Initial commit from mrm8488
Browse files- added_tokens.json +1 -0
- config.json +34 -0
- eval_results_clm.txt +1 -0
- merges.txt +0 -0
- pytorch_model.bin +3 -0
- special_tokens_map.json +1 -0
- tokenizer_config.json +1 -0
- train_results.txt +3 -0
- trainer_state.json +3136 -0
- training_args.bin +3 -0
- vocab.json +0 -0
added_tokens.json
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
{"<|UNKNOWN|>": 50001, "<EOL>": 50000}
|
config.json
ADDED
@@ -0,0 +1,34 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_name_or_path": "microsoft/CodeGPT-small-py",
|
3 |
+
"_num_labels": 2,
|
4 |
+
"activation_function": "gelu_new",
|
5 |
+
"architectures": [
|
6 |
+
"GPT2LMHeadModel"
|
7 |
+
],
|
8 |
+
"attn_pdrop": 0.1,
|
9 |
+
"bos_token_id": 0,
|
10 |
+
"embd_pdrop": 0.1,
|
11 |
+
"eos_token_id": 2,
|
12 |
+
"gradient_checkpointing": false,
|
13 |
+
"initializer_range": 0.02,
|
14 |
+
"layer_norm_epsilon": 1e-05,
|
15 |
+
"model_type": "gpt2",
|
16 |
+
"n_ctx": 1024,
|
17 |
+
"n_embd": 768,
|
18 |
+
"n_head": 12,
|
19 |
+
"n_inner": null,
|
20 |
+
"n_layer": 12,
|
21 |
+
"n_positions": 1024,
|
22 |
+
"output_past": true,
|
23 |
+
"pad_token_id": 1,
|
24 |
+
"resid_pdrop": 0.1,
|
25 |
+
"summary_activation": null,
|
26 |
+
"summary_first_dropout": 0.1,
|
27 |
+
"summary_proj_to_labels": true,
|
28 |
+
"summary_type": "cls_index",
|
29 |
+
"summary_use_proj": true,
|
30 |
+
"transformers_version": "4.2.2",
|
31 |
+
"use_cache": true,
|
32 |
+
"vocab_size": 50002,
|
33 |
+
"xla_device": true
|
34 |
+
}
|
eval_results_clm.txt
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
perplexity = 2.50151493424726
|
merges.txt
ADDED
The diff for this file is too large to render.
See raw diff
|
|
pytorch_model.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:6022c5b6847caa82819b999a2ae09b0b58f0caa5af373e114890c6f9d3d1c8d7
|
3 |
+
size 663219890
|
special_tokens_map.json
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
{"bos_token": {"content": "<s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, "eos_token": {"content": "</s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, "unk_token": {"content": "<|UNKNOWN|>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, "sep_token": "<EOL>", "pad_token": "<pad>"}
|
tokenizer_config.json
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
{"unk_token": {"content": "<|UNKNOWN|>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "__type": "AddedToken"}, "bos_token": {"content": "<s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "__type": "AddedToken"}, "eos_token": {"content": "</s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "__type": "AddedToken"}, "add_prefix_space": false, "sep_token": "<EOL>", "pad_token": "<pad>", "special_tokens_map_file": null, "full_tokenizer_file": null, "name_or_path": "microsoft/CodeGPT-small-py", "errors": "replace"}
|
train_results.txt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
epoch = 5.0
|
2 |
+
train_runtime = 11347.9361
|
3 |
+
train_samples_per_second = 2.29
|
trainer_state.json
ADDED
@@ -0,0 +1,3136 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"best_metric": null,
|
3 |
+
"best_model_checkpoint": null,
|
4 |
+
"epoch": 5.0,
|
5 |
+
"global_step": 25985,
|
6 |
+
"is_hyper_param_search": false,
|
7 |
+
"is_local_process_zero": true,
|
8 |
+
"is_world_process_zero": true,
|
9 |
+
"log_history": [
|
10 |
+
{
|
11 |
+
"epoch": 0.01,
|
12 |
+
"learning_rate": 4.9903790648451034e-05,
|
13 |
+
"loss": 1.5568,
|
14 |
+
"step": 50
|
15 |
+
},
|
16 |
+
{
|
17 |
+
"epoch": 0.02,
|
18 |
+
"learning_rate": 4.980758129690206e-05,
|
19 |
+
"loss": 1.3161,
|
20 |
+
"step": 100
|
21 |
+
},
|
22 |
+
{
|
23 |
+
"epoch": 0.03,
|
24 |
+
"learning_rate": 4.971137194535309e-05,
|
25 |
+
"loss": 1.2764,
|
26 |
+
"step": 150
|
27 |
+
},
|
