+
"loss": 1.3464,
|
| 625 |
+
"step": 7200
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 0.692271218587008,
|
| 629 |
+
"grad_norm": 1.5038641691207886,
|
| 630 |
+
"learning_rate": 3.9089449172956484e-05,
|
| 631 |
+
"loss": 1.3315,
|
| 632 |
+
"step": 7300
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 0.7017543859649122,
|
| 636 |
+
"grad_norm": 1.777970314025879,
|
| 637 |
+
"learning_rate": 3.8928858198169265e-05,
|
| 638 |
+
"loss": 1.3549,
|
| 639 |
+
"step": 7400
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.7112375533428165,
|
| 643 |
+
"grad_norm": 1.8796472549438477,
|
| 644 |
+
"learning_rate": 3.8768267223382045e-05,
|
| 645 |
+
"loss": 1.2907,
|
| 646 |
+
"step": 7500
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"epoch": 0.7112375533428165,
|
| 650 |
+
"eval_loss": 1.2450358867645264,
|
| 651 |
+
"eval_runtime": 72.1657,
|
| 652 |
+
"eval_samples_per_second": 129.895,
|
| 653 |
+
"eval_steps_per_second": 16.24,
|
| 654 |
+
"step": 7500
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 0.7207207207207207,
|
| 658 |
+
"grad_norm": 1.7477796077728271,
|
| 659 |
+
"learning_rate": 3.860767624859483e-05,
|
| 660 |
+
"loss": 1.3196,
|
| 661 |
+
"step": 7600
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 0.7302038880986249,
|
| 665 |
+
"grad_norm": 1.6598505973815918,
|
| 666 |
+
"learning_rate": 3.8447085273807614e-05,
|
| 667 |
+
"loss": 1.2799,
|
| 668 |
+
"step": 7700
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 0.7396870554765291,
|
| 672 |
+
"grad_norm": 1.7319283485412598,
|
| 673 |
+
"learning_rate": 3.8286494299020395e-05,
|
| 674 |
+
"loss": 1.3354,
|
| 675 |
+
"step": 7800
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 0.7491702228544334,
|
| 679 |
+
"grad_norm": 1.847347617149353,
|
| 680 |
+
"learning_rate": 3.812590332423318e-05,
|
| 681 |
+
"loss": 1.3034,
|
| 682 |
+
"step": 7900
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 0.7586533902323376,
|
| 686 |
+
"grad_norm": 1.6584995985031128,
|
| 687 |
+
"learning_rate": 3.796531234944596e-05,
|
| 688 |
+
"loss": 1.3092,
|
| 689 |
+
"step": 8000
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 0.7586533902323376,
|
| 693 |
+
"eval_loss": 1.2385543584823608,
|
| 694 |
+
"eval_runtime": 72.1594,
|
| 695 |
+
"eval_samples_per_second": 129.907,
|
| 696 |
+
"eval_steps_per_second": 16.242,
|
| 697 |
+
"step": 8000
|
| 698 |
+
},
|
| 699 |
+
{
|
| 700 |
+
"epoch": 0.7681365576102418,
|
| 701 |
+
"grad_norm": 1.581036925315857,
|
| 702 |
+
"learning_rate": 3.7804721374658744e-05,
|
| 703 |
+
"loss": 1.3064,
|
| 704 |
+
"step": 8100
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 0.777619724988146,
|
| 708 |
+
"grad_norm": 1.6824501752853394,
|
| 709 |
+
"learning_rate": 3.764413039987153e-05,
|
| 710 |
+
"loss": 1.3039,
|
| 711 |
+
"step": 8200
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 0.7871028923660502,
|
| 715 |
+
"grad_norm": 1.4804019927978516,
|
| 716 |
+
"learning_rate": 3.748353942508431e-05,
|
| 717 |
+
"loss": 1.2774,
|
| 718 |
+
"step": 8300
|
| 719 |
+
},
|
| 720 |
+
{
|
| 721 |
+
"epoch": 0.7965860597439545,
|
| 722 |
+
"grad_norm": 1.5401322841644287,
|
| 723 |
+
"learning_rate": 3.732294845029709e-05,
|
| 724 |
+
"loss": 1.3042,
|
| 725 |
+
"step": 8400
|
| 726 |
+
},
|
| 727 |
+
{
|
| 728 |
+
"epoch": 0.8060692271218587,
|
| 729 |
+
"grad_norm": 1.9226937294006348,
|
| 730 |
+
"learning_rate": 3.716235747550988e-05,
|
| 731 |
+
"loss": 1.3186,
|
| 732 |
+
"step": 8500
|
| 733 |
+
},
|
| 734 |
+
{
|
| 735 |
+
"epoch": 0.8060692271218587,
|
| 736 |
+
"eval_loss": 1.2315117120742798,
|
| 737 |
+
"eval_runtime": 72.0639,
|
| 738 |
+
"eval_samples_per_second": 130.079,
|
| 739 |
+
"eval_steps_per_second": 16.263,
|
| 740 |
+
"step": 8500
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 0.8155523944997629,
|
| 744 |
+
"grad_norm": 1.3993178606033325,
|
| 745 |
+
"learning_rate": 3.700176650072266e-05,
|
| 746 |
+
"loss": 1.3074,
|
| 747 |
+
"step": 8600
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 0.8250355618776671,
|
| 751 |
+
"grad_norm": 1.6044120788574219,
|
| 752 |
+
"learning_rate": 3.684117552593544e-05,
|
| 753 |
+
"loss": 1.2681,
|
| 754 |
+
"step": 8700
|
| 755 |
+
},
|
| 756 |
+
{
|
| 757 |
+
"epoch": 0.8345187292555714,
|
| 758 |
+
"grad_norm": 1.6285070180892944,
|
| 759 |
+
"learning_rate": 3.668058455114823e-05,
|
| 760 |
+
"loss": 1.3198,
|
| 761 |
+
"step": 8800
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"epoch": 0.8440018966334756,
|
| 765 |
+
"grad_norm": 2.002086639404297,
|
| 766 |
+
"learning_rate": 3.651999357636101e-05,
|
| 767 |
+
"loss": 1.3227,
|
| 768 |
+
"step": 8900
|
| 769 |
+
},
|
| 770 |
+
{
|
| 771 |
+
"epoch": 0.8534850640113798,
|
| 772 |
+
"grad_norm": 1.5941271781921387,
|
| 773 |
+
"learning_rate": 3.635940260157379e-05,
|
| 774 |
+
"loss": 1.2914,
|
| 775 |
+
"step": 9000
|
| 776 |
+
},
|
| 777 |
+
{
|
| 778 |
+
"epoch": 0.8534850640113798,
|
| 779 |
+
"eval_loss": 1.2264697551727295,
|
| 780 |
+
"eval_runtime": 72.0482,
|
| 781 |
+
"eval_samples_per_second": 130.107,
|
| 782 |
+
"eval_steps_per_second": 16.267,
|
| 783 |
+
"step": 9000
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 0.862968231389284,
|
| 787 |
+
"grad_norm": 1.5721193552017212,
|
| 788 |
+
"learning_rate": 3.619881162678658e-05,
|
| 789 |
+
"loss": 1.3268,
|
| 790 |
+
"step": 9100
|
| 791 |
+
},
|
| 792 |
+
{
|
| 793 |
+
"epoch": 0.8724513987671882,
|
| 794 |
+
"grad_norm": 1.7066916227340698,
|
| 795 |
+
"learning_rate": 3.603822065199936e-05,
|
| 796 |
+
"loss": 1.2845,
|
| 797 |
+
"step": 9200
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 0.8819345661450925,
|
| 801 |
+
"grad_norm": 1.5683172941207886,
|
| 802 |
+
"learning_rate": 3.587762967721214e-05,
|
| 803 |
+
"loss": 1.2779,
|
| 804 |
+
"step": 9300
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 0.8914177335229967,
|
| 808 |
+
"grad_norm": 1.7200586795806885,
|
| 809 |
+
"learning_rate": 3.571703870242493e-05,
|
| 810 |
+
"loss": 1.3161,
|
| 811 |
+
"step": 9400
|
| 812 |
+
},
|
| 813 |
+
{
|
| 814 |
+
"epoch": 0.9009009009009009,
|
| 815 |
+
"grad_norm": 1.4963386058807373,
|
| 816 |
+
"learning_rate": 3.555644772763771e-05,
|
| 817 |
+
"loss": 1.2668,
|
| 818 |
+
"step": 9500
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 0.9009009009009009,
|
| 822 |
+
"eval_loss": 1.2190866470336914,
|
| 823 |
+
"eval_runtime": 72.0991,
|
| 824 |
+
"eval_samples_per_second": 130.015,
|
| 825 |
+
"eval_steps_per_second": 16.255,
|
| 826 |
+
"step": 9500
|
| 827 |
+
},
|
| 828 |
+
{
|
| 829 |
+
"epoch": 0.9103840682788051,
|
| 830 |
+
"grad_norm": 1.5414083003997803,
|
| 831 |
+
"learning_rate": 3.539585675285049e-05,
|
| 832 |
+
"loss": 1.3185,
|
| 833 |
+
"step": 9600
|
| 834 |
+
},
|
| 835 |
+
{
|
| 836 |
+
"epoch": 0.9198672356567094,
|
| 837 |
+
"grad_norm": 1.46302330493927,
|
| 838 |
+
"learning_rate": 3.523526577806328e-05,
|
| 839 |
+
"loss": 1.2485,
|
| 840 |
+
"step": 9700
|
| 841 |
+
},
|
| 842 |
+
{
|
| 843 |
+
"epoch": 0.9293504030346136,
|
| 844 |
+
"grad_norm": 1.4815856218338013,
|
| 845 |
+
"learning_rate": 3.507467480327606e-05,
|
| 846 |
+
"loss": 1.2912,
|
| 847 |
+
"step": 9800
|
| 848 |
+
},
|
| 849 |
+
{
|
| 850 |
+
"epoch": 0.9388335704125178,
|
| 851 |
+
"grad_norm": 1.5166754722595215,
|
| 852 |
+
"learning_rate": 3.491408382848884e-05,
|
| 853 |
+
"loss": 1.2722,
|
| 854 |
+
"step": 9900
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 0.948316737790422,
|
| 858 |
+
"grad_norm": 1.9628846645355225,
|
| 859 |
+
"learning_rate": 3.475349285370163e-05,
|
| 860 |
+
"loss": 1.2538,
|
| 861 |
+
"step": 10000
|
| 862 |
+
},
|
| 863 |
+
{
|
| 864 |
+
"epoch": 0.948316737790422,
|
| 865 |
+
"eval_loss": 1.2150416374206543,
|
| 866 |
+
"eval_runtime": 72.1513,
|
| 867 |
+
"eval_samples_per_second": 129.921,
|
| 868 |
+
"eval_steps_per_second": 16.244,
|
| 869 |
+
"step": 10000
|
| 870 |
+
},
|
| 871 |
+
{
|
| 872 |
+
"epoch": 0.9577999051683262,
|
| 873 |
+
"grad_norm": 1.6791901588439941,
|
| 874 |
+
"learning_rate": 3.459290187891441e-05,
|
| 875 |
+
"loss": 1.2624,
|
| 876 |
+
"step": 10100
|
| 877 |
+
},
|
| 878 |
+
{
|
| 879 |
+
"epoch": 0.9672830725462305,
|
| 880 |
+
"grad_norm": 1.5026668310165405,
|
| 881 |
+
"learning_rate": 3.443231090412719e-05,
|
| 882 |
+
"loss": 1.2696,
|
| 883 |
+
"step": 10200
|
| 884 |
+
},
|
| 885 |
+
{
|
| 886 |
+
"epoch": 0.9767662399241347,
|
| 887 |
+
"grad_norm": 1.176558017730713,
|
| 888 |
+
"learning_rate": 3.427171992933998e-05,
|
| 889 |
+
"loss": 1.29,
|
| 890 |
+
"step": 10300
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 0.9862494073020389,
|
| 894 |
+
"grad_norm": 1.5698468685150146,
|
| 895 |
+
"learning_rate": 3.411112895455276e-05,
