Upload 13 files
Browse files- config.json +39 -0
- generation_config.json +6 -0
- merges.txt +0 -0
- model.safetensors +3 -0
- optimizer.pt +3 -0
- rng_state.pth +3 -0
- scheduler.pt +3 -0
- special_tokens_map.json +5 -0
- tokenizer.json +0 -0
- tokenizer_config.json +19 -0
- trainer_state.json +1421 -0
- training_args.bin +3 -0
- vocab.json +0 -0
config.json
ADDED
@@ -0,0 +1,39 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_name_or_path": "openai-community/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": "4.40.1",
|
37 |
+
"use_cache": true,
|
38 |
+
"vocab_size": 50257
|
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": "4.40.1"
|
6 |
+
}
|
merges.txt
ADDED
The diff for this file is too large to render.
See raw diff
|
|
model.safetensors
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:f4cc3830150f1b96e749b05aa4896a89c073f25562750ec50d33590ad9858c03
|
3 |
+
size 497774208
|
optimizer.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:dcad6df9b1bc1c01e062177b716a8d4514fb35f9dbfa9c8d9343ff984a5c6635
|
3 |
+
size 995642298
|
rng_state.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:40c45e9f4c00aea5599fd7092cf44a64857d6b6f915890545cb7f11f53e7b0f0
|
3 |
+
size 14244
|
scheduler.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:726fc45979fcbc0b528e0eee04eba68f1b23676fca33f522bab3252706de38d2
|
3 |
+
size 1064
|
special_tokens_map.json
ADDED
@@ -0,0 +1,5 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"bos_token": "<|endoftext|>",
|
3 |
+
"eos_token": "<|endoftext|>",
|
4 |
+
"unk_token": "<|endoftext|>"
|
5 |
+
}
|
tokenizer.json
ADDED
The diff for this file is too large to render.
See raw diff
|
|
tokenizer_config.json
ADDED
@@ -0,0 +1,19 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"add_prefix_space": false,
|
3 |
+
"added_tokens_decoder": {
|
4 |
+
"50256": {
|
5 |
+
"content": "<|endoftext|>",
|
6 |
+
"lstrip": false,
|
7 |
+
"normalized": true,
|
8 |
+
"rstrip": false,
|
9 |
+
"single_word": false,
|
10 |
+
"special": true
|
11 |
+
}
|
12 |
+
},
|
13 |
+
"bos_token": "<|endoftext|>",
|
14 |
+
"clean_up_tokenization_spaces": true,
|
15 |
+
"eos_token": "<|endoftext|>",
|
16 |
+
"model_max_length": 1024,
|
17 |
+
"tokenizer_class": "GPT2Tokenizer",
|
18 |
+
"unk_token": "<|endoftext|>"
|
19 |
+
}
|
trainer_state.json
ADDED
@@ -0,0 +1,1421 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"best_metric": null,
|
3 |
+
"best_model_checkpoint": null,
|
4 |
+
"epoch": 11.889192723814054,
|
5 |
+
"eval_steps": 500,
|
6 |
+
"global_step": 100000,
|
7 |
+
"is_hyper_param_search": false,
|
8 |
+
"is_local_process_zero": true,
|
9 |
+
"is_world_process_zero": true,
|
10 |
+
"log_history": [
|
11 |
+
{
|
12 |
+
"epoch": 0.059445963619070265,
|
13 |
+
"grad_norm": 2.0723960399627686,
|
14 |
+
"learning_rate": 4.985138509095233e-05,
|
15 |
+
"loss": 3.6788,
|
16 |
+
"step": 500
|
17 |
+
},
|
18 |
+
{
|
19 |
+
"epoch": 0.11889192723814053,
|
20 |
+
"grad_norm": 1.9278995990753174,
|
21 |
+
"learning_rate": 4.970277018190465e-05,
|
22 |
+
"loss": 3.4742,
|
23 |
+
"step": 1000
|
24 |
+
},
|
25 |
+
{
|
26 |
+
"epoch": 0.1783378908572108,
|
27 |
+
"grad_norm": 1.4848977327346802,
|
28 |
+
"learning_rate": 4.955415527285698e-05,
|
29 |
+
"loss": 3.3942,
|
30 |
+
"step": 1500
|
31 |
+
},
|
32 |
+
{
|
33 |
+
"epoch": 0.23778385447628106,
|
34 |
+
"grad_norm": 1.3492341041564941,
|
35 |
+
"learning_rate": 4.94055403638093e-05,
|
36 |
+
"loss": 3.3358,
|
37 |
+
"step": 2000
|
38 |
+
},
|
39 |
+
{
|
40 |
+
"epoch": 0.2972298180953513,
|
41 |
+
"grad_norm": 1.212128758430481,
|
42 |
+
"learning_rate": 4.925692545476163e-05,
|
43 |
+
"loss": 3.2851,
|
44 |
+
"step": 2500
|
45 |
+
},
|
46 |
+
{
|
47 |
+
"epoch": 0.3566757817144216,
|
48 |
+
"grad_norm": 1.1597293615341187,
|
49 |
+
"learning_rate": 4.9108310545713945e-05,
|
50 |
+
"loss": 3.2331,
|
51 |
+
"step": 3000
|
52 |
+
},
|
53 |
+
{
|
54 |
+
"epoch": 0.41612174533349183,
|
55 |
+
"grad_norm": 0.9653922319412231,
|
56 |
+
"learning_rate": 4.8959695636666275e-05,
|
57 |
+
"loss": 3.2339,
|
58 |
+
"step": 3500
|
59 |
+
},
|
60 |
+
{
|
61 |
+
"epoch": 0.4755677089525621,
|
62 |
+
"grad_norm": 1.0085793733596802,
|
63 |
+
"learning_rate": 4.88110807276186e-05,
|
64 |
+
"loss": 3.1856,
|
65 |
+
"step": 4000
|
66 |
+
},
|
67 |
+
{
|
68 |
+
"epoch": 0.5350136725716323,
|
69 |
+
"grad_norm": 1.0556505918502808,
|
70 |
+
"learning_rate": 4.866246581857092e-05,
|
71 |
+
"loss": 3.1748,
|
72 |
+
"step": 4500
|
73 |
+
},
|
74 |
+
{
|
75 |
+
"epoch": 0.5944596361907026,
|
76 |
+
"grad_norm": 0.9526228904724121,
|
77 |
+
"learning_rate": 4.851385090952324e-05,
|
78 |
+
"loss": 3.1529,
|
79 |
+
"step": 5000
|
80 |
+
},
|
81 |
+
{
|
82 |
+
"epoch": 0.6539055998097729,
|
83 |
+
"grad_norm": 0.984980046749115,
|
84 |
+
"learning_rate": 4.836523600047557e-05,
|
85 |
+
"loss": 3.1378,
|
86 |
+
"step": 5500
|
87 |
+
},
|
88 |
+
{
|
89 |
+
"epoch": 0.7133515634288432,
|
90 |
+
"grad_norm": 1.0135027170181274,
|
91 |
+
"learning_rate": 4.8216621091427895e-05,
|
92 |
+
"loss": 3.0848,
|
93 |
+
"step": 6000
|
94 |
+
},
|
95 |
+
{
|
96 |
+
"epoch": 0.7727975270479135,
|
97 |
+
"grad_norm": 0.9454924464225769,
|
98 |
+
"learning_rate": 4.806800618238022e-05,
|
99 |
+
"loss": 3.0916,
|
100 |
+
"step": 6500
|
101 |
+
},
|
102 |
+
{
|
103 |
+
"epoch": 0.8322434906669837,
|
104 |
+
"grad_norm": 0.9793129563331604,
|
105 |
+
"learning_rate": 4.791939127333254e-05,
|
106 |
+
"loss": 3.0642,
|
107 |
+
"step": 7000
|
108 |
+
},
|
109 |
+
{
|
110 |
+
"epoch": 0.891689454286054,
|
111 |
+
"grad_norm": 0.9016062617301941,
|
112 |
+
"learning_rate": 4.777077636428487e-05,