28 |
+
{
|
29 |
+
"epoch": 0.04,
|
30 |
+
"learning_rate": 4.961516259380412e-05,
|
31 |
+
"loss": 1.2186,
|
32 |
+
"step": 200
|
33 |
+
},
|
34 |
+
{
|
35 |
+
"epoch": 0.05,
|
36 |
+
"learning_rate": 4.951895324225515e-05,
|
37 |
+
"loss": 1.2072,
|
38 |
+
"step": 250
|
39 |
+
},
|
40 |
+
{
|
41 |
+
"epoch": 0.06,
|
42 |
+
"learning_rate": 4.942274389070618e-05,
|
43 |
+
"loss": 1.2057,
|
44 |
+
"step": 300
|
45 |
+
},
|
46 |
+
{
|
47 |
+
"epoch": 0.07,
|
48 |
+
"learning_rate": 4.932653453915721e-05,
|
49 |
+
"loss": 1.2048,
|
50 |
+
"step": 350
|
51 |
+
},
|
52 |
+
{
|
53 |
+
"epoch": 0.08,
|
54 |
+
"learning_rate": 4.923032518760824e-05,
|
55 |
+
"loss": 1.1741,
|
56 |
+
"step": 400
|
57 |
+
},
|
58 |
+
{
|
59 |
+
"epoch": 0.09,
|
60 |
+
"learning_rate": 4.913411583605927e-05,
|
61 |
+
"loss": 1.2347,
|
62 |
+
"step": 450
|
63 |
+
},
|
64 |
+
{
|
65 |
+
"epoch": 0.1,
|
66 |
+
"learning_rate": 4.9037906484510296e-05,
|
67 |
+
"loss": 1.1057,
|
68 |
+
"step": 500
|
69 |
+
},
|
70 |
+
{
|
71 |
+
"epoch": 0.11,
|
72 |
+
"learning_rate": 4.894169713296133e-05,
|
73 |
+
"loss": 1.2702,
|
74 |
+
"step": 550
|
75 |
+
},
|
76 |
+
{
|
77 |
+
"epoch": 0.12,
|
78 |
+
"learning_rate": 4.884548778141236e-05,
|
79 |
+
"loss": 1.1257,
|
80 |
+
"step": 600
|
81 |
+
},
|
82 |
+
{
|
83 |
+
"epoch": 0.13,
|
84 |
+
"learning_rate": 4.874927842986339e-05,
|
85 |
+
"loss": 1.093,
|
86 |
+
"step": 650
|
87 |
+
},
|
88 |
+
{
|
89 |
+
"epoch": 0.13,
|
90 |
+
"learning_rate": 4.8653069078314415e-05,
|
91 |
+
"loss": 1.1402,
|
92 |
+
"step": 700
|
93 |
+
},
|
94 |
+
{
|
95 |
+
"epoch": 0.14,
|
96 |
+
"learning_rate": 4.855685972676544e-05,
|
97 |
+
"loss": 1.1661,
|
98 |
+
"step": 750
|
99 |
+
},
|
100 |
+
{
|
101 |
+
"epoch": 0.15,
|
102 |
+
"learning_rate": 4.846065037521647e-05,
|
103 |
+
"loss": 1.1504,
|
104 |
+
"step": 800
|
105 |
+
},
|
106 |
+
{
|
107 |
+
"epoch": 0.16,
|
108 |
+
"learning_rate": 4.83644410236675e-05,
|
109 |
+
"loss": 1.1579,
|
110 |
+
"step": 850
|
111 |
+
},
|
112 |
+
{
|
113 |
+
"epoch": 0.17,
|
114 |
+
"learning_rate": 4.826823167211853e-05,
|
115 |
+
"loss": 1.1235,
|
116 |
+
"step": 900
|
117 |
+
},
|
118 |
+
{
|
119 |
+
"epoch": 0.18,
|
120 |
+
"learning_rate": 4.817202232056956e-05,
|
121 |
+
"loss": 1.1364,
|
122 |
+
"step": 950
|
123 |
+
},
|
124 |
+
{
|
125 |
+
"epoch": 0.19,
|
126 |
+
"learning_rate": 4.807581296902059e-05,
|
127 |
+
"loss": 1.1273,
|
128 |
+
"step": 1000
|
129 |
+
},
|
130 |
+
{
|
131 |
+
"epoch": 0.2,
|
132 |
+
"learning_rate": 4.797960361747162e-05,
|
133 |
+
"loss": 1.1337,
|
134 |
+
"step": 1050
|
135 |
+
},
|
136 |
+
{
|
137 |
+
"epoch": 0.21,
|
138 |
+
"learning_rate": 4.7883394265922646e-05,
|
139 |
+
"loss": 1.0788,
|
140 |
+
"step": 1100
|
141 |
+
},
|
142 |
+
{
|
143 |
+
"epoch": 0.22,
|
144 |
+
"learning_rate": 4.778718491437368e-05,
|
145 |
+
"loss": 1.1334,
|
146 |
+
"step": 1150
|
147 |
+
},
|
148 |
+
{
|
149 |
+
"epoch": 0.23,
|
150 |
+
"learning_rate": 4.769097556282471e-05,
|
151 |
+
"loss": 1.0406,
|
152 |
+
"step": 1200
|
153 |
+
},
|
154 |
+
{
|
155 |
+
"epoch": 0.24,
|
156 |
+
"learning_rate": 4.759476621127574e-05,
|
157 |
+
"loss": 1.0592,
|
158 |
+
"step": 1250
|
159 |
+
},
|
160 |
+
{
|
161 |
+
"epoch": 0.25,
|
162 |
+
"learning_rate": 4.7498556859726765e-05,
|
163 |
+
"loss": 1.1162,
|
164 |
+
"step": 1300
|
165 |
+
},
|
166 |
+
{
|
167 |
+
"epoch": 0.26,
|
168 |
+
"learning_rate": 4.7402347508177797e-05,
|
169 |
+
"loss": 1.1041,
|
170 |
+
"step": 1350
|
171 |
+
},
|
172 |
+
{
|
173 |
+
"epoch": 0.27,
|
174 |
+
"learning_rate": 4.730613815662883e-05,
|
175 |
+
"loss": 1.1012,
|
176 |
+
"step": 1400
|
177 |
+
},
|
178 |
+
{
|
179 |
+
"epoch": 0.28,
|
180 |
+
"learning_rate": 4.720992880507986e-05,
|
181 |
+
"loss": 1.0975,
|
182 |
+
"step": 1450
|
183 |
+
},
|
184 |
+
{
|
185 |