|
| 896 |
+
"loss": 1.2874,
|
| 897 |
+
"step": 10400
|
| 898 |
+
},
|
| 899 |
+
{
|
| 900 |
+
"epoch": 0.9957325746799431,
|
| 901 |
+
"grad_norm": 1.4970085620880127,
|
| 902 |
+
"learning_rate": 3.395053797976554e-05,
|
| 903 |
+
"loss": 1.2874,
|
| 904 |
+
"step": 10500
|
| 905 |
+
},
|
| 906 |
+
{
|
| 907 |
+
"epoch": 0.9957325746799431,
|
| 908 |
+
"eval_loss": 1.2110899686813354,
|
| 909 |
+
"eval_runtime": 72.0475,
|
| 910 |
+
"eval_samples_per_second": 130.109,
|
| 911 |
+
"eval_steps_per_second": 16.267,
|
| 912 |
+
"step": 10500
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 1.0052157420578474,
|
| 916 |
+
"grad_norm": 1.284839391708374,
|
| 917 |
+
"learning_rate": 3.3789947004978326e-05,
|
| 918 |
+
"loss": 1.2793,
|
| 919 |
+
"step": 10600
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 1.0146989094357515,
|
| 923 |
+
"grad_norm": 1.680851697921753,
|
| 924 |
+
"learning_rate": 3.362935603019111e-05,
|
| 925 |
+
"loss": 1.2487,
|
| 926 |
+
"step": 10700
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 1.0241820768136558,
|
| 930 |
+
"grad_norm": 1.659610629081726,
|
| 931 |
+
"learning_rate": 3.346876505540389e-05,
|
| 932 |
+
"loss": 1.2454,
|
| 933 |
+
"step": 10800
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 1.03366524419156,
|
| 937 |
+
"grad_norm": 1.6641312837600708,
|
| 938 |
+
"learning_rate": 3.330817408061667e-05,
|
| 939 |
+
"loss": 1.2323,
|
| 940 |
+
"step": 10900
|
| 941 |
+
},
|
| 942 |
+
{
|
| 943 |
+
"epoch": 1.0431484115694643,
|
| 944 |
+
"grad_norm": 1.481063723564148,
|
| 945 |
+
"learning_rate": 3.3147583105829456e-05,
|
| 946 |
+
"loss": 1.2646,
|
| 947 |
+
"step": 11000
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 1.0431484115694643,
|
| 951 |
+
"eval_loss": 1.2060637474060059,
|
| 952 |
+
"eval_runtime": 71.9819,
|
| 953 |
+
"eval_samples_per_second": 130.227,
|
| 954 |
+
"eval_steps_per_second": 16.282,
|
| 955 |
+
"step": 11000
|
| 956 |
+
},
|
| 957 |
+
{
|
| 958 |
+
"epoch": 1.0526315789473684,
|
| 959 |
+
"grad_norm": 1.699491024017334,
|
| 960 |
+
"learning_rate": 3.298699213104224e-05,
|
| 961 |
+
"loss": 1.2828,
|
| 962 |
+
"step": 11100
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 1.0621147463252727,
|
| 966 |
+
"grad_norm": 2.0708415508270264,
|
| 967 |
+
"learning_rate": 3.282640115625502e-05,
|
| 968 |
+
"loss": 1.2648,
|
| 969 |
+
"step": 11200
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 1.0715979137031768,
|
| 973 |
+
"grad_norm": 1.4921772480010986,
|
| 974 |
+
"learning_rate": 3.266741609121567e-05,
|
| 975 |
+
"loss": 1.2611,
|
| 976 |
+
"step": 11300
|
| 977 |
+
},
|
| 978 |
+
{
|
| 979 |
+
"epoch": 1.0810810810810811,
|
| 980 |
+
"grad_norm": 1.744384765625,
|
| 981 |
+
"learning_rate": 3.250682511642846e-05,
|
| 982 |
+
"loss": 1.2435,
|
| 983 |
+
"step": 11400
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 1.0905642484589853,
|
| 987 |
+
"grad_norm": 1.1988921165466309,
|
| 988 |
+
"learning_rate": 3.234623414164124e-05,
|
| 989 |
+
"loss": 1.2525,
|
| 990 |
+
"step": 11500
|
| 991 |
+
},
|
| 992 |
+
{
|
| 993 |
+
"epoch": 1.0905642484589853,
|
| 994 |
+
"eval_loss": 1.2018728256225586,
|
| 995 |
+
"eval_runtime": 71.9385,
|
| 996 |
+
"eval_samples_per_second": 130.306,
|
| 997 |
+
"eval_steps_per_second": 16.292,
|
| 998 |
+
"step": 11500
|
| 999 |
+
},
|
| 1000 |
+
{
|
| 1001 |
+
"epoch": 1.1000474158368896,
|
| 1002 |
+
"grad_norm": 1.5618336200714111,
|
| 1003 |
+
"learning_rate": 3.218564316685402e-05,
|
| 1004 |
+
"loss": 1.2387,
|
| 1005 |
+
"step": 11600
|
| 1006 |
+
},
|
| 1007 |
+
{
|
| 1008 |
+
"epoch": 1.1095305832147937,
|
| 1009 |
+
"grad_norm": 1.512651801109314,
|
| 1010 |
+
"learning_rate": 3.202505219206681e-05,
|
| 1011 |
+
"loss": 1.2507,
|
| 1012 |
+
"step": 11700
|
| 1013 |
+
},
|
| 1014 |
+
{
|
| 1015 |
+
"epoch": 1.119013750592698,
|
| 1016 |
+
"grad_norm": 2.1945042610168457,
|
| 1017 |
+
"learning_rate": 3.186446121727959e-05,
|
| 1018 |
+
"loss": 1.2316,
|
| 1019 |
+
"step": 11800
|
| 1020 |
+
},
|
| 1021 |
+
{
|
| 1022 |
+
"epoch": 1.1284969179706021,
|
| 1023 |
+
"grad_norm": 1.3046265840530396,
|
| 1024 |
+
"learning_rate": 3.170387024249237e-05,
|
| 1025 |
+
"loss": 1.2352,
|
| 1026 |
+
"step": 11900
|
| 1027 |
+
},
|
| 1028 |
+
{
|
| 1029 |
+
"epoch": 1.1379800853485065,
|
| 1030 |
+
"grad_norm": 1.5922869443893433,
|
| 1031 |
+
"learning_rate": 3.154327926770516e-05,
|
| 1032 |
+
"loss": 1.2361,
|
| 1033 |
+
"step": 12000
|
| 1034 |
+
},
|
| 1035 |
+
{
|
| 1036 |
+
"epoch": 1.1379800853485065,
|
| 1037 |
+
"eval_loss": 1.1982355117797852,
|
| 1038 |
+
"eval_runtime": 72.0166,
|
| 1039 |
+
"eval_samples_per_second": 130.164,
|
| 1040 |
+
"eval_steps_per_second": 16.274,
|
| 1041 |
+
"step": 12000
|
| 1042 |
+
},
|
| 1043 |
+
{
|
| 1044 |
+
"epoch": 1.1474632527264106,
|
| 1045 |
+
"grad_norm": 1.2342475652694702,
|
| 1046 |
+
"learning_rate": 3.138268829291794e-05,
|
| 1047 |
+
"loss": 1.2318,
|
| 1048 |
+
"step": 12100
|
| 1049 |
+
},
|
| 1050 |
+
{
|
| 1051 |
+
"epoch": 1.156946420104315,
|
| 1052 |
+
"grad_norm": 1.630129337310791,
|
| 1053 |
+
"learning_rate": 3.122209731813072e-05,
|
| 1054 |
+
"loss": 1.2185,
|
| 1055 |
+
"step": 12200
|
| 1056 |
+
},
|
| 1057 |
+
{
|
| 1058 |
+
"epoch": 1.166429587482219,
|
| 1059 |
+
"grad_norm": 1.4030356407165527,
|
| 1060 |
+
"learning_rate": 3.106150634334351e-05,
|
| 1061 |
+
"loss": 1.2635,
|
| 1062 |
+
"step": 12300
|
| 1063 |
+
},
|
| 1064 |
+
{
|
| 1065 |
+
"epoch": 1.1759127548601234,
|
| 1066 |
+
"grad_norm": 1.372003436088562,
|
| 1067 |
+
"learning_rate": 3.090091536855629e-05,
|
| 1068 |
+
"loss": 1.2131,
|
| 1069 |
+
"step": 12400
|
| 1070 |
+
},
|
| 1071 |
+
{
|
| 1072 |
+
"epoch": 1.1853959222380275,
|
| 1073 |
+
"grad_norm": 1.1380951404571533,
|
| 1074 |
+
"learning_rate": 3.074032439376907e-05,
|
| 1075 |
+
"loss": 1.2553,
|
| 1076 |
+
"step": 12500
|
| 1077 |
+
},
|
| 1078 |
+
{
|
| 1079 |
+
"epoch": 1.1853959222380275,
|
| 1080 |
+
"eval_loss": 1.1942973136901855,
|
| 1081 |
+
"eval_runtime": 71.9892,
|
| 1082 |
+
"eval_samples_per_second": 130.214,
|
| 1083 |
+
"eval_steps_per_second": 16.28,
|
| 1084 |
+
"step": 12500
|
| 1085 |
+
},
|
| 1086 |
+
{
|
| 1087 |
+
"epoch": 1.1948790896159318,
|
| 1088 |
+
"grad_norm": 1.8760716915130615,
|
| 1089 |
+
"learning_rate": 3.057973341898186e-05,
|
| 1090 |
+
"loss": 1.2479,
|
| 1091 |
+
"step": 12600
|
| 1092 |
+
},
|
| 1093 |
+
{
|
| 1094 |
+
"epoch": 1.204362256993836,
|
| 1095 |
+
"grad_norm": 1.7070045471191406,
|
| 1096 |
+
"learning_rate": 3.0419142444194638e-05,
|
| 1097 |
+
"loss": 1.2283,
|
| 1098 |
+
"step": 12700
|
| 1099 |
+
},
|
| 1100 |
+
{
|
| 1101 |
+
"epoch": 1.2138454243717403,
|
| 1102 |
+
"grad_norm": 1.6677838563919067,
|
| 1103 |
+
"learning_rate": 3.025855146940742e-05,
|
| 1104 |
+
"loss": 1.2527,
|
| 1105 |
+
"step": 12800
|
| 1106 |
+
},
|
| 1107 |
+
{
|
| 1108 |
+
"epoch": 1.2233285917496444,
|
| 1109 |
+
"grad_norm": 1.5015747547149658,
|
| 1110 |
+
"learning_rate": 3.0097960494620203e-05,
|
| 1111 |
+
"loss": 1.2402,
|
| 1112 |
+
"step": 12900
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 1.2328117591275487,
|
| 1116 |
+
"grad_norm": 1.613587737083435,
|
| 1117 |
+
"learning_rate": 2.9937369519832987e-05,
|
| 1118 |
+
"loss": 1.2288,
|
| 1119 |
+
"step": 13000
|
| 1120 |
+
},
|
| 1121 |
+
{
|
| 1122 |
+
"epoch": 1.2328117591275487,
|
| 1123 |
+
"eval_loss": 1.1904593706130981,
|
| 1124 |
+
"eval_runtime": 72.0827,
|
| 1125 |
+
"eval_samples_per_second": 130.045,
|
| 1126 |
+
"eval_steps_per_second": 16.259,
|
| 1127 |
+
"step": 13000
|
| 1128 |
+
},
|
| 1129 |
+
{
|
| 1130 |
+
"epoch": 1.2422949265054528,
|
| 1131 |
+
"grad_norm": 1.7170720100402832,
|
| 1132 |
+
"learning_rate": 2.9776778545045768e-05,
|
| 1133 |
+
"loss": 1.2199,
|
| 1134 |
+
"step": 13100
|
| 1135 |
+
},
|
| 1136 |
+
{
|
| 1137 |
+
"epoch": 1.251778093883357,
|
| 1138 |
+
"grad_norm": 1.3260998725891113,
|
| 1139 |
+
"learning_rate": 2.9616187570258552e-05,
|
| 1140 |
+
"loss": 1.2575,
|
| 1141 |
+
"step": 13200
|
| 1142 |
+
},
|
| 1143 |
+
{
|
| 1144 |
+
"epoch": 1.2612612612612613,
|
| 1145 |
+
"grad_norm": 1.450626254081726,
|
| 1146 |
+
"learning_rate": 2.9455596595471337e-05,
|
| 1147 |
+
"loss": 1.2267,
|
| 1148 |
+
"step": 13300
|
| 1149 |
+
},
|
| 1150 |
+
{
|
| 1151 |
+
"epoch": 1.2707444286391656,
|
| 1152 |
+
"grad_norm": 1.51180899143219,
|
| 1153 |
+
"learning_rate": 2.9295005620684118e-05,
|
| 1154 |
+
"loss": 1.2546,
|
| 1155 |
+
"step": 13400
|
| 1156 |
+
},
|
| 1157 |
+
{
|
| 1158 |
+
"epoch": 1.2802275960170697,
|
| 1159 |
+
"grad_norm": 1.846704125404358,
|
| 1160 |
+
"learning_rate": 2.9134414645896902e-05,
|
| 1161 |
+
"loss": 1.2216,
|
| 1162 |
+
"step": 13500
|