|
113 |
+
"loss": 3.0657,
|
114 |
+
"step": 7500
|
115 |
+
},
|
116 |
+
{
|
117 |
+
"epoch": 0.9511354179051242,
|
118 |
+
"grad_norm": 0.8690605163574219,
|
119 |
+
"learning_rate": 4.762216145523719e-05,
|
120 |
+
"loss": 3.0281,
|
121 |
+
"step": 8000
|
122 |
+
},
|
123 |
+
{
|
124 |
+
"epoch": 1.0105813815241944,
|
125 |
+
"grad_norm": 0.891808271408081,
|
126 |
+
"learning_rate": 4.7473546546189516e-05,
|
127 |
+
"loss": 3.0155,
|
128 |
+
"step": 8500
|
129 |
+
},
|
130 |
+
{
|
131 |
+
"epoch": 1.0700273451432647,
|
132 |
+
"grad_norm": 0.9521974325180054,
|
133 |
+
"learning_rate": 4.732493163714184e-05,
|
134 |
+
"loss": 2.9713,
|
135 |
+
"step": 9000
|
136 |
+
},
|
137 |
+
{
|
138 |
+
"epoch": 1.129473308762335,
|
139 |
+
"grad_norm": 0.9132643938064575,
|
140 |
+
"learning_rate": 4.717631672809417e-05,
|
141 |
+
"loss": 2.9663,
|
142 |
+
"step": 9500
|
143 |
+
},
|
144 |
+
{
|
145 |
+
"epoch": 1.1889192723814053,
|
146 |
+
"grad_norm": 0.909182608127594,
|
147 |
+
"learning_rate": 4.702770181904649e-05,
|
148 |
+
"loss": 2.9616,
|
149 |
+
"step": 10000
|
150 |
+
},
|
151 |
+
{
|
152 |
+
"epoch": 1.2483652360004756,
|
153 |
+
"grad_norm": 0.912726104259491,
|
154 |
+
"learning_rate": 4.687908690999881e-05,
|
155 |
+
"loss": 2.9653,
|
156 |
+
"step": 10500
|
157 |
+
},
|
158 |
+
{
|
159 |
+
"epoch": 1.3078111996195458,
|
160 |
+
"grad_norm": 0.8568936586380005,
|
161 |
+
"learning_rate": 4.6730472000951136e-05,
|
162 |
+
"loss": 2.9486,
|
163 |
+
"step": 11000
|
164 |
+
},
|
165 |
+
{
|
166 |
+
"epoch": 1.3672571632386161,
|
167 |
+
"grad_norm": 0.9120291471481323,
|
168 |
+
"learning_rate": 4.6581857091903465e-05,
|
169 |
+
"loss": 2.932,
|
170 |
+
"step": 11500
|
171 |
+
},
|
172 |
+
{
|
173 |
+
"epoch": 1.4267031268576864,
|
174 |
+
"grad_norm": 0.981961190700531,
|
175 |
+
"learning_rate": 4.643324218285579e-05,
|
176 |
+
"loss": 2.9345,
|
177 |
+
"step": 12000
|
178 |
+
},
|
179 |
+
{
|
180 |
+
"epoch": 1.4861490904767567,
|
181 |
+
"grad_norm": 0.9763424396514893,
|
182 |
+
"learning_rate": 4.628462727380811e-05,
|
183 |
+
"loss": 2.9193,
|
184 |
+
"step": 12500
|
185 |
+
},
|
186 |
+
{
|
187 |
+
"epoch": 1.545595054095827,
|
188 |
+
"grad_norm": 0.8868328332901001,
|
189 |
+
"learning_rate": 4.6136012364760434e-05,
|
190 |
+
"loss": 2.9164,
|
191 |
+
"step": 13000
|
192 |
+
},
|
193 |
+
{
|
194 |
+
"epoch": 1.605041017714897,
|
195 |
+
"grad_norm": 0.9175488352775574,
|
196 |
+
"learning_rate": 4.598739745571276e-05,
|
197 |
+
"loss": 2.8932,
|
198 |
+
"step": 13500
|
199 |
+
},
|
200 |
+
{
|
201 |
+
"epoch": 1.6644869813339676,
|
202 |
+
"grad_norm": 0.890186607837677,
|
203 |
+
"learning_rate": 4.583878254666508e-05,
|
204 |
+
"loss": 2.8933,
|
205 |
+
"step": 14000
|
206 |
+
},
|
207 |
+
{
|
208 |
+
"epoch": 1.7239329449530376,
|
209 |
+
"grad_norm": 0.9198343753814697,
|
210 |
+
"learning_rate": 4.569016763761741e-05,
|
211 |
+
"loss": 2.881,
|
212 |
+
"step": 14500
|
213 |
+
},
|
214 |
+
{
|
215 |
+
"epoch": 1.783378908572108,
|
216 |
+
"grad_norm": 0.9706104397773743,
|
217 |
+
"learning_rate": 4.554155272856973e-05,
|
218 |
+
"loss": 2.8705,
|
219 |
+
"step": 15000
|
220 |
+
},
|
221 |
+
{
|
222 |
+
"epoch": 1.8428248721911782,
|
223 |
+
"grad_norm": 0.9355807304382324,
|
224 |
+
"learning_rate": 4.539293781952206e-05,
|
225 |
+
"loss": 2.8601,
|
226 |
+
"step": 15500
|
227 |
+
},
|
228 |
+
{
|
229 |
+
"epoch": 1.9022708358102485,
|
230 |
+
"grad_norm": 0.8972137570381165,
|
231 |
+
"learning_rate": 4.524432291047438e-05,
|
232 |
+
"loss": 2.8632,
|
233 |
+
"step": 16000
|
234 |
+
},
|
235 |
+
{
|
236 |
+
"epoch": 1.9617167994293188,
|
237 |
+
"grad_norm": 0.8553013801574707,
|
238 |
+
"learning_rate": 4.5095708001426706e-05,
|
239 |
+
"loss": 2.8696,
|
240 |
+
"step": 16500
|
241 |
+
},
|
242 |
+
{
|
243 |
+
"epoch": 2.021162763048389,
|
244 |
+
"grad_norm": 0.8952363133430481,
|
245 |
+
"learning_rate": 4.494709309237903e-05,
|
246 |
+
"loss": 2.8541,
|
247 |
+
"step": 17000
|
248 |
+
},
|
249 |
+
{
|
250 |
+
"epoch": 2.0806087266674593,
|
251 |
+
"grad_norm": 0.8947279453277588,
|
252 |
+
"learning_rate": 4.479847818333135e-05,
|
253 |
+
"loss": 2.8203,
|
254 |
+
"step": 17500
|
255 |
+
},
|
256 |
+
{
|
257 |
+
"epoch": 2.1400546902865294,
|
258 |
+
"grad_norm": 0.8680304884910583,
|
259 |
+
"learning_rate": 4.4649863274283674e-05,
|
260 |
+
"loss": 2.8088,
|
261 |
+
"step": 18000
|
262 |
+
},
|
263 |
+
{
|
264 |
+
"epoch": 2.1995006539056,
|
265 |
+
"grad_norm": 0.8425644040107727,
|
266 |
+
"learning_rate": 4.4501248365236004e-05,
|
267 |
+
"loss": 2.8064,
|
268 |
+
"step": 18500
|
269 |
+
},
|
270 |
+
{
|
271 |
+
"epoch": 2.25894661752467,
|
272 |
+
"grad_norm": 0.9474213719367981,
|
273 |
+
"learning_rate": 4.4352633456188327e-05,
|
274 |
+
"loss": 2.7851,
|
275 |
+
"step": 19000
|
276 |
+
},
|
277 |
+
{
|
278 |
+
"epoch": 2.3183925811437405,
|
279 |
+
"grad_norm": 0.9292487502098083,
|
280 |
+
"learning_rate": 4.420401854714065e-05,
|
281 |
+
"loss": 2.8062,
|
282 |
+
"step": 19500
|
283 |
+
},
|
284 |
+
{
|
285 |
+
"epoch": 2.3778385447628105,
|
286 |
+
"grad_norm": 0.8527488708496094,
|
287 |
+
"learning_rate": 4.405540363809297e-05,
|
288 |
+
"loss": 2.7851,
|
289 |
+
"step": 20000
|
290 |
+
},
|
291 |
+
{
|
292 |
+
"epoch": 2.437284508381881,
|
293 |
+
"grad_norm": 0.9439261555671692,
|
294 |
+
"learning_rate": 4.39067887290453e-05,
|
295 |
+
"loss": 2.7873,
|
296 |
+
"step": 20500
|
297 |
+
},
|
298 |
+
{
|
299 |
+
"epoch": 2.496730472000951,
|
300 |
+
"grad_norm": 0.9343836903572083,
|
301 |
+
"learning_rate": 4.3758173819997624e-05,
|
302 |
+