+
"epoch": 0.29,
|
186 |
+
"learning_rate": 4.7113719453530884e-05,
|
187 |
+
"loss": 1.0853,
|
188 |
+
"step": 1500
|
189 |
+
},
|
190 |
+
{
|
191 |
+
"epoch": 0.3,
|
192 |
+
"learning_rate": 4.7017510101981915e-05,
|
193 |
+
"loss": 1.1258,
|
194 |
+
"step": 1550
|
195 |
+
},
|
196 |
+
{
|
197 |
+
"epoch": 0.31,
|
198 |
+
"learning_rate": 4.692130075043295e-05,
|
199 |
+
"loss": 1.0874,
|
200 |
+
"step": 1600
|
201 |
+
},
|
202 |
+
{
|
203 |
+
"epoch": 0.32,
|
204 |
+
"learning_rate": 4.682509139888397e-05,
|
205 |
+
"loss": 1.0703,
|
206 |
+
"step": 1650
|
207 |
+
},
|
208 |
+
{
|
209 |
+
"epoch": 0.33,
|
210 |
+
"learning_rate": 4.6728882047335e-05,
|
211 |
+
"loss": 1.078,
|
212 |
+
"step": 1700
|
213 |
+
},
|
214 |
+
{
|
215 |
+
"epoch": 0.34,
|
216 |
+
"learning_rate": 4.6632672695786034e-05,
|
217 |
+
"loss": 1.0578,
|
218 |
+
"step": 1750
|
219 |
+
},
|
220 |
+
{
|
221 |
+
"epoch": 0.35,
|
222 |
+
"learning_rate": 4.6536463344237066e-05,
|
223 |
+
"loss": 1.0748,
|
224 |
+
"step": 1800
|
225 |
+
},
|
226 |
+
{
|
227 |
+
"epoch": 0.36,
|
228 |
+
"learning_rate": 4.644025399268809e-05,
|
229 |
+
"loss": 1.0652,
|
230 |
+
"step": 1850
|
231 |
+
},
|
232 |
+
{
|
233 |
+
"epoch": 0.37,
|
234 |
+
"learning_rate": 4.634404464113912e-05,
|
235 |
+
"loss": 1.0861,
|
236 |
+
"step": 1900
|
237 |
+
},
|
238 |
+
{
|
239 |
+
"epoch": 0.38,
|
240 |
+
"learning_rate": 4.624783528959015e-05,
|
241 |
+
"loss": 1.1258,
|
242 |
+
"step": 1950
|
243 |
+
},
|
244 |
+
{
|
245 |
+
"epoch": 0.38,
|
246 |
+
"learning_rate": 4.6151625938041185e-05,
|
247 |
+
"loss": 1.0661,
|
248 |
+
"step": 2000
|
249 |
+
},
|
250 |
+
{
|
251 |
+
"epoch": 0.39,
|
252 |
+
"learning_rate": 4.605541658649221e-05,
|
253 |
+
"loss": 1.0718,
|
254 |
+
"step": 2050
|
255 |
+
},
|
256 |
+
{
|
257 |
+
"epoch": 0.4,
|
258 |
+
"learning_rate": 4.595920723494324e-05,
|
259 |
+
"loss": 1.0186,
|
260 |
+
"step": 2100
|
261 |
+
},
|
262 |
+
{
|
263 |
+
"epoch": 0.41,
|
264 |
+
"learning_rate": 4.586299788339427e-05,
|
265 |
+
"loss": 1.0994,
|
266 |
+
"step": 2150
|
267 |
+
},
|
268 |
+
{
|
269 |
+
"epoch": 0.42,
|
270 |
+
"learning_rate": 4.57667885318453e-05,
|
271 |
+
"loss": 1.063,
|
272 |
+
"step": 2200
|
273 |
+
},
|
274 |
+
{
|
275 |
+
"epoch": 0.43,
|
276 |
+
"learning_rate": 4.567057918029632e-05,
|
277 |
+
"loss": 1.0748,
|
278 |
+
"step": 2250
|
279 |
+
},
|
280 |
+
{
|
281 |
+
"epoch": 0.44,
|
282 |
+
"learning_rate": 4.557436982874735e-05,
|
283 |
+
"loss": 1.0637,
|
284 |
+
"step": 2300
|
285 |
+
},
|
286 |
+
{
|
287 |
+
"epoch": 0.45,
|
288 |
+
"learning_rate": 4.5478160477198384e-05,
|
289 |
+
"loss": 1.0256,
|
290 |
+
"step": 2350
|
291 |
+
},
|
292 |
+
{
|
293 |
+
"epoch": 0.46,
|
294 |
+
"learning_rate": 4.5381951125649416e-05,
|
295 |
+
"loss": 1.0493,
|
296 |
+
"step": 2400
|
297 |
+
},
|
298 |
+
{
|
299 |
+
"epoch": 0.47,
|
300 |
+
"learning_rate": 4.528574177410044e-05,
|
301 |
+
"loss": 1.0789,
|
302 |
+
"step": 2450
|
303 |
+
},
|
304 |
+
{
|
305 |
+
"epoch": 0.48,
|
306 |
+
"learning_rate": 4.518953242255147e-05,
|
307 |
+
"loss": 1.0564,
|
308 |
+
"step": 2500
|
309 |
+
},
|
310 |
+
{
|
311 |
+
"epoch": 0.49,
|
312 |
+
"learning_rate": 4.50933230710025e-05,
|
313 |
+
"loss": 1.0308,
|
314 |
+
"step": 2550
|
315 |
+
},
|
316 |
+
{
|
317 |
+
"epoch": 0.5,
|
318 |
+
"learning_rate": 4.4997113719453535e-05,
|
319 |
+
"loss": 1.0008,
|
320 |
+
"step": 2600
|
321 |
+
},
|
322 |
+
{
|
323 |
+
"epoch": 0.51,
|
324 |
+
"learning_rate": 4.490090436790456e-05,
|
325 |
+
"loss": 0.9971,
|
326 |
+
"step": 2650
|
327 |
+
},
|
328 |
+
{
|
329 |
+
"epoch": 0.52,
|
330 |
+
"learning_rate": 4.480469501635559e-05,
|
331 |
+
"loss": 1.02,
|
332 |
+
"step": 2700
|
333 |
+
},
|
334 |
+
{
|
335 |
+
"epoch": 0.53,
|
336 |
+
"learning_rate": 4.470848566480662e-05,
|
337 |
+