| 1163 |
+
},
|
| 1164 |
+
{
|
| 1165 |
+
"epoch": 1.2802275960170697,
|
| 1166 |
+
"eval_loss": 1.1853208541870117,
|
| 1167 |
+
"eval_runtime": 72.0024,
|
| 1168 |
+
"eval_samples_per_second": 130.19,
|
| 1169 |
+
"eval_steps_per_second": 16.277,
|
| 1170 |
+
"step": 13500
|
| 1171 |
+
},
|
| 1172 |
+
{
|
| 1173 |
+
"epoch": 1.2897107633949738,
|
| 1174 |
+
"grad_norm": 1.5088779926300049,
|
| 1175 |
+
"learning_rate": 2.8973823671109686e-05,
|
| 1176 |
+
"loss": 1.2028,
|
| 1177 |
+
"step": 13600
|
| 1178 |
+
},
|
| 1179 |
+
{
|
| 1180 |
+
"epoch": 1.2991939307728781,
|
| 1181 |
+
"grad_norm": 1.2047330141067505,
|
| 1182 |
+
"learning_rate": 2.8813232696322467e-05,
|
| 1183 |
+
"loss": 1.2326,
|
| 1184 |
+
"step": 13700
|
| 1185 |
+
},
|
| 1186 |
+
{
|
| 1187 |
+
"epoch": 1.3086770981507825,
|
| 1188 |
+
"grad_norm": 1.6895666122436523,
|
| 1189 |
+
"learning_rate": 2.865264172153525e-05,
|
| 1190 |
+
"loss": 1.2032,
|
| 1191 |
+
"step": 13800
|
| 1192 |
+
},
|
| 1193 |
+
{
|
| 1194 |
+
"epoch": 1.3181602655286866,
|
| 1195 |
+
"grad_norm": 1.3885574340820312,
|
| 1196 |
+
"learning_rate": 2.8492050746748032e-05,
|
| 1197 |
+
"loss": 1.2438,
|
| 1198 |
+
"step": 13900
|
| 1199 |
+
},
|
| 1200 |
+
{
|
| 1201 |
+
"epoch": 1.3276434329065907,
|
| 1202 |
+
"grad_norm": 1.5129587650299072,
|
| 1203 |
+
"learning_rate": 2.8331459771960816e-05,
|
| 1204 |
+
"loss": 1.2099,
|
| 1205 |
+
"step": 14000
|
| 1206 |
+
},
|
| 1207 |
+
{
|
| 1208 |
+
"epoch": 1.3276434329065907,
|
| 1209 |
+
"eval_loss": 1.1841365098953247,
|
| 1210 |
+
"eval_runtime": 72.0289,
|
| 1211 |
+
"eval_samples_per_second": 130.142,
|
| 1212 |
+
"eval_steps_per_second": 16.271,
|
| 1213 |
+
"step": 14000
|
| 1214 |
+
},
|
| 1215 |
+
{
|
| 1216 |
+
"epoch": 1.337126600284495,
|
| 1217 |
+
"grad_norm": 1.5244189500808716,
|
| 1218 |
+
"learning_rate": 2.81708687971736e-05,
|
| 1219 |
+
"loss": 1.2528,
|
| 1220 |
+
"step": 14100
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 1.3466097676623994,
|
| 1224 |
+
"grad_norm": 1.6656090021133423,
|
| 1225 |
+
"learning_rate": 2.801027782238638e-05,
|
| 1226 |
+
"loss": 1.2437,
|
| 1227 |
+
"step": 14200
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 1.3560929350403035,
|
| 1231 |
+
"grad_norm": 1.6365015506744385,
|
| 1232 |
+
"learning_rate": 2.7849686847599165e-05,
|
| 1233 |
+
"loss": 1.2481,
|
| 1234 |
+
"step": 14300
|
| 1235 |
+
},
|
| 1236 |
+
{
|
| 1237 |
+
"epoch": 1.3655761024182076,
|
| 1238 |
+
"grad_norm": 1.729038953781128,
|
| 1239 |
+
"learning_rate": 2.768909587281195e-05,
|
| 1240 |
+
"loss": 1.2363,
|
| 1241 |
+
"step": 14400
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 1.375059269796112,
|
| 1245 |
+
"grad_norm": 1.663041114807129,
|
| 1246 |
+
"learning_rate": 2.752850489802473e-05,
|
| 1247 |
+
"loss": 1.2371,
|
| 1248 |
+
"step": 14500
|
| 1249 |
+
},
|
| 1250 |
+
{
|
| 1251 |
+
"epoch": 1.375059269796112,
|
| 1252 |
+
"eval_loss": 1.1793495416641235,
|
| 1253 |
+
"eval_runtime": 72.0339,
|
| 1254 |
+
"eval_samples_per_second": 130.133,
|
| 1255 |
+
"eval_steps_per_second": 16.27,
|
| 1256 |
+
"step": 14500
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 1.384542437174016,
|
| 1260 |
+
"grad_norm": 1.5626816749572754,
|
| 1261 |
+
"learning_rate": 2.7367913923237515e-05,
|
| 1262 |
+
"loss": 1.2287,
|
| 1263 |
+
"step": 14600
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 1.3940256045519204,
|
| 1267 |
+
"grad_norm": 1.2476764917373657,
|
| 1268 |
+
"learning_rate": 2.72073229484503e-05,
|
| 1269 |
+
"loss": 1.2129,
|
| 1270 |
+
"step": 14700
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 1.4035087719298245,
|
| 1274 |
+
"grad_norm": 1.4796671867370605,
|
| 1275 |
+
"learning_rate": 2.704673197366308e-05,
|
| 1276 |
+
"loss": 1.2143,
|
| 1277 |
+
"step": 14800
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 1.4129919393077288,
|
| 1281 |
+
"grad_norm": 1.8260607719421387,
|
| 1282 |
+
"learning_rate": 2.6886140998875864e-05,
|
| 1283 |
+
"loss": 1.2411,
|
| 1284 |
+
"step": 14900
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 1.422475106685633,
|
| 1288 |
+
"grad_norm": 1.6393589973449707,
|
| 1289 |
+
"learning_rate": 2.6725550024088648e-05,
|
| 1290 |
+
"loss": 1.2128,
|
| 1291 |
+
"step": 15000
|
| 1292 |
+
},
|
| 1293 |
+
{
|
| 1294 |
+
"epoch": 1.422475106685633,
|
| 1295 |
+
"eval_loss": 1.1766639947891235,
|
| 1296 |
+
"eval_runtime": 72.0436,
|
| 1297 |
+
"eval_samples_per_second": 130.116,
|
| 1298 |
+
"eval_steps_per_second": 16.268,
|
| 1299 |
+
"step": 15000
|
| 1300 |
+
},
|
| 1301 |
+
{
|
| 1302 |
+
"epoch": 1.4319582740635373,
|
| 1303 |
+
"grad_norm": 1.2327754497528076,
|
| 1304 |
+
"learning_rate": 2.656495904930143e-05,
|
| 1305 |
+
"loss": 1.2218,
|
| 1306 |
+
"step": 15100
|
| 1307 |
+
},
|
| 1308 |
+
{
|
| 1309 |
+
"epoch": 1.4414414414414414,
|
| 1310 |
+
"grad_norm": 1.4845291376113892,
|
| 1311 |
+
"learning_rate": 2.6405973984262084e-05,
|
| 1312 |
+
"loss": 1.2158,
|
| 1313 |
+
"step": 15200
|
| 1314 |
+
},
|
| 1315 |
+
{
|
| 1316 |
+
"epoch": 1.4509246088193457,
|
| 1317 |
+
"grad_norm": 1.5115349292755127,
|
| 1318 |
+
"learning_rate": 2.6245383009474868e-05,
|
| 1319 |
+
"loss": 1.2597,
|
| 1320 |
+
"step": 15300
|
| 1321 |
+
},
|
| 1322 |
+
{
|
| 1323 |
+
"epoch": 1.4604077761972498,
|
| 1324 |
+
"grad_norm": 1.2558484077453613,
|
| 1325 |
+
"learning_rate": 2.608479203468765e-05,
|
| 1326 |
+
"loss": 1.2293,
|
| 1327 |
+
"step": 15400
|
| 1328 |
+
},
|
| 1329 |
+
{
|
| 1330 |
+
"epoch": 1.4698909435751542,
|
| 1331 |
+
"grad_norm": 1.412372350692749,
|
| 1332 |
+
"learning_rate": 2.5924201059900433e-05,
|
| 1333 |
+
"loss": 1.2078,
|
| 1334 |
+
"step": 15500
|
| 1335 |
+
},
|
| 1336 |
+
{
|
| 1337 |
+
"epoch": 1.4698909435751542,
|
| 1338 |
+
"eval_loss": 1.175757646560669,
|
| 1339 |
+
"eval_runtime": 72.1719,
|
| 1340 |
+
"eval_samples_per_second": 129.884,
|
| 1341 |
+
"eval_steps_per_second": 16.239,
|
| 1342 |
+
"step": 15500
|
| 1343 |
+
},
|
| 1344 |
+
{
|
| 1345 |
+
"epoch": 1.4793741109530583,
|
| 1346 |
+
"grad_norm": 1.1586443185806274,
|
| 1347 |
+
"learning_rate": 2.5763610085113217e-05,
|
| 1348 |
+
"loss": 1.2167,
|
| 1349 |
+
"step": 15600
|
| 1350 |
+
},
|
| 1351 |
+
{
|
| 1352 |
+
"epoch": 1.4888572783309626,
|
| 1353 |
+
"grad_norm": 1.535499095916748,
|
| 1354 |
+
"learning_rate": 2.5603019110325998e-05,
|
| 1355 |
+
"loss": 1.2177,
|
| 1356 |
+
"step": 15700
|
| 1357 |
+
},
|
| 1358 |
+
{
|
| 1359 |
+
"epoch": 1.4983404457088667,
|
| 1360 |
+
"grad_norm": 1.3925201892852783,
|
| 1361 |
+
"learning_rate": 2.5442428135538782e-05,
|
| 1362 |
+
"loss": 1.2089,
|
| 1363 |
+
"step": 15800
|
| 1364 |
+
},
|
| 1365 |
+
{
|
| 1366 |
+
"epoch": 1.5078236130867708,
|
| 1367 |
+
"grad_norm": 1.239797592163086,
|
| 1368 |
+
"learning_rate": 2.5281837160751563e-05,
|
| 1369 |
+
"loss": 1.2183,
|
| 1370 |
+
"step": 15900
|
| 1371 |
+
},
|
| 1372 |
+
{
|
| 1373 |
+
"epoch": 1.5173067804646752,
|
| 1374 |
+
"grad_norm": 1.4727925062179565,
|
| 1375 |
+
"learning_rate": 2.5121246185964347e-05,
|
| 1376 |
+
"loss": 1.2382,
|
| 1377 |
+
"step": 16000
|
| 1378 |
+
},
|
| 1379 |
+
{
|
| 1380 |
+
"epoch": 1.5173067804646752,
|
| 1381 |
+
"eval_loss": 1.1705734729766846,
|
| 1382 |
+
"eval_runtime": 72.2315,
|
| 1383 |
+
"eval_samples_per_second": 129.777,
|
| 1384 |
+
"eval_steps_per_second": 16.226,
|
| 1385 |
+
"step": 16000
|
| 1386 |
+
},
|
| 1387 |
+
{
|
| 1388 |
+
"epoch": 1.5267899478425795,
|
| 1389 |
+
"grad_norm": 1.9122114181518555,
|
| 1390 |
+
"learning_rate": 2.4960655211177135e-05,
|
| 1391 |
+
"loss": 1.2062,
|
| 1392 |
+
"step": 16100
|
| 1393 |
+
},
|
| 1394 |
+
{
|
| 1395 |
+
"epoch": 1.5362731152204836,
|
| 1396 |
+
"grad_norm": 1.705417275428772,
|
| 1397 |
+
"learning_rate": 2.4800064236389916e-05,
|
| 1398 |
+
"loss": 1.2002,
|
| 1399 |
+
"step": 16200
|
| 1400 |
+
},
|
| 1401 |
+
{
|
| 1402 |
+
"epoch": 1.5457562825983877,
|
| 1403 |
+
"grad_norm": 1.4141908884048462,
|
| 1404 |
+
"learning_rate": 2.46394732616027e-05,
|
| 1405 |
+
"loss": 1.2323,
|
| 1406 |
+
"step": 16300
|
| 1407 |
+
},
|
| 1408 |
+
{
|
| 1409 |
+
"epoch": 1.555239449976292,
|
| 1410 |
+
"grad_norm": 2.050583839416504,
|
| 1411 |
+
"learning_rate": 2.4478882286815484e-05,
|
| 1412 |
+
"loss": 1.2145,
|
| 1413 |
+
"step": 16400
|
| 1414 |
+
},
|
| 1415 |
+
{
|
| 1416 |
+
"epoch": 1.5647226173541964,
|
| 1417 |
+
"grad_norm": 1.495006799697876,
|
| 1418 |
+
"learning_rate": 2.4318291312028265e-05,
|
| 1419 |
+
"loss": 1.2041,
|
| 1420 |
+
"step": 16500
|
| 1421 |
+
},
|
| 1422 |
+
{
|
| 1423 |
+
"epoch": 1.5647226173541964,
|
| 1424 |
+
"eval_loss": 1.1694616079330444,
|
| 1425 |
+
"eval_runtime": 71.9712,
|
| 1426 |
+
"eval_samples_per_second": 130.247,
|