"loss": 2.7611,
|
303 |
+
"step": 21000
|
304 |
+
},
|
305 |
+
{
|
306 |
+
"epoch": 2.5561764356200216,
|
307 |
+
"grad_norm": 0.9050599932670593,
|
308 |
+
"learning_rate": 4.360955891094995e-05,
|
309 |
+
"loss": 2.767,
|
310 |
+
"step": 21500
|
311 |
+
},
|
312 |
+
{
|
313 |
+
"epoch": 2.6156223992390917,
|
314 |
+
"grad_norm": 0.9053699374198914,
|
315 |
+
"learning_rate": 4.346094400190227e-05,
|
316 |
+
"loss": 2.7873,
|
317 |
+
"step": 22000
|
318 |
+
},
|
319 |
+
{
|
320 |
+
"epoch": 2.6750683628581617,
|
321 |
+
"grad_norm": 0.9282116293907166,
|
322 |
+
"learning_rate": 4.33123290928546e-05,
|
323 |
+
"loss": 2.7607,
|
324 |
+
"step": 22500
|
325 |
+
},
|
326 |
+
{
|
327 |
+
"epoch": 2.7345143264772322,
|
328 |
+
"grad_norm": 0.9617480635643005,
|
329 |
+
"learning_rate": 4.316371418380692e-05,
|
330 |
+
"loss": 2.7678,
|
331 |
+
"step": 23000
|
332 |
+
},
|
333 |
+
{
|
334 |
+
"epoch": 2.7939602900963023,
|
335 |
+
"grad_norm": 0.9725137948989868,
|
336 |
+
"learning_rate": 4.3015099274759244e-05,
|
337 |
+
"loss": 2.7665,
|
338 |
+
"step": 23500
|
339 |
+
},
|
340 |
+
{
|
341 |
+
"epoch": 2.853406253715373,
|
342 |
+
"grad_norm": 0.9514666199684143,
|
343 |
+
"learning_rate": 4.286648436571157e-05,
|
344 |
+
"loss": 2.7534,
|
345 |
+
"step": 24000
|
346 |
+
},
|
347 |
+
{
|
348 |
+
"epoch": 2.912852217334443,
|
349 |
+
"grad_norm": 0.9485461115837097,
|
350 |
+
"learning_rate": 4.27178694566639e-05,
|
351 |
+
"loss": 2.7306,
|
352 |
+
"step": 24500
|
353 |
+
},
|
354 |
+
{
|
355 |
+
"epoch": 2.9722981809535134,
|
356 |
+
"grad_norm": 1.014106035232544,
|
357 |
+
"learning_rate": 4.256925454761622e-05,
|
358 |
+
"loss": 2.736,
|
359 |
+
"step": 25000
|
360 |
+
},
|
361 |
+
{
|
362 |
+
"epoch": 3.0317441445725835,
|
363 |
+
"grad_norm": 0.9117903113365173,
|
364 |
+
"learning_rate": 4.242063963856854e-05,
|
365 |
+
"loss": 2.7278,
|
366 |
+
"step": 25500
|
367 |
+
},
|
368 |
+
{
|
369 |
+
"epoch": 3.091190108191654,
|
370 |
+
"grad_norm": 0.8904880881309509,
|
371 |
+
"learning_rate": 4.2272024729520865e-05,
|
372 |
+
"loss": 2.7156,
|
373 |
+
"step": 26000
|
374 |
+
},
|
375 |
+
{
|
376 |
+
"epoch": 3.150636071810724,
|
377 |
+
"grad_norm": 0.8653568625450134,
|
378 |
+
"learning_rate": 4.2123409820473194e-05,
|
379 |
+
"loss": 2.7137,
|
380 |
+
"step": 26500
|
381 |
+
},
|
382 |
+
{
|
383 |
+
"epoch": 3.210082035429794,
|
384 |
+
"grad_norm": 0.9386480450630188,
|
385 |
+
"learning_rate": 4.197479491142551e-05,
|
386 |
+
"loss": 2.7021,
|
387 |
+
"step": 27000
|
388 |
+
},
|
389 |
+
{
|
390 |
+
"epoch": 3.2695279990488646,
|
391 |
+
"grad_norm": 1.0122427940368652,
|
392 |
+
"learning_rate": 4.182618000237784e-05,
|
393 |
+
"loss": 2.699,
|
394 |
+
"step": 27500
|
395 |
+
},
|
396 |
+
{
|
397 |
+
"epoch": 3.3289739626679347,
|
398 |
+
"grad_norm": 0.9319558143615723,
|
399 |
+
"learning_rate": 4.167756509333017e-05,
|
400 |
+
"loss": 2.689,
|
401 |
+
"step": 28000
|
402 |
+
},
|
403 |
+
{
|
404 |
+
"epoch": 3.388419926287005,
|
405 |
+
"grad_norm": 0.9281746745109558,
|
406 |
+
"learning_rate": 4.152895018428249e-05,
|
407 |
+
"loss": 2.7027,
|
408 |
+
"step": 28500
|
409 |
+
},
|
410 |
+
{
|
411 |
+
"epoch": 3.4478658899060752,
|
412 |
+
"grad_norm": 0.9750462770462036,
|
413 |
+
"learning_rate": 4.1380335275234815e-05,
|
414 |
+
"loss": 2.6947,
|
415 |
+
"step": 29000
|
416 |
+
},
|
417 |
+
{
|
418 |
+
"epoch": 3.5073118535251457,
|
419 |
+
"grad_norm": 0.8887720704078674,
|
420 |
+
"learning_rate": 4.123172036618714e-05,
|
421 |
+
"loss": 2.6864,
|
422 |
+
"step": 29500
|
423 |
+
},
|
424 |
+
{
|
425 |
+
"epoch": 3.566757817144216,
|
426 |
+
"grad_norm": 0.9884176254272461,
|
427 |
+
"learning_rate": 4.108310545713947e-05,
|
428 |
+
"loss": 2.6893,
|
429 |
+
"step": 30000
|
430 |
+
},
|
431 |
+
{
|
432 |
+
"epoch": 3.6262037807632863,
|
433 |
+
"grad_norm": 0.9995080828666687,
|
434 |
+
"learning_rate": 4.093449054809178e-05,
|
435 |
+
"loss": 2.6734,
|
436 |
+
"step": 30500
|
437 |
+
},
|
438 |
+
{
|
439 |
+
"epoch": 3.6856497443823564,
|
440 |
+
"grad_norm": 1.0068608522415161,
|
441 |
+
"learning_rate": 4.078587563904411e-05,
|
442 |
+
"loss": 2.6766,
|
443 |
+
"step": 31000
|
444 |
+
},
|
445 |
+
{
|
446 |
+
"epoch": 3.7450957080014264,
|
447 |
+
"grad_norm": 1.0225422382354736,
|
448 |
+
"learning_rate": 4.0637260729996435e-05,
|
449 |
+
"loss": 2.6757,
|
450 |
+
"step": 31500
|
451 |
+
},
|
452 |
+
{
|
453 |
+
"epoch": 3.804541671620497,
|
454 |
+
"grad_norm": 0.9354658126831055,
|
455 |
+
"learning_rate": 4.0488645820948765e-05,
|
456 |
+
"loss": 2.6593,
|
457 |
+
"step": 32000
|
458 |
+
},
|
459 |
+
{
|
460 |
+
"epoch": 3.8639876352395675,
|
461 |
+
"grad_norm": 0.9209592938423157,
|
462 |
+
"learning_rate": 4.034003091190108e-05,
|
463 |
+
"loss": 2.6547,
|
464 |
+
"step": 32500
|
465 |
+
},
|
466 |
+
{
|
467 |
+
"epoch": 3.9234335988586375,
|
468 |
+
"grad_norm": 0.8945015668869019,
|
469 |
+
"learning_rate": 4.019141600285341e-05,
|
470 |
+
"loss": 2.6719,
|
471 |
+
"step": 33000
|
472 |
+
},
|
473 |
+
{
|
474 |
+
"epoch": 3.9828795624777076,
|
475 |
+
"grad_norm": 0.9823748469352722,
|
476 |
+
"learning_rate": 4.004280109380573e-05,
|
477 |
+
"loss": 2.6781,
|
478 |
+
"step": 33500
|
479 |
+
},
|
480 |
+
{
|
481 |
+
"epoch": 4.042325526096778,
|
482 |
+
"grad_norm": 1.0186822414398193,
|
483 |
+
"learning_rate": 3.989418618475806e-05,
|
484 |
+
"loss": 2.6469,
|
485 |
+
"step": 34000
|
486 |
+
},
|
487 |
+
{
|
488 |
+
"epoch": 4.101771489715849,
|
489 |
+
"grad_norm": 0.9255732297897339,
|
490 |
+
"learning_rate": 3.974557127571038e-05,
|
491 |
+
"loss": 2.6296,
|
492 |
+