"loss": 1.0672,
|
338 |
+
"step": 2750
|
339 |
+
},
|
340 |
+
{
|
341 |
+
"epoch": 0.54,
|
342 |
+
"learning_rate": 4.4612276313257654e-05,
|
343 |
+
"loss": 1.0252,
|
344 |
+
"step": 2800
|
345 |
+
},
|
346 |
+
{
|
347 |
+
"epoch": 0.55,
|
348 |
+
"learning_rate": 4.451606696170868e-05,
|
349 |
+
"loss": 1.0235,
|
350 |
+
"step": 2850
|
351 |
+
},
|
352 |
+
{
|
353 |
+
"epoch": 0.56,
|
354 |
+
"learning_rate": 4.441985761015971e-05,
|
355 |
+
"loss": 1.0952,
|
356 |
+
"step": 2900
|
357 |
+
},
|
358 |
+
{
|
359 |
+
"epoch": 0.57,
|
360 |
+
"learning_rate": 4.432364825861074e-05,
|
361 |
+
"loss": 1.0167,
|
362 |
+
"step": 2950
|
363 |
+
},
|
364 |
+
{
|
365 |
+
"epoch": 0.58,
|
366 |
+
"learning_rate": 4.4227438907061766e-05,
|
367 |
+
"loss": 1.0493,
|
368 |
+
"step": 3000
|
369 |
+
},
|
370 |
+
{
|
371 |
+
"epoch": 0.59,
|
372 |
+
"learning_rate": 4.41312295555128e-05,
|
373 |
+
"loss": 1.0299,
|
374 |
+
"step": 3050
|
375 |
+
},
|
376 |
+
{
|
377 |
+
"epoch": 0.6,
|
378 |
+
"learning_rate": 4.403502020396383e-05,
|
379 |
+
"loss": 1.0101,
|
380 |
+
"step": 3100
|
381 |
+
},
|
382 |
+
{
|
383 |
+
"epoch": 0.61,
|
384 |
+
"learning_rate": 4.393881085241486e-05,
|
385 |
+
"loss": 1.0351,
|
386 |
+
"step": 3150
|
387 |
+
},
|
388 |
+
{
|
389 |
+
"epoch": 0.62,
|
390 |
+
"learning_rate": 4.3842601500865885e-05,
|
391 |
+
"loss": 1.0602,
|
392 |
+
"step": 3200
|
393 |
+
},
|
394 |
+
{
|
395 |
+
"epoch": 0.63,
|
396 |
+
"learning_rate": 4.3746392149316916e-05,
|
397 |
+
"loss": 1.0591,
|
398 |
+
"step": 3250
|
399 |
+
},
|
400 |
+
{
|
401 |
+
"epoch": 0.63,
|
402 |
+
"learning_rate": 4.365018279776795e-05,
|
403 |
+
"loss": 1.1272,
|
404 |
+
"step": 3300
|
405 |
+
},
|
406 |
+
{
|
407 |
+
"epoch": 0.64,
|
408 |
+
"learning_rate": 4.355397344621898e-05,
|
409 |
+
"loss": 1.0016,
|
410 |
+
"step": 3350
|
411 |
+
},
|
412 |
+
{
|
413 |
+
"epoch": 0.65,
|
414 |
+
"learning_rate": 4.3457764094670004e-05,
|
415 |
+
"loss": 1.0356,
|
416 |
+
"step": 3400
|
417 |
+
},
|
418 |
+
{
|
419 |
+
"epoch": 0.66,
|
420 |
+
"learning_rate": 4.3361554743121035e-05,
|
421 |
+
"loss": 1.0675,
|
422 |
+
"step": 3450
|
423 |
+
},
|
424 |
+
{
|
425 |
+
"epoch": 0.67,
|
426 |
+
"learning_rate": 4.3265345391572066e-05,
|
427 |
+
"loss": 1.081,
|
428 |
+
"step": 3500
|
429 |
+
},
|
430 |
+
{
|
431 |
+
"epoch": 0.68,
|
432 |
+
"learning_rate": 4.31691360400231e-05,
|
433 |
+
"loss": 1.0639,
|
434 |
+
"step": 3550
|
435 |
+
},
|
436 |
+
{
|
437 |
+
"epoch": 0.69,
|
438 |
+
"learning_rate": 4.307292668847412e-05,
|
439 |
+
"loss": 1.0362,
|
440 |
+
"step": 3600
|
441 |
+
},
|
442 |
+
{
|
443 |
+
"epoch": 0.7,
|
444 |
+
"learning_rate": 4.297671733692515e-05,
|
445 |
+
"loss": 1.0322,
|
446 |
+
"step": 3650
|
447 |
+
},
|
448 |
+
{
|
449 |
+
"epoch": 0.71,
|
450 |
+
"learning_rate": 4.288050798537618e-05,
|
451 |
+
"loss": 1.047,
|
452 |
+
"step": 3700
|
453 |
+
},
|
454 |
+
{
|
455 |
+
"epoch": 0.72,
|
456 |
+
"learning_rate": 4.278429863382721e-05,
|
457 |
+
"loss": 1.0495,
|
458 |
+
"step": 3750
|
459 |
+
},
|
460 |
+
{
|
461 |
+
"epoch": 0.73,
|
462 |
+
"learning_rate": 4.2688089282278235e-05,
|
463 |
+
"loss": 0.9846,
|
464 |
+
"step": 3800
|
465 |
+
},
|
466 |
+
{
|
467 |
+
"epoch": 0.74,
|
468 |
+
"learning_rate": 4.2591879930729266e-05,
|
469 |
+
"loss": 1.0603,
|
470 |
+
"step": 3850
|
471 |
+
},
|
472 |
+
{
|
473 |
+
"epoch": 0.75,
|
474 |
+
"learning_rate": 4.24956705791803e-05,
|
475 |
+
"loss": 1.0081,
|
476 |
+
"step": 3900
|
477 |
+
},
|
478 |
+
{
|
479 |
+
"epoch": 0.76,
|
480 |
+
"learning_rate": 4.239946122763133e-05,
|
481 |
+
"loss": 1.0277,
|
482 |
+
"step": 3950
|
483 |
+
},
|
484 |
+
{
|
485 |
+
"epoch": 0.77,
|
486 |
+
"learning_rate": 4.2303251876082353e-05,
|
487 |
+
"loss": 1.0452,
|
488 |
+
"step": 4000
|
489 |
+
},
|
490 |
+
{
|
491 |
+