| 1427 |
+
"eval_steps_per_second": 16.284,
|
| 1428 |
+
"step": 16500
|
| 1429 |
+
},
|
| 1430 |
+
{
|
| 1431 |
+
"epoch": 1.5742057847321005,
|
| 1432 |
+
"grad_norm": 1.4379011392593384,
|
| 1433 |
+
"learning_rate": 2.415770033724105e-05,
|
| 1434 |
+
"loss": 1.2045,
|
| 1435 |
+
"step": 16600
|
| 1436 |
+
},
|
| 1437 |
+
{
|
| 1438 |
+
"epoch": 1.5836889521100046,
|
| 1439 |
+
"grad_norm": 1.6558938026428223,
|
| 1440 |
+
"learning_rate": 2.399710936245383e-05,
|
| 1441 |
+
"loss": 1.2234,
|
| 1442 |
+
"step": 16700
|
| 1443 |
+
},
|
| 1444 |
+
{
|
| 1445 |
+
"epoch": 1.593172119487909,
|
| 1446 |
+
"grad_norm": 1.6931570768356323,
|
| 1447 |
+
"learning_rate": 2.3836518387666614e-05,
|
| 1448 |
+
"loss": 1.2061,
|
| 1449 |
+
"step": 16800
|
| 1450 |
+
},
|
| 1451 |
+
{
|
| 1452 |
+
"epoch": 1.6026552868658133,
|
| 1453 |
+
"grad_norm": 1.445521593093872,
|
| 1454 |
+
"learning_rate": 2.36759274128794e-05,
|
| 1455 |
+
"loss": 1.2243,
|
| 1456 |
+
"step": 16900
|
| 1457 |
+
},
|
| 1458 |
+
{
|
| 1459 |
+
"epoch": 1.6121384542437174,
|
| 1460 |
+
"grad_norm": 1.4067689180374146,
|
| 1461 |
+
"learning_rate": 2.351533643809218e-05,
|
| 1462 |
+
"loss": 1.2154,
|
| 1463 |
+
"step": 17000
|
| 1464 |
+
},
|
| 1465 |
+
{
|
| 1466 |
+
"epoch": 1.6121384542437174,
|
| 1467 |
+
"eval_loss": 1.1659753322601318,
|
| 1468 |
+
"eval_runtime": 72.1888,
|
| 1469 |
+
"eval_samples_per_second": 129.854,
|
| 1470 |
+
"eval_steps_per_second": 16.235,
|
| 1471 |
+
"step": 17000
|
| 1472 |
+
},
|
| 1473 |
+
{
|
| 1474 |
+
"epoch": 1.6216216216216215,
|
| 1475 |
+
"grad_norm": 1.0550585985183716,
|
| 1476 |
+
"learning_rate": 2.3354745463304964e-05,
|
| 1477 |
+
"loss": 1.2333,
|
| 1478 |
+
"step": 17100
|
| 1479 |
+
},
|
| 1480 |
+
{
|
| 1481 |
+
"epoch": 1.6311047889995258,
|
| 1482 |
+
"grad_norm": 1.5547784566879272,
|
| 1483 |
+
"learning_rate": 2.3194154488517748e-05,
|
| 1484 |
+
"loss": 1.2088,
|
| 1485 |
+
"step": 17200
|
| 1486 |
+
},
|
| 1487 |
+
{
|
| 1488 |
+
"epoch": 1.6405879563774302,
|
| 1489 |
+
"grad_norm": 2.006110191345215,
|
| 1490 |
+
"learning_rate": 2.303356351373053e-05,
|
| 1491 |
+
"loss": 1.1881,
|
| 1492 |
+
"step": 17300
|
| 1493 |
+
},
|
| 1494 |
+
{
|
| 1495 |
+
"epoch": 1.6500711237553343,
|
| 1496 |
+
"grad_norm": 1.6522830724716187,
|
| 1497 |
+
"learning_rate": 2.2872972538943313e-05,
|
| 1498 |
+
"loss": 1.2158,
|
| 1499 |
+
"step": 17400
|
| 1500 |
+
},
|
| 1501 |
+
{
|
| 1502 |
+
"epoch": 1.6595542911332384,
|
| 1503 |
+
"grad_norm": 1.2928231954574585,
|
| 1504 |
+
"learning_rate": 2.2712381564156097e-05,
|
| 1505 |
+
"loss": 1.2303,
|
| 1506 |
+
"step": 17500
|
| 1507 |
+
},
|
| 1508 |
+
{
|
| 1509 |
+
"epoch": 1.6595542911332384,
|
| 1510 |
+
"eval_loss": 1.1643718481063843,
|
| 1511 |
+
"eval_runtime": 72.2381,
|
| 1512 |
+
"eval_samples_per_second": 129.765,
|
| 1513 |
+
"eval_steps_per_second": 16.224,
|
| 1514 |
+
"step": 17500
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 1.6690374585111427,
|
| 1518 |
+
"grad_norm": 1.38106107711792,
|
| 1519 |
+
"learning_rate": 2.2551790589368878e-05,
|
| 1520 |
+
"loss": 1.1969,
|
| 1521 |
+
"step": 17600
|
| 1522 |
+
},
|
| 1523 |
+
{
|
| 1524 |
+
"epoch": 1.678520625889047,
|
| 1525 |
+
"grad_norm": 1.3726710081100464,
|
| 1526 |
+
"learning_rate": 2.2391199614581662e-05,
|
| 1527 |
+
"loss": 1.2122,
|
| 1528 |
+
"step": 17700
|
| 1529 |
+
},
|
| 1530 |
+
{
|
| 1531 |
+
"epoch": 1.6880037932669512,
|
| 1532 |
+
"grad_norm": 1.2017816305160522,
|
| 1533 |
+
"learning_rate": 2.2230608639794447e-05,
|
| 1534 |
+
"loss": 1.2331,
|
| 1535 |
+
"step": 17800
|
| 1536 |
+
},
|
| 1537 |
+
{
|
| 1538 |
+
"epoch": 1.6974869606448553,
|
| 1539 |
+
"grad_norm": 1.329315423965454,
|
| 1540 |
+
"learning_rate": 2.2070017665007227e-05,
|
| 1541 |
+
"loss": 1.2339,
|
| 1542 |
+
"step": 17900
|
| 1543 |
+
},
|
| 1544 |
+
{
|
| 1545 |
+
"epoch": 1.7069701280227596,
|
| 1546 |
+
"grad_norm": 1.5352445840835571,
|
| 1547 |
+
"learning_rate": 2.190942669022001e-05,
|
| 1548 |
+
"loss": 1.2429,
|
| 1549 |
+
"step": 18000
|
| 1550 |
+
},
|
| 1551 |
+
{
|
| 1552 |
+
"epoch": 1.7069701280227596,
|
| 1553 |
+
"eval_loss": 1.1619985103607178,
|
| 1554 |
+
"eval_runtime": 72.1286,
|
| 1555 |
+
"eval_samples_per_second": 129.962,
|
| 1556 |
+
"eval_steps_per_second": 16.249,
|
| 1557 |
+
"step": 18000
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 1.716453295400664,
|
| 1561 |
+
"grad_norm": 1.5836015939712524,
|
| 1562 |
+
"learning_rate": 2.1748835715432796e-05,
|
| 1563 |
+
"loss": 1.1925,
|
| 1564 |
+
"step": 18100
|
| 1565 |
+
},
|
| 1566 |
+
{
|
| 1567 |
+
"epoch": 1.725936462778568,
|
| 1568 |
+
"grad_norm": 1.7755178213119507,
|
| 1569 |
+
"learning_rate": 2.1588244740645577e-05,
|
| 1570 |
+
"loss": 1.2146,
|
| 1571 |
+
"step": 18200
|
| 1572 |
+
},
|
| 1573 |
+
{
|
| 1574 |
+
"epoch": 1.7354196301564722,
|
| 1575 |
+
"grad_norm": 1.3868217468261719,
|
| 1576 |
+
"learning_rate": 2.142765376585836e-05,
|
| 1577 |
+
"loss": 1.2082,
|
| 1578 |
+
"step": 18300
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 1.7449027975343765,
|
| 1582 |
+
"grad_norm": 1.320333480834961,
|
| 1583 |
+
"learning_rate": 2.1267062791071142e-05,
|
| 1584 |
+
"loss": 1.213,
|
| 1585 |
+
"step": 18400
|
| 1586 |
+
},
|
| 1587 |
+
{
|
| 1588 |
+
"epoch": 1.7543859649122808,
|
| 1589 |
+
"grad_norm": 1.5032850503921509,
|
| 1590 |
+
"learning_rate": 2.1106471816283926e-05,
|
| 1591 |
+
"loss": 1.2048,
|
| 1592 |
+
"step": 18500
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 1.7543859649122808,
|
| 1596 |
+
"eval_loss": 1.1578137874603271,
|
| 1597 |
+
"eval_runtime": 72.0841,
|
| 1598 |
+
"eval_samples_per_second": 130.043,
|
| 1599 |
+
"eval_steps_per_second": 16.259,
|
| 1600 |
+
"step": 18500
|
| 1601 |
+
},
|
| 1602 |
+
{
|
| 1603 |
+
"epoch": 1.763869132290185,
|
| 1604 |
+
"grad_norm": 1.5423904657363892,
|
| 1605 |
+
"learning_rate": 2.094588084149671e-05,
|
| 1606 |
+
"loss": 1.2282,
|
| 1607 |
+
"step": 18600
|
| 1608 |
+
},
|
| 1609 |
+
{
|
| 1610 |
+
"epoch": 1.773352299668089,
|
| 1611 |
+
"grad_norm": 1.439765453338623,
|
| 1612 |
+
"learning_rate": 2.078528986670949e-05,
|
| 1613 |
+
"loss": 1.2171,
|
| 1614 |
+
"step": 18700
|
| 1615 |
+
},
|
| 1616 |
+
{
|
| 1617 |
+
"epoch": 1.7828354670459934,
|
| 1618 |
+
"grad_norm": 1.573088526725769,
|
| 1619 |
+
"learning_rate": 2.0624698891922275e-05,
|
| 1620 |
+
"loss": 1.2149,
|
| 1621 |
+
"step": 18800
|
| 1622 |
+
},
|
| 1623 |
+
{
|
| 1624 |
+
"epoch": 1.7923186344238977,
|
| 1625 |
+
"grad_norm": 1.4882514476776123,
|
| 1626 |
+
"learning_rate": 2.046410791713506e-05,
|
| 1627 |
+
"loss": 1.2278,
|
| 1628 |
+
"step": 18900
|
| 1629 |
+
},
|
| 1630 |
+
{
|
| 1631 |
+
"epoch": 1.8018018018018018,
|
| 1632 |
+
"grad_norm": 1.9028195142745972,
|
| 1633 |
+
"learning_rate": 2.030351694234784e-05,
|
| 1634 |
+
"loss": 1.2247,
|
| 1635 |
+
"step": 19000
|
| 1636 |
+
},
|
| 1637 |
+
{
|
| 1638 |
+
"epoch": 1.8018018018018018,
|
| 1639 |
+
"eval_loss": 1.157362937927246,
|
| 1640 |
+
"eval_runtime": 72.1036,
|
| 1641 |
+
"eval_samples_per_second": 130.007,
|
| 1642 |
+
"eval_steps_per_second": 16.254,
|
| 1643 |
+
"step": 19000
|
| 1644 |
+
},
|
| 1645 |
+
{
|
| 1646 |
+
"epoch": 1.811284969179706,
|
| 1647 |
+
"grad_norm": 1.289600133895874,
|
| 1648 |
+
"learning_rate": 2.0142925967560625e-05,
|
| 1649 |
+
"loss": 1.215,
|
| 1650 |
+
"step": 19100
|
| 1651 |
+
},
|
| 1652 |
+
{
|
| 1653 |
+
"epoch": 1.8207681365576103,
|
| 1654 |
+
"grad_norm": 1.4183131456375122,
|
| 1655 |
+
"learning_rate": 1.998233499277341e-05,
|
| 1656 |
+
"loss": 1.2284,
|
| 1657 |
+
"step": 19200
|
| 1658 |
+
},
|
| 1659 |
+
{
|
| 1660 |
+
"epoch": 1.8302513039355146,
|
| 1661 |
+
"grad_norm": 1.235146403312683,
|
| 1662 |
+
"learning_rate": 1.982174401798619e-05,
|
| 1663 |
+
"loss": 1.2067,
|
| 1664 |
+
"step": 19300
|
| 1665 |
+
},
|
| 1666 |
+
{
|
| 1667 |
+
"epoch": 1.8397344713134187,
|
| 1668 |
+
"grad_norm": 1.486122488975525,
|
| 1669 |
+
"learning_rate": 1.9661153043198974e-05,
|
| 1670 |
+
"loss": 1.183,
|
| 1671 |
+
"step": 19400
|
| 1672 |
+
},
|
| 1673 |
+
{
|
| 1674 |
+
"epoch": 1.8492176386913228,
|
| 1675 |
+
"grad_norm": 1.4615782499313354,
|
| 1676 |
+
"learning_rate": 1.9500562068411758e-05,
|
| 1677 |
+
"loss": 1.1847,
|
| 1678 |
+
"step": 19500
|
| 1679 |
+
},
|
| 1680 |
+
{
|
| 1681 |
+
"epoch": 1.8492176386913228,
|
| 1682 |
+
"eval_loss": 1.1544617414474487,
|
| 1683 |
+
"eval_runtime": 72.1411,
|
| 1684 |
+
"eval_samples_per_second": 129.94,
|
| 1685 |
+
"eval_steps_per_second": 16.246,
|
| 1686 |
+
"step": 19500
|
| 1687 |
+
},
|
| 1688 |
+
{
|
| 1689 |
+
"epoch": 1.8587008060692272,
|
| 1690 |
+
"grad_norm": 1.3062597513198853,
|
| 1691 |
+