"step": 34500
|
493 |
+
},
|
494 |
+
{
|
495 |
+
"epoch": 4.161217453334919,
|
496 |
+
"grad_norm": 1.0235294103622437,
|
497 |
+
"learning_rate": 3.959695636666271e-05,
|
498 |
+
"loss": 2.6358,
|
499 |
+
"step": 35000
|
500 |
+
},
|
501 |
+
{
|
502 |
+
"epoch": 4.220663416953989,
|
503 |
+
"grad_norm": 0.911547064781189,
|
504 |
+
"learning_rate": 3.944834145761503e-05,
|
505 |
+
"loss": 2.6354,
|
506 |
+
"step": 35500
|
507 |
+
},
|
508 |
+
{
|
509 |
+
"epoch": 4.280109380573059,
|
510 |
+
"grad_norm": 1.0124516487121582,
|
511 |
+
"learning_rate": 3.929972654856735e-05,
|
512 |
+
"loss": 2.6416,
|
513 |
+
"step": 36000
|
514 |
+
},
|
515 |
+
{
|
516 |
+
"epoch": 4.33955534419213,
|
517 |
+
"grad_norm": 1.0222316980361938,
|
518 |
+
"learning_rate": 3.9151111639519676e-05,
|
519 |
+
"loss": 2.6188,
|
520 |
+
"step": 36500
|
521 |
+
},
|
522 |
+
{
|
523 |
+
"epoch": 4.3990013078112,
|
524 |
+
"grad_norm": 0.9710135459899902,
|
525 |
+
"learning_rate": 3.9002496730472005e-05,
|
526 |
+
"loss": 2.6228,
|
527 |
+
"step": 37000
|
528 |
+
},
|
529 |
+
{
|
530 |
+
"epoch": 4.45844727143027,
|
531 |
+
"grad_norm": 1.0287182331085205,
|
532 |
+
"learning_rate": 3.885388182142433e-05,
|
533 |
+
"loss": 2.6067,
|
534 |
+
"step": 37500
|
535 |
+
},
|
536 |
+
{
|
537 |
+
"epoch": 4.51789323504934,
|
538 |
+
"grad_norm": 0.9699456095695496,
|
539 |
+
"learning_rate": 3.870526691237665e-05,
|
540 |
+
"loss": 2.6385,
|
541 |
+
"step": 38000
|
542 |
+
},
|
543 |
+
{
|
544 |
+
"epoch": 4.57733919866841,
|
545 |
+
"grad_norm": 0.9066009521484375,
|
546 |
+
"learning_rate": 3.855665200332897e-05,
|
547 |
+
"loss": 2.6284,
|
548 |
+
"step": 38500
|
549 |
+
},
|
550 |
+
{
|
551 |
+
"epoch": 4.636785162287481,
|
552 |
+
"grad_norm": 0.8537769317626953,
|
553 |
+
"learning_rate": 3.84080370942813e-05,
|
554 |
+
"loss": 2.6135,
|
555 |
+
"step": 39000
|
556 |
+
},
|
557 |
+
{
|
558 |
+
"epoch": 4.696231125906551,
|
559 |
+
"grad_norm": 1.0666980743408203,
|
560 |
+
"learning_rate": 3.8259422185233626e-05,
|
561 |
+
"loss": 2.6312,
|
562 |
+
"step": 39500
|
563 |
+
},
|
564 |
+
{
|
565 |
+
"epoch": 4.755677089525621,
|
566 |
+
"grad_norm": 1.0641474723815918,
|
567 |
+
"learning_rate": 3.811080727618595e-05,
|
568 |
+
"loss": 2.6127,
|
569 |
+
"step": 40000
|
570 |
+
},
|
571 |
+
{
|
572 |
+
"epoch": 4.815123053144691,
|
573 |
+
"grad_norm": 1.076323390007019,
|
574 |
+
"learning_rate": 3.796219236713827e-05,
|
575 |
+
"loss": 2.6184,
|
576 |
+
"step": 40500
|
577 |
+
},
|
578 |
+
{
|
579 |
+
"epoch": 4.874569016763762,
|
580 |
+
"grad_norm": 0.8963558077812195,
|
581 |
+
"learning_rate": 3.78135774580906e-05,
|
582 |
+
"loss": 2.6165,
|
583 |
+
"step": 41000
|
584 |
+
},
|
585 |
+
{
|
586 |
+
"epoch": 4.934014980382832,
|
587 |
+
"grad_norm": 0.968908429145813,
|
588 |
+
"learning_rate": 3.766496254904292e-05,
|
589 |
+
"loss": 2.6009,
|
590 |
+
"step": 41500
|
591 |
+
},
|
592 |
+
{
|
593 |
+
"epoch": 4.993460944001902,
|
594 |
+
"grad_norm": 0.9362033605575562,
|
595 |
+
"learning_rate": 3.7516347639995246e-05,
|
596 |
+
"loss": 2.5956,
|
597 |
+
"step": 42000
|
598 |
+
},
|
599 |
+
{
|
600 |
+
"epoch": 5.052906907620972,
|
601 |
+
"grad_norm": 1.1101199388504028,
|
602 |
+
"learning_rate": 3.736773273094757e-05,
|
603 |
+
"loss": 2.5755,
|
604 |
+
"step": 42500
|
605 |
+
},
|
606 |
+
{
|
607 |
+
"epoch": 5.112352871240043,
|
608 |
+
"grad_norm": 1.2178868055343628,
|
609 |
+
"learning_rate": 3.72191178218999e-05,
|
610 |
+
"loss": 2.5724,
|
611 |
+
"step": 43000
|
612 |
+
},
|
613 |
+
{
|
614 |
+
"epoch": 5.171798834859113,
|
615 |
+
"grad_norm": 1.0143418312072754,
|
616 |
+
"learning_rate": 3.707050291285222e-05,
|
617 |
+
"loss": 2.5834,
|
618 |
+
"step": 43500
|
619 |
+
},
|
620 |
+
{
|
621 |
+
"epoch": 5.231244798478183,
|
622 |
+
"grad_norm": 0.9720271825790405,
|
623 |
+
"learning_rate": 3.6921888003804544e-05,
|
624 |
+
"loss": 2.586,
|
625 |
+
"step": 44000
|
626 |
+
},
|
627 |
+
{
|
628 |
+
"epoch": 5.290690762097253,
|
629 |
+
"grad_norm": 0.8847070932388306,
|
630 |
+
"learning_rate": 3.6773273094756866e-05,
|
631 |
+
"loss": 2.5953,
|
632 |
+
"step": 44500
|
633 |
+
},
|
634 |
+
{
|
635 |
+
"epoch": 5.3501367257163235,
|
636 |
+
"grad_norm": 0.9654759764671326,
|
637 |
+
"learning_rate": 3.6624658185709196e-05,
|
638 |
+
"loss": 2.5777,
|
639 |
+
"step": 45000
|
640 |
+
},
|
641 |
+
{
|
642 |
+
"epoch": 5.409582689335394,
|
643 |
+
"grad_norm": 0.9272730350494385,
|
644 |
+
"learning_rate": 3.647604327666151e-05,
|
645 |
+
"loss": 2.5774,
|
646 |
+
"step": 45500
|
647 |
+
},
|
648 |
+
{
|
649 |
+
"epoch": 5.4690286529544645,
|
650 |
+
"grad_norm": 0.9674676656723022,
|
651 |
+
"learning_rate": 3.632742836761384e-05,
|
652 |
+
"loss": 2.5779,
|
653 |
+
"step": 46000
|
654 |
+
},
|
655 |
+
{
|
656 |
+
"epoch": 5.528474616573535,
|
657 |
+
"grad_norm": 1.0238367319107056,
|
658 |
+
"learning_rate": 3.6178813458566164e-05,
|
659 |
+
"loss": 2.5683,
|
660 |
+
"step": 46500
|
661 |
+
},
|
662 |
+
{
|
663 |
+
"epoch": 5.587920580192605,
|
664 |
+
"grad_norm": 1.1663753986358643,
|
665 |
+
"learning_rate": 3.603019854951849e-05,
|
666 |
+
"loss": 2.5802,
|
667 |
+
"step": 47000
|
668 |
+
},
|
669 |
+
{
|
670 |
+
"epoch": 5.647366543811675,
|
671 |
+
"grad_norm": 0.8961432576179504,
|
672 |
+
"learning_rate": 3.588158364047081e-05,
|
673 |
+
"loss": 2.5726,
|
674 |
+
"step": 47500
|
675 |
+
},
|
676 |
+
{
|
677 |
+
"epoch": 5.706812507430746,
|
678 |
+
"grad_norm": 1.1115467548370361,
|
679 |
+
"learning_rate": 3.573296873142314e-05,
|
680 |
+
"loss": 2.5719,
|
681 |
+
"step": 48000
|
682 |
+
},
|
683 |
+
{
|
684 |
+