"epoch": 0.78,
|
492 |
+
"learning_rate": 4.2207042524533385e-05,
|
493 |
+
"loss": 1.0459,
|
494 |
+
"step": 4050
|
495 |
+
},
|
496 |
+
{
|
497 |
+
"epoch": 0.79,
|
498 |
+
"learning_rate": 4.2110833172984416e-05,
|
499 |
+
"loss": 1.0238,
|
500 |
+
"step": 4100
|
501 |
+
},
|
502 |
+
{
|
503 |
+
"epoch": 0.8,
|
504 |
+
"learning_rate": 4.201462382143545e-05,
|
505 |
+
"loss": 1.0076,
|
506 |
+
"step": 4150
|
507 |
+
},
|
508 |
+
{
|
509 |
+
"epoch": 0.81,
|
510 |
+
"learning_rate": 4.191841446988647e-05,
|
511 |
+
"loss": 1.0593,
|
512 |
+
"step": 4200
|
513 |
+
},
|
514 |
+
{
|
515 |
+
"epoch": 0.82,
|
516 |
+
"learning_rate": 4.1822205118337504e-05,
|
517 |
+
"loss": 0.9765,
|
518 |
+
"step": 4250
|
519 |
+
},
|
520 |
+
{
|
521 |
+
"epoch": 0.83,
|
522 |
+
"learning_rate": 4.1725995766788535e-05,
|
523 |
+
"loss": 1.0129,
|
524 |
+
"step": 4300
|
525 |
+
},
|
526 |
+
{
|
527 |
+
"epoch": 0.84,
|
528 |
+
"learning_rate": 4.162978641523957e-05,
|
529 |
+
"loss": 1.0691,
|
530 |
+
"step": 4350
|
531 |
+
},
|
532 |
+
{
|
533 |
+
"epoch": 0.85,
|
534 |
+
"learning_rate": 4.153357706369059e-05,
|
535 |
+
"loss": 1.0025,
|
536 |
+
"step": 4400
|
537 |
+
},
|
538 |
+
{
|
539 |
+
"epoch": 0.86,
|
540 |
+
"learning_rate": 4.143736771214162e-05,
|
541 |
+
"loss": 1.0469,
|
542 |
+
"step": 4450
|
543 |
+
},
|
544 |
+
{
|
545 |
+
"epoch": 0.87,
|
546 |
+
"learning_rate": 4.1341158360592654e-05,
|
547 |
+
"loss": 0.9761,
|
548 |
+
"step": 4500
|
549 |
+
},
|
550 |
+
{
|
551 |
+
"epoch": 0.88,
|
552 |
+
"learning_rate": 4.124494900904368e-05,
|
553 |
+
"loss": 1.0338,
|
554 |
+
"step": 4550
|
555 |
+
},
|
556 |
+
{
|
557 |
+
"epoch": 0.89,
|
558 |
+
"learning_rate": 4.114873965749471e-05,
|
559 |
+
"loss": 1.0535,
|
560 |
+
"step": 4600
|
561 |
+
},
|
562 |
+
{
|
563 |
+
"epoch": 0.89,
|
564 |
+
"learning_rate": 4.105253030594574e-05,
|
565 |
+
"loss": 1.0271,
|
566 |
+
"step": 4650
|
567 |
+
},
|
568 |
+
{
|
569 |
+
"epoch": 0.9,
|
570 |
+
"learning_rate": 4.095632095439677e-05,
|
571 |
+
"loss": 0.9932,
|
572 |
+
"step": 4700
|
573 |
+
},
|
574 |
+
{
|
575 |
+
"epoch": 0.91,
|
576 |
+
"learning_rate": 4.08601116028478e-05,
|
577 |
+
"loss": 1.046,
|
578 |
+
"step": 4750
|
579 |
+
},
|
580 |
+
{
|
581 |
+
"epoch": 0.92,
|
582 |
+
"learning_rate": 4.076390225129883e-05,
|
583 |
+
"loss": 1.0436,
|
584 |
+
"step": 4800
|
585 |
+
},
|
586 |
+
{
|
587 |
+
"epoch": 0.93,
|
588 |
+
"learning_rate": 4.066769289974986e-05,
|
589 |
+
"loss": 1.0411,
|
590 |
+
"step": 4850
|
591 |
+
},
|
592 |
+
{
|
593 |
+
"epoch": 0.94,
|
594 |
+
"learning_rate": 4.057148354820089e-05,
|
595 |
+
"loss": 1.0217,
|
596 |
+
"step": 4900
|
597 |
+
},
|
598 |
+
{
|
599 |
+
"epoch": 0.95,
|
600 |
+
"learning_rate": 4.0475274196651917e-05,
|
601 |
+
"loss": 1.0135,
|
602 |
+
"step": 4950
|
603 |
+
},
|
604 |
+
{
|
605 |
+
"epoch": 0.96,
|
606 |
+
"learning_rate": 4.037906484510295e-05,
|
607 |
+
"loss": 1.0147,
|
608 |
+
"step": 5000
|
609 |
+
},
|
610 |
+
{
|
611 |
+
"epoch": 0.97,
|
612 |
+
"learning_rate": 4.028285549355398e-05,
|
613 |
+
"loss": 1.0569,
|
614 |
+
"step": 5050
|
615 |
+
},
|
616 |
+
{
|
617 |
+
"epoch": 0.98,
|
618 |
+
"learning_rate": 4.0186646142005004e-05,
|
619 |
+
"loss": 0.9712,
|
620 |
+
"step": 5100
|
621 |
+
},
|
622 |
+
{
|
623 |
+
"epoch": 0.99,
|
624 |
+
"learning_rate": 4.009043679045603e-05,
|
625 |
+
"loss": 1.0107,
|
626 |
+
"step": 5150
|
627 |
+
},
|
628 |
+
{
|
629 |
+
"epoch": 1.0,
|
630 |
+
"learning_rate": 3.999422743890706e-05,
|
631 |
+
"loss": 0.9829,
|
632 |
+
"step": 5200
|
633 |
+
},
|
634 |
+
{
|
635 |
+
"epoch": 1.01,
|
636 |
+
"learning_rate": 3.989801808735809e-05,
|
637 |
+
"loss": 1.0263,
|
638 |
+
"step": 5250
|
639 |
+
},
|
640 |
+
{
|
641 |
+
"epoch": 1.02,
|
642 |
+
"learning_rate": 3.980180873580912e-05,
|
643 |
+