"learning_rate": 1.933997109362454e-05,
|
| 1692 |
+
"loss": 1.1998,
|
| 1693 |
+
"step": 19600
|
| 1694 |
+
},
|
| 1695 |
+
{
|
| 1696 |
+
"epoch": 1.8681839734471315,
|
| 1697 |
+
"grad_norm": 1.7676483392715454,
|
| 1698 |
+
"learning_rate": 1.9180986028585193e-05,
|
| 1699 |
+
"loss": 1.1985,
|
| 1700 |
+
"step": 19700
|
| 1701 |
+
},
|
| 1702 |
+
{
|
| 1703 |
+
"epoch": 1.8776671408250356,
|
| 1704 |
+
"grad_norm": 1.55678129196167,
|
| 1705 |
+
"learning_rate": 1.9020395053797978e-05,
|
| 1706 |
+
"loss": 1.2155,
|
| 1707 |
+
"step": 19800
|
| 1708 |
+
},
|
| 1709 |
+
{
|
| 1710 |
+
"epoch": 1.8871503082029397,
|
| 1711 |
+
"grad_norm": 1.2260453701019287,
|
| 1712 |
+
"learning_rate": 1.885980407901076e-05,
|
| 1713 |
+
"loss": 1.2282,
|
| 1714 |
+
"step": 19900
|
| 1715 |
+
},
|
| 1716 |
+
{
|
| 1717 |
+
"epoch": 1.896633475580844,
|
| 1718 |
+
"grad_norm": 1.6828114986419678,
|
| 1719 |
+
"learning_rate": 1.8699213104223543e-05,
|
| 1720 |
+
"loss": 1.2183,
|
| 1721 |
+
"step": 20000
|
| 1722 |
+
},
|
| 1723 |
+
{
|
| 1724 |
+
"epoch": 1.896633475580844,
|
| 1725 |
+
"eval_loss": 1.1521168947219849,
|
| 1726 |
+
"eval_runtime": 72.1018,
|
| 1727 |
+
"eval_samples_per_second": 130.011,
|
| 1728 |
+
"eval_steps_per_second": 16.255,
|
| 1729 |
+
"step": 20000
|
| 1730 |
+
},
|
| 1731 |
+
{
|
| 1732 |
+
"epoch": 1.9061166429587484,
|
| 1733 |
+
"grad_norm": 1.6691786050796509,
|
| 1734 |
+
"learning_rate": 1.8538622129436327e-05,
|
| 1735 |
+
"loss": 1.1651,
|
| 1736 |
+
"step": 20100
|
| 1737 |
+
},
|
| 1738 |
+
{
|
| 1739 |
+
"epoch": 1.9155998103366523,
|
| 1740 |
+
"grad_norm": 1.4728951454162598,
|
| 1741 |
+
"learning_rate": 1.8378031154649108e-05,
|
| 1742 |
+
"loss": 1.2022,
|
| 1743 |
+
"step": 20200
|
| 1744 |
+
},
|
| 1745 |
+
{
|
| 1746 |
+
"epoch": 1.9250829777145566,
|
| 1747 |
+
"grad_norm": 1.6341995000839233,
|
| 1748 |
+
"learning_rate": 1.8217440179861892e-05,
|
| 1749 |
+
"loss": 1.1777,
|
| 1750 |
+
"step": 20300
|
| 1751 |
+
},
|
| 1752 |
+
{
|
| 1753 |
+
"epoch": 1.934566145092461,
|
| 1754 |
+
"grad_norm": 1.4492669105529785,
|
| 1755 |
+
"learning_rate": 1.8056849205074676e-05,
|
| 1756 |
+
"loss": 1.2081,
|
| 1757 |
+
"step": 20400
|
| 1758 |
+
},
|
| 1759 |
+
{
|
| 1760 |
+
"epoch": 1.944049312470365,
|
| 1761 |
+
"grad_norm": 1.6642097234725952,
|
| 1762 |
+
"learning_rate": 1.7896258230287457e-05,
|
| 1763 |
+
"loss": 1.1848,
|
| 1764 |
+
"step": 20500
|
| 1765 |
+
},
|
| 1766 |
+
{
|
| 1767 |
+
"epoch": 1.944049312470365,
|
| 1768 |
+
"eval_loss": 1.150140404701233,
|
| 1769 |
+
"eval_runtime": 72.0779,
|
| 1770 |
+
"eval_samples_per_second": 130.054,
|
| 1771 |
+
"eval_steps_per_second": 16.26,
|
| 1772 |
+
"step": 20500
|
| 1773 |
+
},
|
| 1774 |
+
{
|
| 1775 |
+
"epoch": 1.9535324798482692,
|
| 1776 |
+
"grad_norm": 1.8986822366714478,
|
| 1777 |
+
"learning_rate": 1.773566725550024e-05,
|
| 1778 |
+
"loss": 1.2223,
|
| 1779 |
+
"step": 20600
|
| 1780 |
+
},
|
| 1781 |
+
{
|
| 1782 |
+
"epoch": 1.9630156472261735,
|
| 1783 |
+
"grad_norm": 1.390931248664856,
|
| 1784 |
+
"learning_rate": 1.7575076280713022e-05,
|
| 1785 |
+
"loss": 1.2068,
|
| 1786 |
+
"step": 20700
|
| 1787 |
+
},
|
| 1788 |
+
{
|
| 1789 |
+
"epoch": 1.9724988146040778,
|
| 1790 |
+
"grad_norm": 1.3856289386749268,
|
| 1791 |
+
"learning_rate": 1.7414485305925806e-05,
|
| 1792 |
+
"loss": 1.1828,
|
| 1793 |
+
"step": 20800
|
| 1794 |
+
},
|
| 1795 |
+
{
|
| 1796 |
+
"epoch": 1.981981981981982,
|
| 1797 |
+
"grad_norm": 1.2241305112838745,
|
| 1798 |
+
"learning_rate": 1.725389433113859e-05,
|
| 1799 |
+
"loss": 1.1938,
|
| 1800 |
+
"step": 20900
|
| 1801 |
+
},
|
| 1802 |
+
{
|
| 1803 |
+
"epoch": 1.991465149359886,
|
| 1804 |
+
"grad_norm": 1.5855077505111694,
|
| 1805 |
+
"learning_rate": 1.709330335635137e-05,
|
| 1806 |
+
"loss": 1.206,
|
| 1807 |
+
"step": 21000
|
| 1808 |
+
},
|
| 1809 |
+
{
|
| 1810 |
+
"epoch": 1.991465149359886,
|
| 1811 |
+
"eval_loss": 1.1497843265533447,
|
| 1812 |
+
"eval_runtime": 72.1674,
|
| 1813 |
+
"eval_samples_per_second": 129.893,
|
| 1814 |
+
"eval_steps_per_second": 16.24,
|
| 1815 |
+
"step": 21000
|
| 1816 |
+
},
|
| 1817 |
+
{
|
| 1818 |
+
"epoch": 2.0009483167377904,
|
| 1819 |
+
"grad_norm": 2.0832741260528564,
|
| 1820 |
+
"learning_rate": 1.6932712381564156e-05,
|
| 1821 |
+
"loss": 1.1805,
|
| 1822 |
+
"step": 21100
|
| 1823 |
+
},
|
| 1824 |
+
{
|
| 1825 |
+
"epoch": 2.0104314841156947,
|
| 1826 |
+
"grad_norm": 1.893350601196289,
|
| 1827 |
+
"learning_rate": 1.677212140677694e-05,
|
| 1828 |
+
"loss": 1.1757,
|
| 1829 |
+
"step": 21200
|
| 1830 |
+
},
|
| 1831 |
+
{
|
| 1832 |
+
"epoch": 2.019914651493599,
|
| 1833 |
+
"grad_norm": 1.346118688583374,
|
| 1834 |
+
"learning_rate": 1.661153043198972e-05,
|
| 1835 |
+
"loss": 1.1938,
|
| 1836 |
+
"step": 21300
|
| 1837 |
+
},
|
| 1838 |
+
{
|
| 1839 |
+
"epoch": 2.029397818871503,
|
| 1840 |
+
"grad_norm": 1.658034086227417,
|
| 1841 |
+
"learning_rate": 1.6450939457202505e-05,
|
| 1842 |
+
"loss": 1.1773,
|
| 1843 |
+
"step": 21400
|
| 1844 |
+
},
|
| 1845 |
+
{
|
| 1846 |
+
"epoch": 2.0388809862494073,
|
| 1847 |
+
"grad_norm": 1.4759783744812012,
|
| 1848 |
+
"learning_rate": 1.629034848241529e-05,
|
| 1849 |
+
"loss": 1.1735,
|
| 1850 |
+
"step": 21500
|
| 1851 |
+
},
|
| 1852 |
+
{
|
| 1853 |
+
"epoch": 2.0388809862494073,
|
| 1854 |
+
"eval_loss": 1.1474945545196533,
|
| 1855 |
+
"eval_runtime": 71.9179,
|
| 1856 |
+
"eval_samples_per_second": 130.343,
|
| 1857 |
+
"eval_steps_per_second": 16.296,
|
| 1858 |
+
"step": 21500
|
| 1859 |
+
},
|
| 1860 |
+
{
|
| 1861 |
+
"epoch": 2.0483641536273116,
|
| 1862 |
+
"grad_norm": 1.2887206077575684,
|
| 1863 |
+
"learning_rate": 1.612975750762807e-05,
|
| 1864 |
+
"loss": 1.1701,
|
| 1865 |
+
"step": 21600
|
| 1866 |
+
},
|
| 1867 |
+
{
|
| 1868 |
+
"epoch": 2.057847321005216,
|
| 1869 |
+
"grad_norm": 1.552646279335022,
|
| 1870 |
+
"learning_rate": 1.5969166532840854e-05,
|
| 1871 |
+
"loss": 1.1734,
|
| 1872 |
+
"step": 21700
|
| 1873 |
+
},
|
| 1874 |
+
{
|
| 1875 |
+
"epoch": 2.06733048838312,
|
| 1876 |
+
"grad_norm": 1.6683566570281982,
|
| 1877 |
+
"learning_rate": 1.581018146780151e-05,
|
| 1878 |
+
"loss": 1.1883,
|
| 1879 |
+
"step": 21800
|
| 1880 |
+
},
|
| 1881 |
+
{
|
| 1882 |
+
"epoch": 2.076813655761024,
|
| 1883 |
+
"grad_norm": 1.4613324403762817,
|
| 1884 |
+
"learning_rate": 1.5649590493014293e-05,
|
| 1885 |
+
"loss": 1.1845,
|
| 1886 |
+
"step": 21900
|
| 1887 |
+
},
|
| 1888 |
+
{
|
| 1889 |
+
"epoch": 2.0862968231389285,
|
| 1890 |
+
"grad_norm": 1.5622040033340454,
|
| 1891 |
+
"learning_rate": 1.5488999518227077e-05,
|
| 1892 |
+
"loss": 1.1584,
|
| 1893 |
+
"step": 22000
|
| 1894 |
+
},
|
| 1895 |
+
{
|
| 1896 |
+
"epoch": 2.0862968231389285,
|
| 1897 |
+
"eval_loss": 1.1467849016189575,
|
| 1898 |
+
"eval_runtime": 72.0497,
|
| 1899 |
+
"eval_samples_per_second": 130.105,
|
| 1900 |
+
"eval_steps_per_second": 16.267,
|
| 1901 |
+
"step": 22000
|
| 1902 |
+
},
|
| 1903 |
+
{
|
| 1904 |
+
"epoch": 2.095779990516833,
|
| 1905 |
+
"grad_norm": 1.721030831336975,
|
| 1906 |
+
"learning_rate": 1.5328408543439858e-05,
|
| 1907 |
+
"loss": 1.2018,
|
| 1908 |
+
"step": 22100
|
| 1909 |
+
},
|
| 1910 |
+
{
|
| 1911 |
+
"epoch": 2.1052631578947367,
|
| 1912 |
+
"grad_norm": 1.3872593641281128,
|
| 1913 |
+
"learning_rate": 1.5167817568652642e-05,
|
| 1914 |
+
"loss": 1.1659,
|
| 1915 |
+
"step": 22200
|
| 1916 |
+
},
|
| 1917 |
+
{
|
| 1918 |
+
"epoch": 2.114746325272641,
|
| 1919 |
+
"grad_norm": 1.655704140663147,
|
| 1920 |
+
"learning_rate": 1.5007226593865425e-05,
|
| 1921 |
+
"loss": 1.1503,
|
| 1922 |
+
"step": 22300
|
| 1923 |
+
},
|
| 1924 |
+
{
|
| 1925 |
+
"epoch": 2.1242294926505454,
|
| 1926 |
+
"grad_norm": 1.5672900676727295,
|
| 1927 |
+
"learning_rate": 1.4848241528826081e-05,
|
| 1928 |
+
"loss": 1.1879,
|
| 1929 |
+
"step": 22400
|
| 1930 |
+
},
|
| 1931 |
+
{
|
| 1932 |
+
"epoch": 2.1337126600284497,
|
| 1933 |
+
"grad_norm": 1.6815894842147827,
|
| 1934 |
+
"learning_rate": 1.4687650554038865e-05,
|
| 1935 |
+
"loss": 1.1719,
|
| 1936 |
+
"step": 22500
|
| 1937 |
+
},
|
| 1938 |
+
{
|
| 1939 |
+
"epoch": 2.1337126600284497,
|
| 1940 |
+
"eval_loss": 1.1450951099395752,
|
| 1941 |
+
"eval_runtime": 72.1598,
|
| 1942 |
+
"eval_samples_per_second": 129.906,
|
| 1943 |
+
"eval_steps_per_second": 16.242,
|
| 1944 |
+
"step": 22500
|
| 1945 |
+
},
|
| 1946 |
+
{
|
| 1947 |
+
"epoch": 2.1431958274063536,
|
| 1948 |
+
"grad_norm": 1.040648102760315,
|
| 1949 |
+
"learning_rate": 1.4527059579251648e-05,
|
| 1950 |
+
"loss": 1.1629,
|
| 1951 |
+
"step": 22600
|
| 1952 |
+
},
|
| 1953 |
+
{
|
| 1954 |
+
"epoch": 2.152678994784258,
|
| 1955 |
+
"grad_norm": 1.5001453161239624,
|
| 1956 |
+