"epoch": 5.766258471049816,
|
685 |
+
"grad_norm": 1.00434148311615,
|
686 |
+
"learning_rate": 3.558435382237546e-05,
|
687 |
+
"loss": 2.556,
|
688 |
+
"step": 48500
|
689 |
+
},
|
690 |
+
{
|
691 |
+
"epoch": 5.825704434668886,
|
692 |
+
"grad_norm": 1.1120518445968628,
|
693 |
+
"learning_rate": 3.5435738913327784e-05,
|
694 |
+
"loss": 2.5627,
|
695 |
+
"step": 49000
|
696 |
+
},
|
697 |
+
{
|
698 |
+
"epoch": 5.885150398287957,
|
699 |
+
"grad_norm": 0.9611983299255371,
|
700 |
+
"learning_rate": 3.528712400428011e-05,
|
701 |
+
"loss": 2.5568,
|
702 |
+
"step": 49500
|
703 |
+
},
|
704 |
+
{
|
705 |
+
"epoch": 5.944596361907027,
|
706 |
+
"grad_norm": 1.1176481246948242,
|
707 |
+
"learning_rate": 3.5138509095232436e-05,
|
708 |
+
"loss": 2.5634,
|
709 |
+
"step": 50000
|
710 |
+
},
|
711 |
+
{
|
712 |
+
"epoch": 6.004042325526097,
|
713 |
+
"grad_norm": 0.8676426410675049,
|
714 |
+
"learning_rate": 3.498989418618476e-05,
|
715 |
+
"loss": 2.5551,
|
716 |
+
"step": 50500
|
717 |
+
},
|
718 |
+
{
|
719 |
+
"epoch": 6.063488289145167,
|
720 |
+
"grad_norm": 0.8983253240585327,
|
721 |
+
"learning_rate": 3.484127927713708e-05,
|
722 |
+
"loss": 2.5442,
|
723 |
+
"step": 51000
|
724 |
+
},
|
725 |
+
{
|
726 |
+
"epoch": 6.122934252764237,
|
727 |
+
"grad_norm": 0.9558296203613281,
|
728 |
+
"learning_rate": 3.4692664368089405e-05,
|
729 |
+
"loss": 2.5415,
|
730 |
+
"step": 51500
|
731 |
+
},
|
732 |
+
{
|
733 |
+
"epoch": 6.182380216383308,
|
734 |
+
"grad_norm": 1.1759629249572754,
|
735 |
+
"learning_rate": 3.4544049459041734e-05,
|
736 |
+
"loss": 2.5186,
|
737 |
+
"step": 52000
|
738 |
+
},
|
739 |
+
{
|
740 |
+
"epoch": 6.241826180002378,
|
741 |
+
"grad_norm": 1.186232089996338,
|
742 |
+
"learning_rate": 3.439543454999406e-05,
|
743 |
+
"loss": 2.5437,
|
744 |
+
"step": 52500
|
745 |
+
},
|
746 |
+
{
|
747 |
+
"epoch": 6.301272143621448,
|
748 |
+
"grad_norm": 1.1072938442230225,
|
749 |
+
"learning_rate": 3.424681964094638e-05,
|
750 |
+
"loss": 2.5442,
|
751 |
+
"step": 53000
|
752 |
+
},
|
753 |
+
{
|
754 |
+
"epoch": 6.360718107240518,
|
755 |
+
"grad_norm": 1.1854956150054932,
|
756 |
+
"learning_rate": 3.40982047318987e-05,
|
757 |
+
"loss": 2.5265,
|
758 |
+
"step": 53500
|
759 |
+
},
|
760 |
+
{
|
761 |
+
"epoch": 6.420164070859588,
|
762 |
+
"grad_norm": 1.037420392036438,
|
763 |
+
"learning_rate": 3.394958982285103e-05,
|
764 |
+
"loss": 2.5101,
|
765 |
+
"step": 54000
|
766 |
+
},
|
767 |
+
{
|
768 |
+
"epoch": 6.479610034478659,
|
769 |
+
"grad_norm": 1.0414271354675293,
|
770 |
+
"learning_rate": 3.3800974913803354e-05,
|
771 |
+
"loss": 2.5291,
|
772 |
+
"step": 54500
|
773 |
+
},
|
774 |
+
{
|
775 |
+
"epoch": 6.539055998097729,
|
776 |
+
"grad_norm": 0.8827362656593323,
|
777 |
+
"learning_rate": 3.365236000475568e-05,
|
778 |
+
"loss": 2.5187,
|
779 |
+
"step": 55000
|
780 |
+
},
|
781 |
+
{
|
782 |
+
"epoch": 6.598501961716799,
|
783 |
+
"grad_norm": 0.9146121144294739,
|
784 |
+
"learning_rate": 3.3503745095708007e-05,
|
785 |
+
"loss": 2.5234,
|
786 |
+
"step": 55500
|
787 |
+
},
|
788 |
+
{
|
789 |
+
"epoch": 6.657947925335869,
|
790 |
+
"grad_norm": 1.0134857892990112,
|
791 |
+
"learning_rate": 3.335513018666033e-05,
|
792 |
+
"loss": 2.5199,
|
793 |
+
"step": 56000
|
794 |
+
},
|
795 |
+
{
|
796 |
+
"epoch": 6.71739388895494,
|
797 |
+
"grad_norm": 1.1852586269378662,
|
798 |
+
"learning_rate": 3.320651527761265e-05,
|
799 |
+
"loss": 2.5347,
|
800 |
+
"step": 56500
|
801 |
+
},
|
802 |
+
{
|
803 |
+
"epoch": 6.77683985257401,
|
804 |
+
"grad_norm": 1.0739943981170654,
|
805 |
+
"learning_rate": 3.3057900368564975e-05,
|
806 |
+
"loss": 2.5367,
|
807 |
+
"step": 57000
|
808 |
+
},
|
809 |
+
{
|
810 |
+
"epoch": 6.83628581619308,
|
811 |
+
"grad_norm": 0.9880659580230713,
|
812 |
+
"learning_rate": 3.2909285459517304e-05,
|
813 |
+
"loss": 2.5181,
|
814 |
+
"step": 57500
|
815 |
+
},
|
816 |
+
{
|
817 |
+
"epoch": 6.8957317798121505,
|
818 |
+
"grad_norm": 1.0519931316375732,
|
819 |
+
"learning_rate": 3.276067055046963e-05,
|
820 |
+
"loss": 2.5325,
|
821 |
+
"step": 58000
|
822 |
+
},
|
823 |
+
{
|
824 |
+
"epoch": 6.955177743431221,
|
825 |
+
"grad_norm": 0.9463315010070801,
|
826 |
+
"learning_rate": 3.261205564142195e-05,
|
827 |
+
"loss": 2.5384,
|
828 |
+
"step": 58500
|
829 |
+
},
|
830 |
+
{
|
831 |
+
"epoch": 7.0146237070502915,
|
832 |
+
"grad_norm": 0.9906750917434692,
|
833 |
+
"learning_rate": 3.246344073237427e-05,
|
834 |
+
"loss": 2.5374,
|
835 |
+
"step": 59000
|
836 |
+
},
|
837 |
+
{
|
838 |
+
"epoch": 7.0740696706693615,
|
839 |
+
"grad_norm": 0.9740419983863831,
|
840 |
+
"learning_rate": 3.23148258233266e-05,
|
841 |
+
"loss": 2.4919,
|
842 |
+
"step": 59500
|
843 |
+
},
|
844 |
+
{
|
845 |
+
"epoch": 7.133515634288432,
|
846 |
+
"grad_norm": 1.0209646224975586,
|
847 |
+
"learning_rate": 3.2166210914278925e-05,
|
848 |
+
"loss": 2.5065,
|
849 |
+
"step": 60000
|
850 |
+
},
|
851 |
+
{
|
852 |
+
"epoch": 7.192961597907502,
|
853 |
+
"grad_norm": 1.1537789106369019,
|
854 |
+
"learning_rate": 3.201759600523125e-05,
|
855 |
+
"loss": 2.4888,
|
856 |
+
"step": 60500
|
857 |
+
},
|
858 |
+
{
|
859 |
+
"epoch": 7.252407561526573,
|
860 |
+
"grad_norm": 1.0545387268066406,
|
861 |
+
"learning_rate": 3.186898109618357e-05,
|
862 |
+
"loss": 2.5042,
|
863 |
+
"step": 61000
|
864 |
+
},
|
865 |
+
{
|
866 |
+
"epoch": 7.311853525145643,
|
867 |
+
"grad_norm": 0.8990502953529358,
|
868 |
+
"learning_rate": 3.17203661871359e-05,
|
869 |
+
"loss": 2.4956,
|
870 |
+
"step": 61500
|
871 |
+
},
|
872 |
+
{
|
873 |
+
"epoch": 7.371299488764713,
|
874 |
+