"loss": 1.0463,
|
644 |
+
"step": 5300
|
645 |
+
},
|
646 |
+
{
|
647 |
+
"epoch": 1.03,
|
648 |
+
"learning_rate": 3.970559938426015e-05,
|
649 |
+
"loss": 0.9776,
|
650 |
+
"step": 5350
|
651 |
+
},
|
652 |
+
{
|
653 |
+
"epoch": 1.04,
|
654 |
+
"learning_rate": 3.960939003271118e-05,
|
655 |
+
"loss": 0.9697,
|
656 |
+
"step": 5400
|
657 |
+
},
|
658 |
+
{
|
659 |
+
"epoch": 1.05,
|
660 |
+
"learning_rate": 3.951318068116221e-05,
|
661 |
+
"loss": 0.9636,
|
662 |
+
"step": 5450
|
663 |
+
},
|
664 |
+
{
|
665 |
+
"epoch": 1.06,
|
666 |
+
"learning_rate": 3.941697132961324e-05,
|
667 |
+
"loss": 0.9746,
|
668 |
+
"step": 5500
|
669 |
+
},
|
670 |
+
{
|
671 |
+
"epoch": 1.07,
|
672 |
+
"learning_rate": 3.9320761978064267e-05,
|
673 |
+
"loss": 0.9786,
|
674 |
+
"step": 5550
|
675 |
+
},
|
676 |
+
{
|
677 |
+
"epoch": 1.08,
|
678 |
+
"learning_rate": 3.92245526265153e-05,
|
679 |
+
"loss": 1.0174,
|
680 |
+
"step": 5600
|
681 |
+
},
|
682 |
+
{
|
683 |
+
"epoch": 1.09,
|
684 |
+
"learning_rate": 3.912834327496633e-05,
|
685 |
+
"loss": 0.9955,
|
686 |
+
"step": 5650
|
687 |
+
},
|
688 |
+
{
|
689 |
+
"epoch": 1.1,
|
690 |
+
"learning_rate": 3.903213392341736e-05,
|
691 |
+
"loss": 0.9633,
|
692 |
+
"step": 5700
|
693 |
+
},
|
694 |
+
{
|
695 |
+
"epoch": 1.11,
|
696 |
+
"learning_rate": 3.8935924571868385e-05,
|
697 |
+
"loss": 0.9661,
|
698 |
+
"step": 5750
|
699 |
+
},
|
700 |
+
{
|
701 |
+
"epoch": 1.12,
|
702 |
+
"learning_rate": 3.883971522031942e-05,
|
703 |
+
"loss": 0.9731,
|
704 |
+
"step": 5800
|
705 |
+
},
|
706 |
+
{
|
707 |
+
"epoch": 1.13,
|
708 |
+
"learning_rate": 3.874350586877045e-05,
|
709 |
+
"loss": 0.986,
|
710 |
+
"step": 5850
|
711 |
+
},
|
712 |
+
{
|
713 |
+
"epoch": 1.14,
|
714 |
+
"learning_rate": 3.864729651722147e-05,
|
715 |
+
"loss": 0.9579,
|
716 |
+
"step": 5900
|
717 |
+
},
|
718 |
+
{
|
719 |
+
"epoch": 1.14,
|
720 |
+
"learning_rate": 3.8551087165672504e-05,
|
721 |
+
"loss": 0.9643,
|
722 |
+
"step": 5950
|
723 |
+
},
|
724 |
+
{
|
725 |
+
"epoch": 1.15,
|
726 |
+
"learning_rate": 3.8454877814123536e-05,
|
727 |
+
"loss": 1.0001,
|
728 |
+
"step": 6000
|
729 |
+
},
|
730 |
+
{
|
731 |
+
"epoch": 1.16,
|
732 |
+
"learning_rate": 3.835866846257457e-05,
|
733 |
+
"loss": 0.931,
|
734 |
+
"step": 6050
|
735 |
+
},
|
736 |
+
{
|
737 |
+
"epoch": 1.17,
|
738 |
+
"learning_rate": 3.826245911102559e-05,
|
739 |
+
"loss": 0.9706,
|
740 |
+
"step": 6100
|
741 |
+
},
|
742 |
+
{
|
743 |
+
"epoch": 1.18,
|
744 |
+
"learning_rate": 3.816624975947662e-05,
|
745 |
+
"loss": 1.0086,
|
746 |
+
"step": 6150
|
747 |
+
},
|
748 |
+
{
|
749 |
+
"epoch": 1.19,
|
750 |
+
"learning_rate": 3.8070040407927655e-05,
|
751 |
+
"loss": 0.9136,
|
752 |
+
"step": 6200
|
753 |
+
},
|
754 |
+
{
|
755 |
+
"epoch": 1.2,
|
756 |
+
"learning_rate": 3.7973831056378686e-05,
|
757 |
+
"loss": 0.9657,
|
758 |
+
"step": 6250
|
759 |
+
},
|
760 |
+
{
|
761 |
+
"epoch": 1.21,
|
762 |
+
"learning_rate": 3.787762170482971e-05,
|
763 |
+
"loss": 1.0179,
|
764 |
+
"step": 6300
|
765 |
+
},
|
766 |
+
{
|
767 |
+
"epoch": 1.22,
|
768 |
+
"learning_rate": 3.778141235328074e-05,
|
769 |
+
"loss": 0.9634,
|
770 |
+
"step": 6350
|
771 |
+
},
|
772 |
+
{
|
773 |
+
"epoch": 1.23,
|
774 |
+
"learning_rate": 3.7685203001731774e-05,
|
775 |
+
"loss": 0.956,
|
776 |
+
"step": 6400
|
777 |
+
},
|
778 |
+
{
|
779 |
+
"epoch": 1.24,
|
780 |
+
"learning_rate": 3.7588993650182805e-05,
|
781 |
+
"loss": 0.9316,
|
782 |
+
"step": 6450
|
783 |
+
},
|
784 |
+
{
|
785 |
+
"epoch": 1.25,
|
786 |
+
"learning_rate": 3.749278429863382e-05,
|
787 |
+
"loss": 0.9615,
|
788 |
+
"step": 6500
|
789 |
+
},
|
790 |
+
{
|
791 |
+
"epoch": 1.26,
|
792 |
+
"learning_rate": 3.7396574947084854e-05,
|
793 |
+
"loss": 1.0051,
|
794 |
+
"step": 6550
|
795 |
+
},
|
796 |
+
{
|
797 |
+
"epoch": 1.27,