"learning_rate": 1.436646860446443e-05,
|
| 1957 |
+
"loss": 1.1796,
|
| 1958 |
+
"step": 22700
|
| 1959 |
+
},
|
| 1960 |
+
{
|
| 1961 |
+
"epoch": 2.1621621621621623,
|
| 1962 |
+
"grad_norm": 1.7325968742370605,
|
| 1963 |
+
"learning_rate": 1.4205877629677215e-05,
|
| 1964 |
+
"loss": 1.1757,
|
| 1965 |
+
"step": 22800
|
| 1966 |
+
},
|
| 1967 |
+
{
|
| 1968 |
+
"epoch": 2.171645329540066,
|
| 1969 |
+
"grad_norm": 1.7485188245773315,
|
| 1970 |
+
"learning_rate": 1.4045286654889997e-05,
|
| 1971 |
+
"loss": 1.1485,
|
| 1972 |
+
"step": 22900
|
| 1973 |
+
},
|
| 1974 |
+
{
|
| 1975 |
+
"epoch": 2.1811284969179705,
|
| 1976 |
+
"grad_norm": 1.4972156286239624,
|
| 1977 |
+
"learning_rate": 1.388469568010278e-05,
|
| 1978 |
+
"loss": 1.1667,
|
| 1979 |
+
"step": 23000
|
| 1980 |
+
},
|
| 1981 |
+
{
|
| 1982 |
+
"epoch": 2.1811284969179705,
|
| 1983 |
+
"eval_loss": 1.144049048423767,
|
| 1984 |
+
"eval_runtime": 72.1218,
|
| 1985 |
+
"eval_samples_per_second": 129.975,
|
| 1986 |
+
"eval_steps_per_second": 16.25,
|
| 1987 |
+
"step": 23000
|
| 1988 |
+
},
|
| 1989 |
+
{
|
| 1990 |
+
"epoch": 2.190611664295875,
|
| 1991 |
+
"grad_norm": 1.2919082641601562,
|
| 1992 |
+
"learning_rate": 1.3724104705315564e-05,
|
| 1993 |
+
"loss": 1.1764,
|
| 1994 |
+
"step": 23100
|
| 1995 |
+
},
|
| 1996 |
+
{
|
| 1997 |
+
"epoch": 2.200094831673779,
|
| 1998 |
+
"grad_norm": 1.6442806720733643,
|
| 1999 |
+
"learning_rate": 1.3563513730528346e-05,
|
| 2000 |
+
"loss": 1.174,
|
| 2001 |
+
"step": 23200
|
| 2002 |
+
},
|
| 2003 |
+
{
|
| 2004 |
+
"epoch": 2.209577999051683,
|
| 2005 |
+
"grad_norm": 1.480901837348938,
|
| 2006 |
+
"learning_rate": 1.3402922755741129e-05,
|
| 2007 |
+
"loss": 1.1666,
|
| 2008 |
+
"step": 23300
|
| 2009 |
+
},
|
| 2010 |
+
{
|
| 2011 |
+
"epoch": 2.2190611664295874,
|
| 2012 |
+
"grad_norm": 1.6193006038665771,
|
| 2013 |
+
"learning_rate": 1.3242331780953911e-05,
|
| 2014 |
+
"loss": 1.1975,
|
| 2015 |
+
"step": 23400
|
| 2016 |
+
},
|
| 2017 |
+
{
|
| 2018 |
+
"epoch": 2.2285443338074917,
|
| 2019 |
+
"grad_norm": 1.2970917224884033,
|
| 2020 |
+
"learning_rate": 1.3081740806166696e-05,
|
| 2021 |
+
"loss": 1.1579,
|
| 2022 |
+
"step": 23500
|
| 2023 |
+
},
|
| 2024 |
+
{
|
| 2025 |
+
"epoch": 2.2285443338074917,
|
| 2026 |
+
"eval_loss": 1.1433159112930298,
|
| 2027 |
+
"eval_runtime": 72.0832,
|
| 2028 |
+
"eval_samples_per_second": 130.044,
|
| 2029 |
+
"eval_steps_per_second": 16.259,
|
| 2030 |
+
"step": 23500
|
| 2031 |
+
},
|
| 2032 |
+
{
|
| 2033 |
+
"epoch": 2.238027501185396,
|
| 2034 |
+
"grad_norm": 1.4054538011550903,
|
| 2035 |
+
"learning_rate": 1.2921149831379478e-05,
|
| 2036 |
+
"loss": 1.1779,
|
| 2037 |
+
"step": 23600
|
| 2038 |
+
},
|
| 2039 |
+
{
|
| 2040 |
+
"epoch": 2.2475106685633,
|
| 2041 |
+
"grad_norm": 1.5161010026931763,
|
| 2042 |
+
"learning_rate": 1.276055885659226e-05,
|
| 2043 |
+
"loss": 1.1709,
|
| 2044 |
+
"step": 23700
|
| 2045 |
+
},
|
| 2046 |
+
{
|
| 2047 |
+
"epoch": 2.2569938359412043,
|
| 2048 |
+
"grad_norm": 2.040818929672241,
|
| 2049 |
+
"learning_rate": 1.2599967881805045e-05,
|
| 2050 |
+
"loss": 1.1692,
|
| 2051 |
+
"step": 23800
|
| 2052 |
+
},
|
| 2053 |
+
{
|
| 2054 |
+
"epoch": 2.2664770033191086,
|
| 2055 |
+
"grad_norm": 1.3812401294708252,
|
| 2056 |
+
"learning_rate": 1.2439376907017826e-05,
|
| 2057 |
+
"loss": 1.1733,
|
| 2058 |
+
"step": 23900
|
| 2059 |
+
},
|
| 2060 |
+
{
|
| 2061 |
+
"epoch": 2.275960170697013,
|
| 2062 |
+
"grad_norm": 2.113886833190918,
|
| 2063 |
+
"learning_rate": 1.2278785932230608e-05,
|
| 2064 |
+
"loss": 1.1682,
|
| 2065 |
+
"step": 24000
|
| 2066 |
+
},
|
| 2067 |
+
{
|
| 2068 |
+
"epoch": 2.275960170697013,
|
| 2069 |
+
"eval_loss": 1.1404303312301636,
|
| 2070 |
+
"eval_runtime": 72.1649,
|
| 2071 |
+
"eval_samples_per_second": 129.897,
|
| 2072 |
+
"eval_steps_per_second": 16.241,
|
| 2073 |
+
"step": 24000
|
| 2074 |
+
},
|
| 2075 |
+
{
|
| 2076 |
+
"epoch": 2.285443338074917,
|
| 2077 |
+
"grad_norm": 1.3256770372390747,
|
| 2078 |
+
"learning_rate": 1.2118194957443393e-05,
|
| 2079 |
+
"loss": 1.1847,
|
| 2080 |
+
"step": 24100
|
| 2081 |
+
},
|
| 2082 |
+
{
|
| 2083 |
+
"epoch": 2.294926505452821,
|
| 2084 |
+
"grad_norm": 1.4699623584747314,
|
| 2085 |
+
"learning_rate": 1.1957603982656175e-05,
|
| 2086 |
+
"loss": 1.1576,
|
| 2087 |
+
"step": 24200
|
| 2088 |
+
},
|
| 2089 |
+
{
|
| 2090 |
+
"epoch": 2.3044096728307255,
|
| 2091 |
+
"grad_norm": 1.5492583513259888,
|
| 2092 |
+
"learning_rate": 1.1797013007868958e-05,
|
| 2093 |
+
"loss": 1.1532,
|
| 2094 |
+
"step": 24300
|
| 2095 |
+
},
|
| 2096 |
+
{
|
| 2097 |
+
"epoch": 2.31389284020863,
|
| 2098 |
+
"grad_norm": 1.409488558769226,
|
| 2099 |
+
"learning_rate": 1.1636422033081742e-05,
|
| 2100 |
+
"loss": 1.1626,
|
| 2101 |
+
"step": 24400
|
| 2102 |
+
},
|
| 2103 |
+
{
|
| 2104 |
+
"epoch": 2.3233760075865337,
|
| 2105 |
+
"grad_norm": 1.642247200012207,
|
| 2106 |
+
"learning_rate": 1.1475831058294524e-05,
|
| 2107 |
+
"loss": 1.1943,
|
| 2108 |
+
"step": 24500
|
| 2109 |
+
},
|
| 2110 |
+
{
|
| 2111 |
+
"epoch": 2.3233760075865337,
|
| 2112 |
+
"eval_loss": 1.139186978340149,
|
| 2113 |
+
"eval_runtime": 72.1131,
|
| 2114 |
+
"eval_samples_per_second": 129.99,
|
| 2115 |
+
"eval_steps_per_second": 16.252,
|
| 2116 |
+
"step": 24500
|
| 2117 |
+
},
|
| 2118 |
+
{
|
| 2119 |
+
"epoch": 2.332859174964438,
|
| 2120 |
+
"grad_norm": 1.4776501655578613,
|
| 2121 |
+
"learning_rate": 1.1315240083507307e-05,
|
| 2122 |
+
"loss": 1.1566,
|
| 2123 |
+
"step": 24600
|
| 2124 |
+
},
|
| 2125 |
+
{
|
| 2126 |
+
"epoch": 2.3423423423423424,
|
| 2127 |
+
"grad_norm": 1.475188136100769,
|
| 2128 |
+
"learning_rate": 1.115464910872009e-05,
|
| 2129 |
+
"loss": 1.1743,
|
| 2130 |
+
"step": 24700
|
| 2131 |
+
},
|
| 2132 |
+
{
|
| 2133 |
+
"epoch": 2.3518255097202467,
|
| 2134 |
+
"grad_norm": 1.48451828956604,
|
| 2135 |
+
"learning_rate": 1.0994058133932874e-05,
|
| 2136 |
+
"loss": 1.1539,
|
| 2137 |
+
"step": 24800
|
| 2138 |
+
},
|
| 2139 |
+
{
|
| 2140 |
+
"epoch": 2.3613086770981506,
|
| 2141 |
+
"grad_norm": 1.4650864601135254,
|
| 2142 |
+
"learning_rate": 1.0833467159145656e-05,
|
| 2143 |
+
"loss": 1.2073,
|
| 2144 |
+
"step": 24900
|
| 2145 |
+
},
|
| 2146 |
+
{
|
| 2147 |
+
"epoch": 2.370791844476055,
|
| 2148 |
+
"grad_norm": 1.71983003616333,
|
| 2149 |
+
"learning_rate": 1.0672876184358439e-05,
|
| 2150 |
+
"loss": 1.2021,
|
| 2151 |
+
"step": 25000
|
| 2152 |
+
},
|
| 2153 |
+
{
|
| 2154 |
+
"epoch": 2.370791844476055,
|
| 2155 |
+
"eval_loss": 1.1377766132354736,
|
| 2156 |
+
"eval_runtime": 71.9749,
|
| 2157 |
+
"eval_samples_per_second": 130.24,
|
| 2158 |
+
"eval_steps_per_second": 16.283,
|
| 2159 |
+
"step": 25000
|
| 2160 |
+
},
|
| 2161 |
+
{
|
| 2162 |
+
"epoch": 2.3802750118539593,
|
| 2163 |
+
"grad_norm": 1.3838121891021729,
|
| 2164 |
+
"learning_rate": 1.0512285209571223e-05,
|
| 2165 |
+
"loss": 1.1791,
|
| 2166 |
+
"step": 25100
|
| 2167 |
+
},
|
| 2168 |
+
{
|
| 2169 |
+
"epoch": 2.3897581792318636,
|
| 2170 |
+
"grad_norm": 1.8836325407028198,
|
| 2171 |
+
"learning_rate": 1.0351694234784006e-05,
|
| 2172 |
+
"loss": 1.1834,
|
| 2173 |
+
"step": 25200
|
| 2174 |
+
},
|
| 2175 |
+
{
|
| 2176 |
+
"epoch": 2.3992413466097675,
|
| 2177 |
+
"grad_norm": 1.3679293394088745,
|
| 2178 |
+
"learning_rate": 1.0191103259996788e-05,
|
| 2179 |
+
"loss": 1.183,
|
| 2180 |
+
"step": 25300
|
| 2181 |
+
},
|
| 2182 |
+
{
|
| 2183 |
+
"epoch": 2.408724513987672,
|
| 2184 |
+
"grad_norm": 1.5593743324279785,
|
| 2185 |
+
"learning_rate": 1.003051228520957e-05,
|
| 2186 |
+
"loss": 1.1703,
|
| 2187 |
+
"step": 25400
|
| 2188 |
+
},
|
| 2189 |
+
{
|
| 2190 |
+
"epoch": 2.418207681365576,
|
| 2191 |
+
"grad_norm": 1.4257512092590332,
|
| 2192 |
+
"learning_rate": 9.869921310422355e-06,
|
| 2193 |
+
"loss": 1.172,
|
| 2194 |
+
"step": 25500
|
| 2195 |
+
},
|
| 2196 |
+
{
|
| 2197 |
+
"epoch": 2.418207681365576,
|
| 2198 |
+
"eval_loss": 1.1378742456436157,
|
| 2199 |
+
"eval_runtime": 72.0363,
|
| 2200 |
+
"eval_samples_per_second": 130.129,
|
| 2201 |
+
"eval_steps_per_second": 16.27,
|
| 2202 |
+
"step": 25500
|
| 2203 |
+
},
|
| 2204 |
+
{
|
| 2205 |
+
"epoch": 2.4276908487434805,
|
| 2206 |
+
"grad_norm": 1.771941065788269,
|
| 2207 |
+
"learning_rate": 9.709330335635137e-06,
|
| 2208 |
+
"loss": 1.1676,
|
| 2209 |
+
"step": 25600
|
| 2210 |
+
},
|
| 2211 |
+
{
|
| 2212 |
+
"epoch": 2.4371740161213844,
|
| 2213 |
+
"grad_norm": 1.7247157096862793,
|
| 2214 |
+
"learning_rate": 9.54873936084792e-06,
|
| 2215 |
+
"loss": 1.1753,
|
| 2216 |
+
"step": 25700
|
| 2217 |
+
},
|
| 2218 |
+
{
|
| 2219 |
+
"epoch": 2.4466571834992887,
|
| 2220 |
+
"grad_norm": 1.5509614944458008,
|
| 2221 |
+
"learning_rate": 9.388148386060704e-06,