"grad_norm": 1.0004386901855469,
|
875 |
+
"learning_rate": 3.1571751278088215e-05,
|
876 |
+
"loss": 2.5096,
|
877 |
+
"step": 62000
|
878 |
+
},
|
879 |
+
{
|
880 |
+
"epoch": 7.430745452383783,
|
881 |
+
"grad_norm": 1.192317008972168,
|
882 |
+
"learning_rate": 3.1423136369040545e-05,
|
883 |
+
"loss": 2.5038,
|
884 |
+
"step": 62500
|
885 |
+
},
|
886 |
+
{
|
887 |
+
"epoch": 7.490191416002854,
|
888 |
+
"grad_norm": 0.9577484726905823,
|
889 |
+
"learning_rate": 3.127452145999287e-05,
|
890 |
+
"loss": 2.5113,
|
891 |
+
"step": 63000
|
892 |
+
},
|
893 |
+
{
|
894 |
+
"epoch": 7.549637379621924,
|
895 |
+
"grad_norm": 0.8835137486457825,
|
896 |
+
"learning_rate": 3.11259065509452e-05,
|
897 |
+
"loss": 2.4939,
|
898 |
+
"step": 63500
|
899 |
+
},
|
900 |
+
{
|
901 |
+
"epoch": 7.609083343240994,
|
902 |
+
"grad_norm": 0.8289955258369446,
|
903 |
+
"learning_rate": 3.097729164189751e-05,
|
904 |
+
"loss": 2.4716,
|
905 |
+
"step": 64000
|
906 |
+
},
|
907 |
+
{
|
908 |
+
"epoch": 7.668529306860064,
|
909 |
+
"grad_norm": 0.9576908349990845,
|
910 |
+
"learning_rate": 3.082867673284984e-05,
|
911 |
+
"loss": 2.5109,
|
912 |
+
"step": 64500
|
913 |
+
},
|
914 |
+
{
|
915 |
+
"epoch": 7.727975270479135,
|
916 |
+
"grad_norm": 0.9045142531394958,
|
917 |
+
"learning_rate": 3.0680061823802165e-05,
|
918 |
+
"loss": 2.4811,
|
919 |
+
"step": 65000
|
920 |
+
},
|
921 |
+
{
|
922 |
+
"epoch": 7.787421234098205,
|
923 |
+
"grad_norm": 1.3150789737701416,
|
924 |
+
"learning_rate": 3.053144691475449e-05,
|
925 |
+
"loss": 2.505,
|
926 |
+
"step": 65500
|
927 |
+
},
|
928 |
+
{
|
929 |
+
"epoch": 7.846867197717275,
|
930 |
+
"grad_norm": 0.9815430641174316,
|
931 |
+
"learning_rate": 3.0382832005706814e-05,
|
932 |
+
"loss": 2.4923,
|
933 |
+
"step": 66000
|
934 |
+
},
|
935 |
+
{
|
936 |
+
"epoch": 7.906313161336345,
|
937 |
+
"grad_norm": 1.0355448722839355,
|
938 |
+
"learning_rate": 3.023421709665914e-05,
|
939 |
+
"loss": 2.4867,
|
940 |
+
"step": 66500
|
941 |
+
},
|
942 |
+
{
|
943 |
+
"epoch": 7.965759124955415,
|
944 |
+
"grad_norm": 1.0244001150131226,
|
945 |
+
"learning_rate": 3.0085602187611463e-05,
|
946 |
+
"loss": 2.4973,
|
947 |
+
"step": 67000
|
948 |
+
},
|
949 |
+
{
|
950 |
+
"epoch": 8.025205088574486,
|
951 |
+
"grad_norm": 1.052660584449768,
|
952 |
+
"learning_rate": 2.993698727856379e-05,
|
953 |
+
"loss": 2.4976,
|
954 |
+
"step": 67500
|
955 |
+
},
|
956 |
+
{
|
957 |
+
"epoch": 8.084651052193555,
|
958 |
+
"grad_norm": 1.1590783596038818,
|
959 |
+
"learning_rate": 2.978837236951611e-05,
|
960 |
+
"loss": 2.4631,
|
961 |
+
"step": 68000
|
962 |
+
},
|
963 |
+
{
|
964 |
+
"epoch": 8.144097015812626,
|
965 |
+
"grad_norm": 0.9065755605697632,
|
966 |
+
"learning_rate": 2.9639757460468438e-05,
|
967 |
+
"loss": 2.4494,
|
968 |
+
"step": 68500
|
969 |
+
},
|
970 |
+
{
|
971 |
+
"epoch": 8.203542979431697,
|
972 |
+
"grad_norm": 0.9562356472015381,
|
973 |
+
"learning_rate": 2.9491142551420757e-05,
|
974 |
+
"loss": 2.4728,
|
975 |
+
"step": 69000
|
976 |
+
},
|
977 |
+
{
|
978 |
+
"epoch": 8.262988943050766,
|
979 |
+
"grad_norm": 0.9509665966033936,
|
980 |
+
"learning_rate": 2.9342527642373087e-05,
|
981 |
+
"loss": 2.4747,
|
982 |
+
"step": 69500
|
983 |
+
},
|
984 |
+
{
|
985 |
+
"epoch": 8.322434906669837,
|
986 |
+
"grad_norm": 0.9384153485298157,
|
987 |
+
"learning_rate": 2.9193912733325406e-05,
|
988 |
+
"loss": 2.4745,
|
989 |
+
"step": 70000
|
990 |
+
},
|
991 |
+
{
|
992 |
+
"epoch": 8.381880870288906,
|
993 |
+
"grad_norm": 0.9459151029586792,
|
994 |
+
"learning_rate": 2.9045297824277735e-05,
|
995 |
+
"loss": 2.476,
|
996 |
+
"step": 70500
|
997 |
+
},
|
998 |
+
{
|
999 |
+
"epoch": 8.441326833907977,
|
1000 |
+
"grad_norm": 0.9553677439689636,
|
1001 |
+
"learning_rate": 2.8896682915230055e-05,
|
1002 |
+
"loss": 2.4753,
|
1003 |
+
"step": 71000
|
1004 |
+
},
|
1005 |
+
{
|
1006 |
+
"epoch": 8.500772797527048,
|
1007 |
+
"grad_norm": 1.014932632446289,
|
1008 |
+
"learning_rate": 2.8748068006182384e-05,
|
1009 |
+
"loss": 2.4647,
|
1010 |
+
"step": 71500
|
1011 |
+
},
|
1012 |
+
{
|
1013 |
+
"epoch": 8.560218761146118,
|
1014 |
+
"grad_norm": 0.990463376045227,
|
1015 |
+
"learning_rate": 2.8599453097134704e-05,
|
1016 |
+
"loss": 2.4782,
|
1017 |
+
"step": 72000
|
1018 |
+
},
|
1019 |
+
{
|
1020 |
+
"epoch": 8.619664724765189,
|
1021 |
+
"grad_norm": 0.892906665802002,
|
1022 |
+
"learning_rate": 2.845083818808703e-05,
|
1023 |
+
"loss": 2.4736,
|
1024 |
+
"step": 72500
|
1025 |
+
},
|
1026 |
+
{
|
1027 |
+
"epoch": 8.67911068838426,
|
1028 |
+
"grad_norm": 0.9943811297416687,
|
1029 |
+
"learning_rate": 2.8302223279039352e-05,
|
1030 |
+
"loss": 2.4554,
|
1031 |
+
"step": 73000
|
1032 |
+
},
|
1033 |
+
{
|
1034 |
+
"epoch": 8.738556652003329,
|
1035 |
+
"grad_norm": 0.9325155019760132,
|
1036 |
+
"learning_rate": 2.815360836999168e-05,
|
1037 |
+
"loss": 2.4703,
|
1038 |
+
"step": 73500
|
1039 |
+
},
|
1040 |
+
{
|
1041 |
+
"epoch": 8.7980026156224,
|
1042 |
+
"grad_norm": 0.9389231204986572,
|
1043 |
+
"learning_rate": 2.8004993460944e-05,
|
1044 |
+
"loss": 2.4727,
|
1045 |
+
"step": 74000
|
1046 |
+
},
|
1047 |
+
{
|
1048 |
+
"epoch": 8.857448579241469,
|
1049 |
+
"grad_norm": 0.9121980667114258,
|
1050 |
+
"learning_rate": 2.7856378551896327e-05,
|
1051 |
+
"loss": 2.4533,
|
1052 |
+
"step": 74500
|
1053 |
+
},
|
1054 |
+
{
|
1055 |
+
"epoch": 8.91689454286054,
|
1056 |
+
"grad_norm": 1.046366572380066,
|
1057 |
+
"learning_rate": 2.770776364284865e-05,
|
1058 |
+
"loss": 2.4652,
|
1059 |
+
"step": 75000
|
1060 |
+
},
|
1061 |
+
{
|
1062 |
+
"epoch": 8.97634050647961,