|
798 |
+
"learning_rate": 3.7300365595535886e-05,
|
799 |
+
"loss": 0.9738,
|
800 |
+
"step": 6600
|
801 |
+
},
|
802 |
+
{
|
803 |
+
"epoch": 1.28,
|
804 |
+
"learning_rate": 3.720415624398692e-05,
|
805 |
+
"loss": 0.9919,
|
806 |
+
"step": 6650
|
807 |
+
},
|
808 |
+
{
|
809 |
+
"epoch": 1.29,
|
810 |
+
"learning_rate": 3.710794689243794e-05,
|
811 |
+
"loss": 1.0258,
|
812 |
+
"step": 6700
|
813 |
+
},
|
814 |
+
{
|
815 |
+
"epoch": 1.3,
|
816 |
+
"learning_rate": 3.701173754088897e-05,
|
817 |
+
"loss": 0.95,
|
818 |
+
"step": 6750
|
819 |
+
},
|
820 |
+
{
|
821 |
+
"epoch": 1.31,
|
822 |
+
"learning_rate": 3.6915528189340005e-05,
|
823 |
+
"loss": 0.9876,
|
824 |
+
"step": 6800
|
825 |
+
},
|
826 |
+
{
|
827 |
+
"epoch": 1.32,
|
828 |
+
"learning_rate": 3.6819318837791036e-05,
|
829 |
+
"loss": 0.9641,
|
830 |
+
"step": 6850
|
831 |
+
},
|
832 |
+
{
|
833 |
+
"epoch": 1.33,
|
834 |
+
"learning_rate": 3.672310948624206e-05,
|
835 |
+
"loss": 0.9604,
|
836 |
+
"step": 6900
|
837 |
+
},
|
838 |
+
{
|
839 |
+
"epoch": 1.34,
|
840 |
+
"learning_rate": 3.662690013469309e-05,
|
841 |
+
"loss": 0.9775,
|
842 |
+
"step": 6950
|
843 |
+
},
|
844 |
+
{
|
845 |
+
"epoch": 1.35,
|
846 |
+
"learning_rate": 3.6530690783144124e-05,
|
847 |
+
"loss": 0.9484,
|
848 |
+
"step": 7000
|
849 |
+
},
|
850 |
+
{
|
851 |
+
"epoch": 1.36,
|
852 |
+
"learning_rate": 3.6434481431595155e-05,
|
853 |
+
"loss": 0.9288,
|
854 |
+
"step": 7050
|
855 |
+
},
|
856 |
+
{
|
857 |
+
"epoch": 1.37,
|
858 |
+
"learning_rate": 3.633827208004618e-05,
|
859 |
+
"loss": 0.9688,
|
860 |
+
"step": 7100
|
861 |
+
},
|
862 |
+
{
|
863 |
+
"epoch": 1.38,
|
864 |
+
"learning_rate": 3.624206272849721e-05,
|
865 |
+
"loss": 0.9436,
|
866 |
+
"step": 7150
|
867 |
+
},
|
868 |
+
{
|
869 |
+
"epoch": 1.39,
|
870 |
+
"learning_rate": 3.614585337694824e-05,
|
871 |
+
"loss": 0.9607,
|
872 |
+
"step": 7200
|
873 |
+
},
|
874 |
+
{
|
875 |
+
"epoch": 1.4,
|
876 |
+
"learning_rate": 3.6049644025399274e-05,
|
877 |
+
"loss": 0.9548,
|
878 |
+
"step": 7250
|
879 |
+
},
|
880 |
+
{
|
881 |
+
"epoch": 1.4,
|
882 |
+
"learning_rate": 3.59534346738503e-05,
|
883 |
+
"loss": 0.9638,
|
884 |
+
"step": 7300
|
885 |
+
},
|
886 |
+
{
|
887 |
+
"epoch": 1.41,
|
888 |
+
"learning_rate": 3.585722532230133e-05,
|
889 |
+
"loss": 0.9608,
|
890 |
+
"step": 7350
|
891 |
+
},
|
892 |
+
{
|
893 |
+
"epoch": 1.42,
|
894 |
+
"learning_rate": 3.576101597075236e-05,
|
895 |
+
"loss": 0.9739,
|
896 |
+
"step": 7400
|
897 |
+
},
|
898 |
+
{
|
899 |
+
"epoch": 1.43,
|
900 |
+
"learning_rate": 3.5664806619203386e-05,
|
901 |
+
"loss": 0.9444,
|
902 |
+
"step": 7450
|
903 |
+
},
|
904 |
+
{
|
905 |
+
"epoch": 1.44,
|
906 |
+
"learning_rate": 3.556859726765442e-05,
|
907 |
+
"loss": 1.0029,
|
908 |
+
"step": 7500
|
909 |
+
},
|
910 |
+
{
|
911 |
+
"epoch": 1.45,
|
912 |
+
"learning_rate": 3.547238791610545e-05,
|
913 |
+
"loss": 0.9284,
|
914 |
+
"step": 7550
|
915 |
+
},
|
916 |
+
{
|
917 |
+
"epoch": 1.46,
|
918 |
+
"learning_rate": 3.537617856455648e-05,
|
919 |
+
"loss": 0.9946,
|
920 |
+
"step": 7600
|
921 |
+
},
|
922 |
+
{
|
923 |
+
"epoch": 1.47,
|
924 |
+
"learning_rate": 3.5279969213007505e-05,
|
925 |
+
"loss": 0.9267,
|
926 |
+
"step": 7650
|
927 |
+
},
|
928 |
+
{
|
929 |
+
"epoch": 1.48,
|
930 |
+
"learning_rate": 3.5183759861458536e-05,
|
931 |
+
"loss": 0.9609,
|
932 |
+
"step": 7700
|
933 |
+
},
|
934 |
+
{
|
935 |
+
"epoch": 1.49,
|
936 |
+
"learning_rate": 3.508755050990957e-05,
|
937 |
+
"loss": 1.0015,
|
938 |
+
"step": 7750
|
939 |
+
},
|
940 |
+
{
|
941 |
+
"epoch": 1.5,
|
942 |
+
"learning_rate": 3.49913411583606e-05,
|
943 |
+
"loss": 0.9598,
|
944 |
+
"step": 7800
|
945 |
+
},
|
946 |
+
{
|
947 |
+
"epoch": 1.51,
|
948 |
+
"learning_rate": 3.4895131806811624e-05,
|
949 |
+
"loss": 1.0063,