|
| 2222 |
+
"loss": 1.1705,
|
| 2223 |
+
"step": 25800
|
| 2224 |
+
},
|
| 2225 |
+
{
|
| 2226 |
+
"epoch": 2.456140350877193,
|
| 2227 |
+
"grad_norm": 1.8205307722091675,
|
| 2228 |
+
"learning_rate": 9.227557411273487e-06,
|
| 2229 |
+
"loss": 1.1938,
|
| 2230 |
+
"step": 25900
|
| 2231 |
+
},
|
| 2232 |
+
{
|
| 2233 |
+
"epoch": 2.4656235182550974,
|
| 2234 |
+
"grad_norm": 1.501631498336792,
|
| 2235 |
+
"learning_rate": 9.06696643648627e-06,
|
| 2236 |
+
"loss": 1.1737,
|
| 2237 |
+
"step": 26000
|
| 2238 |
+
},
|
| 2239 |
+
{
|
| 2240 |
+
"epoch": 2.4656235182550974,
|
| 2241 |
+
"eval_loss": 1.1362242698669434,
|
| 2242 |
+
"eval_runtime": 71.9608,
|
| 2243 |
+
"eval_samples_per_second": 130.265,
|
| 2244 |
+
"eval_steps_per_second": 16.287,
|
| 2245 |
+
"step": 26000
|
| 2246 |
+
},
|
| 2247 |
+
{
|
| 2248 |
+
"epoch": 2.4751066856330013,
|
| 2249 |
+
"grad_norm": 1.4233213663101196,
|
| 2250 |
+
"learning_rate": 8.906375461699054e-06,
|
| 2251 |
+
"loss": 1.1728,
|
| 2252 |
+
"step": 26100
|
| 2253 |
+
},
|
| 2254 |
+
{
|
| 2255 |
+
"epoch": 2.4845898530109056,
|
| 2256 |
+
"grad_norm": 1.597785472869873,
|
| 2257 |
+
"learning_rate": 8.745784486911836e-06,
|
| 2258 |
+
"loss": 1.1559,
|
| 2259 |
+
"step": 26200
|
| 2260 |
+
},
|
| 2261 |
+
{
|
| 2262 |
+
"epoch": 2.49407302038881,
|
| 2263 |
+
"grad_norm": 1.2396786212921143,
|
| 2264 |
+
"learning_rate": 8.585193512124619e-06,
|
| 2265 |
+
"loss": 1.1645,
|
| 2266 |
+
"step": 26300
|
| 2267 |
+
},
|
| 2268 |
+
{
|
| 2269 |
+
"epoch": 2.503556187766714,
|
| 2270 |
+
"grad_norm": 1.643211841583252,
|
| 2271 |
+
"learning_rate": 8.424602537337401e-06,
|
| 2272 |
+
"loss": 1.1948,
|
| 2273 |
+
"step": 26400
|
| 2274 |
+
},
|
| 2275 |
+
{
|
| 2276 |
+
"epoch": 2.513039355144618,
|
| 2277 |
+
"grad_norm": 1.688436508178711,
|
| 2278 |
+
"learning_rate": 8.264011562550185e-06,
|
| 2279 |
+
"loss": 1.1875,
|
| 2280 |
+
"step": 26500
|
| 2281 |
+
},
|
| 2282 |
+
{
|
| 2283 |
+
"epoch": 2.513039355144618,
|
| 2284 |
+
"eval_loss": 1.134669303894043,
|
| 2285 |
+
"eval_runtime": 72.1082,
|
| 2286 |
+
"eval_samples_per_second": 129.999,
|
| 2287 |
+
"eval_steps_per_second": 16.253,
|
| 2288 |
+
"step": 26500
|
| 2289 |
+
},
|
| 2290 |
+
{
|
| 2291 |
+
"epoch": 2.5225225225225225,
|
| 2292 |
+
"grad_norm": 1.6127384901046753,
|
| 2293 |
+
"learning_rate": 8.103420587762968e-06,
|
| 2294 |
+
"loss": 1.1657,
|
| 2295 |
+
"step": 26600
|
| 2296 |
+
},
|
| 2297 |
+
{
|
| 2298 |
+
"epoch": 2.532005689900427,
|
| 2299 |
+
"grad_norm": 2.12892484664917,
|
| 2300 |
+
"learning_rate": 7.944435522723622e-06,
|
| 2301 |
+
"loss": 1.1636,
|
| 2302 |
+
"step": 26700
|
| 2303 |
+
},
|
| 2304 |
+
{
|
| 2305 |
+
"epoch": 2.541488857278331,
|
| 2306 |
+
"grad_norm": 1.173686146736145,
|
| 2307 |
+
"learning_rate": 7.783844547936407e-06,
|
| 2308 |
+
"loss": 1.1866,
|
| 2309 |
+
"step": 26800
|
| 2310 |
+
},
|
| 2311 |
+
{
|
| 2312 |
+
"epoch": 2.550972024656235,
|
| 2313 |
+
"grad_norm": 1.4527802467346191,
|
| 2314 |
+
"learning_rate": 7.623253573149189e-06,
|
| 2315 |
+
"loss": 1.1755,
|
| 2316 |
+
"step": 26900
|
| 2317 |
+
},
|
| 2318 |
+
{
|
| 2319 |
+
"epoch": 2.5604551920341394,
|
| 2320 |
+
"grad_norm": 1.6228667497634888,
|
| 2321 |
+
"learning_rate": 7.462662598361972e-06,
|
| 2322 |
+
"loss": 1.1427,
|
| 2323 |
+
"step": 27000
|
| 2324 |
+
},
|
| 2325 |
+
{
|
| 2326 |
+
"epoch": 2.5604551920341394,
|
| 2327 |
+
"eval_loss": 1.134996771812439,
|
| 2328 |
+
"eval_runtime": 72.1853,
|
| 2329 |
+
"eval_samples_per_second": 129.86,
|
| 2330 |
+
"eval_steps_per_second": 16.236,
|
| 2331 |
+
"step": 27000
|
| 2332 |
+
},
|
| 2333 |
+
{
|
| 2334 |
+
"epoch": 2.5699383594120437,
|
| 2335 |
+
"grad_norm": 1.5179518461227417,
|
| 2336 |
+
"learning_rate": 7.302071623574755e-06,
|
| 2337 |
+
"loss": 1.1496,
|
| 2338 |
+
"step": 27100
|
| 2339 |
+
},
|
| 2340 |
+
{
|
| 2341 |
+
"epoch": 2.5794215267899476,
|
| 2342 |
+
"grad_norm": 1.2633978128433228,
|
| 2343 |
+
"learning_rate": 7.141480648787538e-06,
|
| 2344 |
+
"loss": 1.1633,
|
| 2345 |
+
"step": 27200
|
| 2346 |
+
},
|
| 2347 |
+
{
|
| 2348 |
+
"epoch": 2.588904694167852,
|
| 2349 |
+
"grad_norm": 1.3050264120101929,
|
| 2350 |
+
"learning_rate": 6.980889674000321e-06,
|
| 2351 |
+
"loss": 1.1614,
|
| 2352 |
+
"step": 27300
|
| 2353 |
+
},
|
| 2354 |
+
{
|
| 2355 |
+
"epoch": 2.5983878615457563,
|
| 2356 |
+
"grad_norm": 1.432268500328064,
|
| 2357 |
+
"learning_rate": 6.820298699213104e-06,
|
| 2358 |
+
"loss": 1.1684,
|
| 2359 |
+
"step": 27400
|
| 2360 |
+
},
|
| 2361 |
+
{
|
| 2362 |
+
"epoch": 2.6078710289236606,
|
| 2363 |
+
"grad_norm": 1.6904171705245972,
|
| 2364 |
+
"learning_rate": 6.659707724425887e-06,
|
| 2365 |
+
"loss": 1.1673,
|
| 2366 |
+
"step": 27500
|
| 2367 |
+
},
|
| 2368 |
+
{
|
| 2369 |
+
"epoch": 2.6078710289236606,
|
| 2370 |
+
"eval_loss": 1.1333271265029907,
|
| 2371 |
+
"eval_runtime": 72.187,
|
| 2372 |
+
"eval_samples_per_second": 129.857,
|
| 2373 |
+
"eval_steps_per_second": 16.236,
|
| 2374 |
+
"step": 27500
|
| 2375 |
+
},
|
| 2376 |
+
{
|
| 2377 |
+
"epoch": 2.617354196301565,
|
| 2378 |
+
"grad_norm": 1.2229042053222656,
|
| 2379 |
+
"learning_rate": 6.49911674963867e-06,
|
| 2380 |
+
"loss": 1.1793,
|
| 2381 |
+
"step": 27600
|
| 2382 |
+
},
|
| 2383 |
+
{
|
| 2384 |
+
"epoch": 2.626837363679469,
|
| 2385 |
+
"grad_norm": 1.7409764528274536,
|
| 2386 |
+
"learning_rate": 6.338525774851453e-06,
|
| 2387 |
+
"loss": 1.1963,
|
| 2388 |
+
"step": 27700
|
| 2389 |
+
},
|
| 2390 |
+
{
|
| 2391 |
+
"epoch": 2.636320531057373,
|
| 2392 |
+
"grad_norm": 1.4706058502197266,
|
| 2393 |
+
"learning_rate": 6.177934800064237e-06,
|
| 2394 |
+
"loss": 1.1836,
|
| 2395 |
+
"step": 27800
|
| 2396 |
+
},
|
| 2397 |
+
{
|
| 2398 |
+
"epoch": 2.6458036984352775,
|
| 2399 |
+
"grad_norm": 1.3871138095855713,
|
| 2400 |
+
"learning_rate": 6.01734382527702e-06,
|
| 2401 |
+
"loss": 1.1669,
|
| 2402 |
+
"step": 27900
|
| 2403 |
+
},
|
| 2404 |
+
{
|
| 2405 |
+
"epoch": 2.6552868658131814,
|
| 2406 |
+
"grad_norm": 1.5841022729873657,
|
| 2407 |
+
"learning_rate": 5.856752850489803e-06,
|
| 2408 |
+
"loss": 1.1765,
|
| 2409 |
+
"step": 28000
|
| 2410 |
+
},
|
| 2411 |
+
{
|
| 2412 |
+
"epoch": 2.6552868658131814,
|
| 2413 |
+
"eval_loss": 1.1325418949127197,
|
| 2414 |
+
"eval_runtime": 72.1699,
|
| 2415 |
+
"eval_samples_per_second": 129.888,
|
| 2416 |
+
"eval_steps_per_second": 16.239,
|
| 2417 |
+
"step": 28000
|
| 2418 |
+
},
|
| 2419 |
+
{
|
| 2420 |
+
"epoch": 2.6647700331910857,
|
| 2421 |
+
"grad_norm": 1.2488940954208374,
|
| 2422 |
+
"learning_rate": 5.696161875702586e-06,
|
| 2423 |
+
"loss": 1.1581,
|
| 2424 |
+
"step": 28100
|
| 2425 |
+
},
|
| 2426 |
+
{
|
| 2427 |
+
"epoch": 2.67425320056899,
|
| 2428 |
+
"grad_norm": 1.633123517036438,
|
| 2429 |
+
"learning_rate": 5.535570900915369e-06,
|
| 2430 |
+
"loss": 1.1829,
|
| 2431 |
+
"step": 28200
|
| 2432 |
+
},
|
| 2433 |
+
{
|
| 2434 |
+
"epoch": 2.6837363679468944,
|
| 2435 |
+
"grad_norm": 1.558030366897583,
|
| 2436 |
+
"learning_rate": 5.374979926128152e-06,
|
| 2437 |
+
"loss": 1.1816,
|
| 2438 |
+
"step": 28300
|
| 2439 |
+
},
|
| 2440 |
+
{
|
| 2441 |
+
"epoch": 2.6932195353247987,
|
| 2442 |
+
"grad_norm": 1.5178041458129883,
|
| 2443 |
+
"learning_rate": 5.214388951340935e-06,
|
| 2444 |
+
"loss": 1.1789,
|
| 2445 |
+
"step": 28400
|
| 2446 |
+
},
|
| 2447 |
+
{
|
| 2448 |
+
"epoch": 2.7027027027027026,
|
| 2449 |
+
"grad_norm": 1.8317012786865234,
|
| 2450 |
+
"learning_rate": 5.053797976553718e-06,
|
| 2451 |
+
"loss": 1.1612,
|
| 2452 |
+
"step": 28500
|
| 2453 |
+
},
|
| 2454 |
+
{
|
| 2455 |
+
"epoch": 2.7027027027027026,
|
| 2456 |
+
"eval_loss": 1.1320453882217407,
|
| 2457 |
+
"eval_runtime": 72.3445,
|
| 2458 |
+
"eval_samples_per_second": 129.575,
|
| 2459 |
+
"eval_steps_per_second": 16.2,
|
| 2460 |
+
"step": 28500
|
| 2461 |
+
},
|
| 2462 |
+
{
|
| 2463 |
+
"epoch": 2.712185870080607,
|
| 2464 |
+
"grad_norm": 1.4248275756835938,
|
| 2465 |
+
"learning_rate": 4.893207001766502e-06,
|
| 2466 |
+
"loss": 1.1583,
|
| 2467 |
+
"step": 28600
|
| 2468 |
+
},
|
| 2469 |
+
{
|
| 2470 |
+
"epoch": 2.7216690374585113,
|
| 2471 |
+
"grad_norm": 1.3696835041046143,
|
| 2472 |
+
"learning_rate": 4.732616026979284e-06,
|
| 2473 |
+
"loss": 1.1302,
|
| 2474 |
+
"step": 28700
|
| 2475 |
+
},
|
| 2476 |
+
{
|
| 2477 |
+
"epoch": 2.731152204836415,
|
| 2478 |
+
"grad_norm": 1.4212887287139893,
|
| 2479 |
+
"learning_rate": 4.5720250521920675e-06,
|
| 2480 |
+
"loss": 1.1396,
|
| 2481 |
+
"step": 28800
|
| 2482 |
+
},
|
| 2483 |
+
{
|
| 2484 |
+
"epoch": 2.7406353722143195,
|
| 2485 |
+
"grad_norm": 1.6230417490005493,
|
| 2486 |
+
"learning_rate": 4.41143407740485e-06,
|