|
1063 |
+
"grad_norm": 1.0157803297042847,
|
1064 |
+
"learning_rate": 2.7559148733800976e-05,
|
1065 |
+
"loss": 2.4701,
|
1066 |
+
"step": 75500
|
1067 |
+
},
|
1068 |
+
{
|
1069 |
+
"epoch": 9.03578647009868,
|
1070 |
+
"grad_norm": 1.1012301445007324,
|
1071 |
+
"learning_rate": 2.74105338247533e-05,
|
1072 |
+
"loss": 2.4491,
|
1073 |
+
"step": 76000
|
1074 |
+
},
|
1075 |
+
{
|
1076 |
+
"epoch": 9.09523243371775,
|
1077 |
+
"grad_norm": 1.000829815864563,
|
1078 |
+
"learning_rate": 2.7261918915705625e-05,
|
1079 |
+
"loss": 2.4434,
|
1080 |
+
"step": 76500
|
1081 |
+
},
|
1082 |
+
{
|
1083 |
+
"epoch": 9.15467839733682,
|
1084 |
+
"grad_norm": 1.028676986694336,
|
1085 |
+
"learning_rate": 2.7113304006657948e-05,
|
1086 |
+
"loss": 2.4392,
|
1087 |
+
"step": 77000
|
1088 |
+
},
|
1089 |
+
{
|
1090 |
+
"epoch": 9.214124360955891,
|
1091 |
+
"grad_norm": 1.0821462869644165,
|
1092 |
+
"learning_rate": 2.6964689097610274e-05,
|
1093 |
+
"loss": 2.4289,
|
1094 |
+
"step": 77500
|
1095 |
+
},
|
1096 |
+
{
|
1097 |
+
"epoch": 9.273570324574962,
|
1098 |
+
"grad_norm": 0.951738715171814,
|
1099 |
+
"learning_rate": 2.6816074188562596e-05,
|
1100 |
+
"loss": 2.4437,
|
1101 |
+
"step": 78000
|
1102 |
+
},
|
1103 |
+
{
|
1104 |
+
"epoch": 9.333016288194031,
|
1105 |
+
"grad_norm": 0.9170756936073303,
|
1106 |
+
"learning_rate": 2.6667459279514923e-05,
|
1107 |
+
"loss": 2.4507,
|
1108 |
+
"step": 78500
|
1109 |
+
},
|
1110 |
+
{
|
1111 |
+
"epoch": 9.392462251813102,
|
1112 |
+
"grad_norm": 0.9591684937477112,
|
1113 |
+
"learning_rate": 2.6518844370467245e-05,
|
1114 |
+
"loss": 2.4584,
|
1115 |
+
"step": 79000
|
1116 |
+
},
|
1117 |
+
{
|
1118 |
+
"epoch": 9.451908215432173,
|
1119 |
+
"grad_norm": 1.1289016008377075,
|
1120 |
+
"learning_rate": 2.637022946141957e-05,
|
1121 |
+
"loss": 2.4595,
|
1122 |
+
"step": 79500
|
1123 |
+
},
|
1124 |
+
{
|
1125 |
+
"epoch": 9.511354179051242,
|
1126 |
+
"grad_norm": 1.0114785432815552,
|
1127 |
+
"learning_rate": 2.6221614552371894e-05,
|
1128 |
+
"loss": 2.4404,
|
1129 |
+
"step": 80000
|
1130 |
+
},
|
1131 |
+
{
|
1132 |
+
"epoch": 9.570800142670313,
|
1133 |
+
"grad_norm": 1.1835304498672485,
|
1134 |
+
"learning_rate": 2.607299964332422e-05,
|
1135 |
+
"loss": 2.4308,
|
1136 |
+
"step": 80500
|
1137 |
+
},
|
1138 |
+
{
|
1139 |
+
"epoch": 9.630246106289382,
|
1140 |
+
"grad_norm": 0.9822309017181396,
|
1141 |
+
"learning_rate": 2.592438473427654e-05,
|
1142 |
+
"loss": 2.4387,
|
1143 |
+
"step": 81000
|
1144 |
+
},
|
1145 |
+
{
|
1146 |
+
"epoch": 9.689692069908453,
|
1147 |
+
"grad_norm": 1.114311695098877,
|
1148 |
+
"learning_rate": 2.577576982522887e-05,
|
1149 |
+
"loss": 2.4519,
|
1150 |
+
"step": 81500
|
1151 |
+
},
|
1152 |
+
{
|
1153 |
+
"epoch": 9.749138033527524,
|
1154 |
+
"grad_norm": 1.1047866344451904,
|
1155 |
+
"learning_rate": 2.5627154916181195e-05,
|
1156 |
+
"loss": 2.4497,
|
1157 |
+
"step": 82000
|
1158 |
+
},
|
1159 |
+
{
|
1160 |
+
"epoch": 9.808583997146593,
|
1161 |
+
"grad_norm": 0.9930892586708069,
|
1162 |
+
"learning_rate": 2.5478540007133518e-05,
|
1163 |
+
"loss": 2.4489,
|
1164 |
+
"step": 82500
|
1165 |
+
},
|
1166 |
+
{
|
1167 |
+
"epoch": 9.868029960765664,
|
1168 |
+
"grad_norm": 1.1107361316680908,
|
1169 |
+
"learning_rate": 2.5329925098085844e-05,
|
1170 |
+
"loss": 2.4399,
|
1171 |
+
"step": 83000
|
1172 |
+
},
|
1173 |
+
{
|
1174 |
+
"epoch": 9.927475924384733,
|
1175 |
+
"grad_norm": 1.0770343542099,
|
1176 |
+
"learning_rate": 2.5181310189038167e-05,
|
1177 |
+
"loss": 2.4362,
|
1178 |
+
"step": 83500
|
1179 |
+
},
|
1180 |
+
{
|
1181 |
+
"epoch": 9.986921888003804,
|
1182 |
+
"grad_norm": 0.9818819761276245,
|
1183 |
+
"learning_rate": 2.5032695279990493e-05,
|
1184 |
+
"loss": 2.4418,
|
1185 |
+
"step": 84000
|
1186 |
+
},
|
1187 |
+
{
|
1188 |
+
"epoch": 10.046367851622875,
|
1189 |
+
"grad_norm": 1.1135622262954712,
|
1190 |
+
"learning_rate": 2.4884080370942815e-05,
|
1191 |
+
"loss": 2.428,
|
1192 |
+
"step": 84500
|
1193 |
+
},
|
1194 |
+
{
|
1195 |
+
"epoch": 10.105813815241945,
|
1196 |
+
"grad_norm": 1.035888671875,
|
1197 |
+
"learning_rate": 2.4735465461895138e-05,
|
1198 |
+
"loss": 2.4193,
|
1199 |
+
"step": 85000
|
1200 |
+
},
|
1201 |
+
{
|
1202 |
+
"epoch": 10.165259778861016,
|
1203 |
+
"grad_norm": 0.9694905281066895,
|
1204 |
+
"learning_rate": 2.458685055284746e-05,
|
1205 |
+
"loss": 2.4165,
|
1206 |
+
"step": 85500
|
1207 |
+
},
|
1208 |
+
{
|
1209 |
+
"epoch": 10.224705742480086,
|
1210 |
+
"grad_norm": 1.116449236869812,
|
1211 |
+
"learning_rate": 2.4438235643799787e-05,
|
1212 |
+
"loss": 2.4122,
|
1213 |
+
"step": 86000
|
1214 |
+
},
|
1215 |
+
{
|
1216 |
+
"epoch": 10.284151706099156,
|
1217 |
+
"grad_norm": 0.9860423803329468,
|
1218 |
+
"learning_rate": 2.428962073475211e-05,
|
1219 |
+
"loss": 2.4173,
|
1220 |
+
"step": 86500
|
1221 |
+
},
|
1222 |
+
{
|
1223 |
+
"epoch": 10.343597669718227,
|
1224 |
+
"grad_norm": 1.1727473735809326,
|
1225 |
+
"learning_rate": 2.4141005825704436e-05,
|
1226 |
+
"loss": 2.4258,
|
1227 |
+
"step": 87000
|
1228 |
+
},
|
1229 |
+
{
|
1230 |
+
"epoch": 10.403043633337296,
|
1231 |
+
"grad_norm": 1.0731017589569092,
|
1232 |
+
"learning_rate": 2.399239091665676e-05,
|
1233 |
+
"loss": 2.4289,
|
1234 |
+
"step": 87500
|
1235 |
+
},
|
1236 |
+
{
|
1237 |
+
"epoch": 10.462489596956367,
|
1238 |
+
"grad_norm": 1.0740883350372314,
|
1239 |
+
"learning_rate": 2.3843776007609085e-05,
|
1240 |
+
"loss": 2.4142,
|
1241 |
+
"step": 88000
|
1242 |
+
},
|
1243 |
+
{
|
1244 |
+
"epoch": 10.521935560575438,
|
1245 |
+
"grad_norm": 1.1342713832855225,
|
1246 |
+