| 2487 |
+
"loss": 1.167,
|
| 2488 |
+
"step": 28900
|
| 2489 |
+
},
|
| 2490 |
+
{
|
| 2491 |
+
"epoch": 2.750118539592224,
|
| 2492 |
+
"grad_norm": 1.4556254148483276,
|
| 2493 |
+
"learning_rate": 4.252449012365505e-06,
|
| 2494 |
+
"loss": 1.2229,
|
| 2495 |
+
"step": 29000
|
| 2496 |
+
},
|
| 2497 |
+
{
|
| 2498 |
+
"epoch": 2.750118539592224,
|
| 2499 |
+
"eval_loss": 1.1307094097137451,
|
| 2500 |
+
"eval_runtime": 72.2019,
|
| 2501 |
+
"eval_samples_per_second": 129.83,
|
| 2502 |
+
"eval_steps_per_second": 16.232,
|
| 2503 |
+
"step": 29000
|
| 2504 |
+
},
|
| 2505 |
+
{
|
| 2506 |
+
"epoch": 2.759601706970128,
|
| 2507 |
+
"grad_norm": 1.399604082107544,
|
| 2508 |
+
"learning_rate": 4.091858037578288e-06,
|
| 2509 |
+
"loss": 1.183,
|
| 2510 |
+
"step": 29100
|
| 2511 |
+
},
|
| 2512 |
+
{
|
| 2513 |
+
"epoch": 2.769084874348032,
|
| 2514 |
+
"grad_norm": 1.3562369346618652,
|
| 2515 |
+
"learning_rate": 3.931267062791071e-06,
|
| 2516 |
+
"loss": 1.1729,
|
| 2517 |
+
"step": 29200
|
| 2518 |
+
},
|
| 2519 |
+
{
|
| 2520 |
+
"epoch": 2.7785680417259364,
|
| 2521 |
+
"grad_norm": 1.4427545070648193,
|
| 2522 |
+
"learning_rate": 3.7706760880038542e-06,
|
| 2523 |
+
"loss": 1.1636,
|
| 2524 |
+
"step": 29300
|
| 2525 |
+
},
|
| 2526 |
+
{
|
| 2527 |
+
"epoch": 2.7880512091038407,
|
| 2528 |
+
"grad_norm": 1.6153539419174194,
|
| 2529 |
+
"learning_rate": 3.610085113216637e-06,
|
| 2530 |
+
"loss": 1.1608,
|
| 2531 |
+
"step": 29400
|
| 2532 |
+
},
|
| 2533 |
+
{
|
| 2534 |
+
"epoch": 2.797534376481745,
|
| 2535 |
+
"grad_norm": 1.553841233253479,
|
| 2536 |
+
"learning_rate": 3.44949413842942e-06,
|
| 2537 |
+
"loss": 1.1727,
|
| 2538 |
+
"step": 29500
|
| 2539 |
+
},
|
| 2540 |
+
{
|
| 2541 |
+
"epoch": 2.797534376481745,
|
| 2542 |
+
"eval_loss": 1.1305798292160034,
|
| 2543 |
+
"eval_runtime": 72.4369,
|
| 2544 |
+
"eval_samples_per_second": 129.409,
|
| 2545 |
+
"eval_steps_per_second": 16.18,
|
| 2546 |
+
"step": 29500
|
| 2547 |
+
},
|
| 2548 |
+
{
|
| 2549 |
+
"epoch": 2.807017543859649,
|
| 2550 |
+
"grad_norm": 1.4503796100616455,
|
| 2551 |
+
"learning_rate": 3.2889031636422036e-06,
|
| 2552 |
+
"loss": 1.1533,
|
| 2553 |
+
"step": 29600
|
| 2554 |
+
},
|
| 2555 |
+
{
|
| 2556 |
+
"epoch": 2.8165007112375533,
|
| 2557 |
+
"grad_norm": 2.3234095573425293,
|
| 2558 |
+
"learning_rate": 3.1283121888549865e-06,
|
| 2559 |
+
"loss": 1.1849,
|
| 2560 |
+
"step": 29700
|
| 2561 |
+
},
|
| 2562 |
+
{
|
| 2563 |
+
"epoch": 2.8259838786154576,
|
| 2564 |
+
"grad_norm": 1.6692347526550293,
|
| 2565 |
+
"learning_rate": 2.9677212140677695e-06,
|
| 2566 |
+
"loss": 1.1629,
|
| 2567 |
+
"step": 29800
|
| 2568 |
+
},
|
| 2569 |
+
{
|
| 2570 |
+
"epoch": 2.8354670459933615,
|
| 2571 |
+
"grad_norm": 1.6683822870254517,
|
| 2572 |
+
"learning_rate": 2.8071302392805524e-06,
|
| 2573 |
+
"loss": 1.1584,
|
| 2574 |
+
"step": 29900
|
| 2575 |
+
},
|
| 2576 |
+
{
|
| 2577 |
+
"epoch": 2.844950213371266,
|
| 2578 |
+
"grad_norm": 1.371102213859558,
|
| 2579 |
+
"learning_rate": 2.6465392644933354e-06,
|
| 2580 |
+
"loss": 1.1208,
|
| 2581 |
+
"step": 30000
|
| 2582 |
+
},
|
| 2583 |
+
{
|
| 2584 |
+
"epoch": 2.844950213371266,
|
| 2585 |
+
"eval_loss": 1.1299171447753906,
|
| 2586 |
+
"eval_runtime": 72.1885,
|
| 2587 |
+
"eval_samples_per_second": 129.854,
|
| 2588 |
+
"eval_steps_per_second": 16.235,
|
| 2589 |
+
"step": 30000
|
| 2590 |
+
},
|
| 2591 |
+
{
|
| 2592 |
+
"epoch": 2.85443338074917,
|
| 2593 |
+
"grad_norm": 1.9285227060317993,
|
| 2594 |
+
"learning_rate": 2.4859482897061184e-06,
|
| 2595 |
+
"loss": 1.1871,
|
| 2596 |
+
"step": 30100
|
| 2597 |
+
},
|
| 2598 |
+
{
|
| 2599 |
+
"epoch": 2.8639165481270745,
|
| 2600 |
+
"grad_norm": 1.5394768714904785,
|
| 2601 |
+
"learning_rate": 2.3253573149189017e-06,
|
| 2602 |
+
"loss": 1.1786,
|
| 2603 |
+
"step": 30200
|
| 2604 |
+
},
|
| 2605 |
+
{
|
| 2606 |
+
"epoch": 2.873399715504979,
|
| 2607 |
+
"grad_norm": 1.606779932975769,
|
| 2608 |
+
"learning_rate": 2.1647663401316847e-06,
|
| 2609 |
+
"loss": 1.181,
|
| 2610 |
+
"step": 30300
|
| 2611 |
+
},
|
| 2612 |
+
{
|
| 2613 |
+
"epoch": 2.8828828828828827,
|
| 2614 |
+
"grad_norm": 1.6637898683547974,
|
| 2615 |
+
"learning_rate": 2.0041753653444677e-06,
|
| 2616 |
+
"loss": 1.1435,
|
| 2617 |
+
"step": 30400
|
| 2618 |
+
},
|
| 2619 |
+
{
|
| 2620 |
+
"epoch": 2.892366050260787,
|
| 2621 |
+
"grad_norm": 1.4190491437911987,
|
| 2622 |
+
"learning_rate": 1.8435843905572506e-06,
|
| 2623 |
+
"loss": 1.158,
|
| 2624 |
+
"step": 30500
|
| 2625 |
+
},
|
| 2626 |
+
{
|
| 2627 |
+
"epoch": 2.892366050260787,
|
| 2628 |
+
"eval_loss": 1.129961371421814,
|
| 2629 |
+
"eval_runtime": 72.2984,
|
| 2630 |
+
"eval_samples_per_second": 129.657,
|
| 2631 |
+
"eval_steps_per_second": 16.211,
|
| 2632 |
+
"step": 30500
|
| 2633 |
+
},
|
| 2634 |
+
{
|
| 2635 |
+
"epoch": 2.9018492176386914,
|
| 2636 |
+
"grad_norm": 1.3839406967163086,
|
| 2637 |
+
"learning_rate": 1.6829934157700338e-06,
|
| 2638 |
+
"loss": 1.1716,
|
| 2639 |
+
"step": 30600
|
| 2640 |
+
},
|
| 2641 |
+
{
|
| 2642 |
+
"epoch": 2.9113323850165953,
|
| 2643 |
+
"grad_norm": 1.2562811374664307,
|
| 2644 |
+
"learning_rate": 1.5224024409828168e-06,
|
| 2645 |
+
"loss": 1.1466,
|
| 2646 |
+
"step": 30700
|
| 2647 |
+
},
|
| 2648 |
+
{
|
| 2649 |
+
"epoch": 2.9208155523944996,
|
| 2650 |
+
"grad_norm": 1.4180203676223755,
|
| 2651 |
+
"learning_rate": 1.3618114661955997e-06,
|
| 2652 |
+
"loss": 1.1405,
|
| 2653 |
+
"step": 30800
|
| 2654 |
+
},
|
| 2655 |
+
{
|
| 2656 |
+
"epoch": 2.930298719772404,
|
| 2657 |
+
"grad_norm": 1.7891360521316528,
|
| 2658 |
+
"learning_rate": 1.2012204914083829e-06,
|
| 2659 |
+
"loss": 1.1591,
|
| 2660 |
+
"step": 30900
|
| 2661 |
+
},
|
| 2662 |
+
{
|
| 2663 |
+
"epoch": 2.9397818871503083,
|
| 2664 |
+
"grad_norm": 1.7551426887512207,
|
| 2665 |
+
"learning_rate": 1.0406295166211659e-06,
|
| 2666 |
+
"loss": 1.1833,
|
| 2667 |
+
"step": 31000
|
| 2668 |
+
},
|
| 2669 |
+
{
|
| 2670 |
+
"epoch": 2.9397818871503083,
|
| 2671 |
+
"eval_loss": 1.129394292831421,
|
| 2672 |
+
"eval_runtime": 72.2972,
|
| 2673 |
+
"eval_samples_per_second": 129.659,
|
| 2674 |
+
"eval_steps_per_second": 16.211,
|
| 2675 |
+
"step": 31000
|
| 2676 |
+
},
|
| 2677 |
+
{
|
| 2678 |
+
"epoch": 2.9492650545282126,
|
| 2679 |
+
"grad_norm": 1.4321238994598389,
|
| 2680 |
+
"learning_rate": 8.800385418339489e-07,
|
| 2681 |
+
"loss": 1.1879,
|
| 2682 |
+
"step": 31100
|
| 2683 |
+
},
|
| 2684 |
+
{
|
| 2685 |
+
"epoch": 2.9587482219061165,
|
| 2686 |
+
"grad_norm": 1.732853651046753,
|
| 2687 |
+
"learning_rate": 7.210534767946041e-07,
|
| 2688 |
+
"loss": 1.1682,
|
| 2689 |
+
"step": 31200
|
| 2690 |
+
},
|
| 2691 |
+
{
|
| 2692 |
+
"epoch": 2.968231389284021,
|
| 2693 |
+
"grad_norm": 1.473656415939331,
|
| 2694 |
+
"learning_rate": 5.604625020073872e-07,
|
| 2695 |
+
"loss": 1.1708,
|
| 2696 |
+
"step": 31300
|
| 2697 |
+
},
|
| 2698 |
+
{
|
| 2699 |
+
"epoch": 2.977714556661925,
|
| 2700 |
+
"grad_norm": 1.2021667957305908,
|
| 2701 |
+
"learning_rate": 3.998715272201702e-07,
|
| 2702 |
+
"loss": 1.1679,
|
| 2703 |
+
"step": 31400
|
| 2704 |
+
},
|
| 2705 |
+
{
|
| 2706 |
+
"epoch": 2.987197724039829,
|
| 2707 |
+
"grad_norm": 1.4972681999206543,
|
| 2708 |
+
"learning_rate": 2.4088646218082545e-07,
|
| 2709 |
+
"loss": 1.1678,
|
| 2710 |
+
"step": 31500
|
| 2711 |
+
},
|
| 2712 |
+
{
|
| 2713 |
+
"epoch": 2.987197724039829,
|
| 2714 |
+
"eval_loss": 1.129324197769165,
|
| 2715 |
+
"eval_runtime": 72.2219,
|
| 2716 |
+
"eval_samples_per_second": 129.794,
|
| 2717 |
+
"eval_steps_per_second": 16.228,
|
| 2718 |
+
"step": 31500
|
| 2719 |
+
},
|
| 2720 |
+
{
|
| 2721 |
+
"epoch": 2.9966808914177334,
|
| 2722 |
+
"grad_norm": 1.7410774230957031,
|
| 2723 |
+
"learning_rate": 8.029548739360848e-08,
|
| 2724 |
+
"loss": 1.1645,
|
| 2725 |
+
"step": 31600
|
| 2726 |
+
}
|
| 2727 |
+
],
|
| 2728 |
+
"logging_steps": 100,
|
| 2729 |
+
"max_steps": 31635,
|
| 2730 |
+
"num_input_tokens_seen": 0,
|
| 2731 |
+
"num_train_epochs": 3,
|
| 2732 |
+
"save_steps": 500,
|
| 2733 |
+
"stateful_callbacks": {
|
| 2734 |
+
"TrainerControl": {
|
| 2735 |
+
"args": {
|
| 2736 |
+
"should_epoch_stop": false,
|
| 2737 |
+
"should_evaluate": false,
|
| 2738 |
+
"should_log": false,
|
| 2739 |
+
"should_save": true,
|
| 2740 |
+
"should_training_stop": true
|
| 2741 |
+
},
|
| 2742 |
+
"attributes": {}
|
| 2743 |
+
}
|
| 2744 |
+
},
|
| 2745 |
+
"total_flos": 1.32254007164928e+17,
|
| 2746 |
+
"train_batch_size": 8,
|
| 2747 |
+
"trial_name": null,
|
| 2748 |
+
"trial_params": null
|
| 2749 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d41c01d752ebacb9caebf09f81c9bf077c60c828f6470cae02f424d092e9b86c
|
| 3 |
+
size 5368
|