"learning_rate": 2.3695161098561407e-05,
|
1247 |
+
"loss": 2.4315,
|
1248 |
+
"step": 88500
|
1249 |
+
},
|
1250 |
+
{
|
1251 |
+
"epoch": 10.581381524194507,
|
1252 |
+
"grad_norm": 1.0230334997177124,
|
1253 |
+
"learning_rate": 2.3546546189513733e-05,
|
1254 |
+
"loss": 2.4352,
|
1255 |
+
"step": 89000
|
1256 |
+
},
|
1257 |
+
{
|
1258 |
+
"epoch": 10.640827487813578,
|
1259 |
+
"grad_norm": 1.0113749504089355,
|
1260 |
+
"learning_rate": 2.3397931280466056e-05,
|
1261 |
+
"loss": 2.4128,
|
1262 |
+
"step": 89500
|
1263 |
+
},
|
1264 |
+
{
|
1265 |
+
"epoch": 10.700273451432647,
|
1266 |
+
"grad_norm": 1.0363703966140747,
|
1267 |
+
"learning_rate": 2.3249316371418382e-05,
|
1268 |
+
"loss": 2.4343,
|
1269 |
+
"step": 90000
|
1270 |
+
},
|
1271 |
+
{
|
1272 |
+
"epoch": 10.759719415051718,
|
1273 |
+
"grad_norm": 1.0065736770629883,
|
1274 |
+
"learning_rate": 2.3100701462370705e-05,
|
1275 |
+
"loss": 2.4268,
|
1276 |
+
"step": 90500
|
1277 |
+
},
|
1278 |
+
{
|
1279 |
+
"epoch": 10.819165378670789,
|
1280 |
+
"grad_norm": 0.949798047542572,
|
1281 |
+
"learning_rate": 2.295208655332303e-05,
|
1282 |
+
"loss": 2.4114,
|
1283 |
+
"step": 91000
|
1284 |
+
},
|
1285 |
+
{
|
1286 |
+
"epoch": 10.878611342289858,
|
1287 |
+
"grad_norm": 0.9772433042526245,
|
1288 |
+
"learning_rate": 2.2803471644275354e-05,
|
1289 |
+
"loss": 2.4187,
|
1290 |
+
"step": 91500
|
1291 |
+
},
|
1292 |
+
{
|
1293 |
+
"epoch": 10.938057305908929,
|
1294 |
+
"grad_norm": 0.9436720609664917,
|
1295 |
+
"learning_rate": 2.2654856735227677e-05,
|
1296 |
+
"loss": 2.4151,
|
1297 |
+
"step": 92000
|
1298 |
+
},
|
1299 |
+
{
|
1300 |
+
"epoch": 10.997503269528,
|
1301 |
+
"grad_norm": 0.9903433918952942,
|
1302 |
+
"learning_rate": 2.2506241826180003e-05,
|
1303 |
+
"loss": 2.4332,
|
1304 |
+
"step": 92500
|
1305 |
+
},
|
1306 |
+
{
|
1307 |
+
"epoch": 11.05694923314707,
|
1308 |
+
"grad_norm": 0.9285963177680969,
|
1309 |
+
"learning_rate": 2.2357626917132325e-05,
|
1310 |
+
"loss": 2.3895,
|
1311 |
+
"step": 93000
|
1312 |
+
},
|
1313 |
+
{
|
1314 |
+
"epoch": 11.11639519676614,
|
1315 |
+
"grad_norm": 1.0996205806732178,
|
1316 |
+
"learning_rate": 2.220901200808465e-05,
|
1317 |
+
"loss": 2.3858,
|
1318 |
+
"step": 93500
|
1319 |
+
},
|
1320 |
+
{
|
1321 |
+
"epoch": 11.17584116038521,
|
1322 |
+
"grad_norm": 0.9550360441207886,
|
1323 |
+
"learning_rate": 2.2060397099036974e-05,
|
1324 |
+
"loss": 2.4016,
|
1325 |
+
"step": 94000
|
1326 |
+
},
|
1327 |
+
{
|
1328 |
+
"epoch": 11.23528712400428,
|
1329 |
+
"grad_norm": 1.3018606901168823,
|
1330 |
+
"learning_rate": 2.19117821899893e-05,
|
1331 |
+
"loss": 2.4031,
|
1332 |
+
"step": 94500
|
1333 |
+
},
|
1334 |
+
{
|
1335 |
+
"epoch": 11.294733087623351,
|
1336 |
+
"grad_norm": 0.9388914704322815,
|
1337 |
+
"learning_rate": 2.1763167280941626e-05,
|
1338 |
+
"loss": 2.4094,
|
1339 |
+
"step": 95000
|
1340 |
+
},
|
1341 |
+
{
|
1342 |
+
"epoch": 11.35417905124242,
|
1343 |
+
"grad_norm": 0.9850655794143677,
|
1344 |
+
"learning_rate": 2.161455237189395e-05,
|
1345 |
+
"loss": 2.4054,
|
1346 |
+
"step": 95500
|
1347 |
+
},
|
1348 |
+
{
|
1349 |
+
"epoch": 11.413625014861491,
|
1350 |
+
"grad_norm": 1.038522481918335,
|
1351 |
+
"learning_rate": 2.1465937462846275e-05,
|
1352 |
+
"loss": 2.3895,
|
1353 |
+
"step": 96000
|
1354 |
+
},
|
1355 |
+
{
|
1356 |
+
"epoch": 11.47307097848056,
|
1357 |
+
"grad_norm": 1.0989197492599487,
|
1358 |
+
"learning_rate": 2.1317322553798598e-05,
|
1359 |
+
"loss": 2.4019,
|
1360 |
+
"step": 96500
|
1361 |
+
},
|
1362 |
+
{
|
1363 |
+
"epoch": 11.532516942099631,
|
1364 |
+
"grad_norm": 1.0527700185775757,
|
1365 |
+
"learning_rate": 2.1168707644750924e-05,
|
1366 |
+
"loss": 2.399,
|
1367 |
+
"step": 97000
|
1368 |
+
},
|
1369 |
+
{
|
1370 |
+
"epoch": 11.591962905718702,
|
1371 |
+
"grad_norm": 1.273655652999878,
|
1372 |
+
"learning_rate": 2.1020092735703247e-05,
|
1373 |
+
"loss": 2.4259,
|
1374 |
+
"step": 97500
|
1375 |
+
},
|
1376 |
+
{
|
1377 |
+
"epoch": 11.651408869337772,
|
1378 |
+
"grad_norm": 1.002064824104309,
|
1379 |
+
"learning_rate": 2.0871477826655573e-05,
|
1380 |
+
"loss": 2.4073,
|
1381 |
+
"step": 98000
|
1382 |
+
},
|
1383 |
+
{
|
1384 |
+
"epoch": 11.710854832956842,
|
1385 |
+
"grad_norm": 0.9922045469284058,
|
1386 |
+
"learning_rate": 2.0722862917607896e-05,
|
1387 |
+
"loss": 2.4059,
|
1388 |
+
"step": 98500
|
1389 |
+
},
|
1390 |
+
{
|
1391 |
+
"epoch": 11.770300796575913,
|
1392 |
+
"grad_norm": 0.9962035417556763,
|
1393 |
+
"learning_rate": 2.057424800856022e-05,
|
1394 |
+
"loss": 2.4174,
|
1395 |
+
"step": 99000
|
1396 |
+
},
|
1397 |
+
{
|
1398 |
+
"epoch": 11.829746760194983,
|
1399 |
+
"grad_norm": 1.0998961925506592,
|
1400 |
+
"learning_rate": 2.0425633099512544e-05,
|
1401 |
+
"loss": 2.4133,
|
1402 |
+
"step": 99500
|
1403 |
+
},
|
1404 |
+
{
|
1405 |
+
"epoch": 11.889192723814054,
|
1406 |
+
"grad_norm": 1.0380686521530151,
|
1407 |
+
"learning_rate": 2.027701819046487e-05,
|
1408 |
+
"loss": 2.414,
|
1409 |
+
"step": 100000
|
1410 |
+
}
|
1411 |
+
],
|
1412 |
+
"logging_steps": 500,
|
1413 |
+
"max_steps": 168220,
|
1414 |
+
"num_input_tokens_seen": 0,
|
1415 |
+
"num_train_epochs": 20,
|
1416 |
+
"save_steps": 500,
|
1417 |
+
"total_flos": 4.1803850907648e+17,
|
1418 |
+
"train_batch_size": 8,
|
1419 |
+
"trial_name": null,
|
1420 |
+
"trial_params": null
|
1421 |
+
}
|
training_args.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:7163e8e93997d9542b4813116cd402aa38eade3ba35ae0069435235054002545
|
3 |
+
size 4920
|
vocab.json
ADDED
The diff for this file is too large to render.
See raw diff
|
|