|
950 |
+
"step": 7850
|
951 |
+
},
|
952 |
+
{
|
953 |
+
"epoch": 1.52,
|
954 |
+
"learning_rate": 3.4798922455262655e-05,
|
955 |
+
"loss": 0.9482,
|
956 |
+
"step": 7900
|
957 |
+
},
|
958 |
+
{
|
959 |
+
"epoch": 1.53,
|
960 |
+
"learning_rate": 3.470271310371368e-05,
|
961 |
+
"loss": 0.9251,
|
962 |
+
"step": 7950
|
963 |
+
},
|
964 |
+
{
|
965 |
+
"epoch": 1.54,
|
966 |
+
"learning_rate": 3.460650375216471e-05,
|
967 |
+
"loss": 1.0041,
|
968 |
+
"step": 8000
|
969 |
+
},
|
970 |
+
{
|
971 |
+
"epoch": 1.55,
|
972 |
+
"learning_rate": 3.4510294400615736e-05,
|
973 |
+
"loss": 0.9382,
|
974 |
+
"step": 8050
|
975 |
+
},
|
976 |
+
{
|
977 |
+
"epoch": 1.56,
|
978 |
+
"learning_rate": 3.441408504906677e-05,
|
979 |
+
"loss": 0.9844,
|
980 |
+
"step": 8100
|
981 |
+
},
|
982 |
+
{
|
983 |
+
"epoch": 1.57,
|
984 |
+
"learning_rate": 3.43178756975178e-05,
|
985 |
+
"loss": 0.9732,
|
986 |
+
"step": 8150
|
987 |
+
},
|
988 |
+
{
|
989 |
+
"epoch": 1.58,
|
990 |
+
"learning_rate": 3.422166634596883e-05,
|
991 |
+
"loss": 0.9647,
|
992 |
+
"step": 8200
|
993 |
+
},
|
994 |
+
{
|
995 |
+
"epoch": 1.59,
|
996 |
+
"learning_rate": 3.4125456994419855e-05,
|
997 |
+
"loss": 0.9859,
|
998 |
+
"step": 8250
|
999 |
+
},
|
1000 |
+
{
|
1001 |
+
"epoch": 1.6,
|
1002 |
+
"learning_rate": 3.4029247642870886e-05,
|
1003 |
+
"loss": 0.9316,
|
1004 |
+
"step": 8300
|
1005 |
+
},
|
1006 |
+
{
|
1007 |
+
"epoch": 1.61,
|
1008 |
+
"learning_rate": 3.393303829132192e-05,
|
1009 |
+
"loss": 0.9852,
|
1010 |
+
"step": 8350
|
1011 |
+
},
|
1012 |
+
{
|
1013 |
+
"epoch": 1.62,
|
1014 |
+
"learning_rate": 3.383682893977295e-05,
|
1015 |
+
"loss": 0.9839,
|
1016 |
+
"step": 8400
|
1017 |
+
},
|
1018 |
+
{
|
1019 |
+
"epoch": 1.63,
|
1020 |
+
"learning_rate": 3.3740619588223974e-05,
|
1021 |
+
"loss": 0.958,
|
1022 |
+
"step": 8450
|
1023 |
+
},
|
1024 |
+
{
|
1025 |
+
"epoch": 1.64,
|
1026 |
+
"learning_rate": 3.3644410236675005e-05,
|
1027 |
+
"loss": 0.9693,
|
1028 |
+
"step": 8500
|
1029 |
+
},
|
1030 |
+
{
|
1031 |
+
"epoch": 1.65,
|
1032 |
+
"learning_rate": 3.354820088512604e-05,
|
1033 |
+
"loss": 0.99,
|
1034 |
+
"step": 8550
|
1035 |
+
},
|
1036 |
+
{
|
1037 |
+
"epoch": 1.65,
|
1038 |
+
"learning_rate": 3.345199153357707e-05,
|
1039 |
+
"loss": 0.9832,
|
1040 |
+
"step": 8600
|
1041 |
+
},
|
1042 |
+
{
|
1043 |
+
"epoch": 1.66,
|
1044 |
+
"learning_rate": 3.335578218202809e-05,
|
1045 |
+
"loss": 0.9417,
|
1046 |
+
"step": 8650
|
1047 |
+
},
|
1048 |
+
{
|
1049 |
+
"epoch": 1.67,
|
1050 |
+
"learning_rate": 3.3259572830479124e-05,
|
1051 |
+
"loss": 0.9372,
|
1052 |
+
"step": 8700
|
1053 |
+
},
|
1054 |
+
{
|
1055 |
+
"epoch": 1.68,
|
1056 |
+
"learning_rate": 3.3163363478930156e-05,
|
1057 |
+
"loss": 0.9493,
|
1058 |
+
"step": 8750
|
1059 |
+
},
|
1060 |
+
{
|
1061 |
+
"epoch": 1.69,
|
1062 |
+
"learning_rate": 3.306715412738119e-05,
|
1063 |
+
"loss": 0.9727,
|
1064 |
+
"step": 8800
|
1065 |
+
},
|
1066 |
+
{
|
1067 |
+
"epoch": 1.7,
|
1068 |
+
"learning_rate": 3.297094477583221e-05,
|
1069 |
+
"loss": 0.9129,
|
1070 |
+
"step": 8850
|
1071 |
+
},
|
1072 |
+
{
|
1073 |
+
"epoch": 1.71,
|
1074 |
+
"learning_rate": 3.287473542428324e-05,
|
1075 |
+
"loss": 0.9653,
|
1076 |
+
"step": 8900
|
1077 |
+
},
|
1078 |
+
{
|
1079 |
+
"epoch": 1.72,
|
1080 |
+
"learning_rate": 3.2778526072734274e-05,
|
1081 |
+
"loss": 0.9745,
|
1082 |
+
"step": 8950
|
1083 |
+
},
|
1084 |
+
{
|
1085 |
+
"epoch": 1.73,
|
1086 |
+
"learning_rate": 3.26823167211853e-05,
|
1087 |
+
"loss": 0.9281,
|
1088 |
+
"step": 9000
|
1089 |
+
},
|
1090 |
+
{
|
1091 |
+
"epoch": 1.74,
|
1092 |
+
"learning_rate": 3.258610736963633e-05,
|
1093 |
+
"loss": 0.9438,
|
1094 |
+
"step": 9050
|
1095 |
+
},
|
1096 |
+
{
|
1097 |
+
"epoch": 1.75,
|
1098 |
+
"learning_rate": 3.248989801808736e-05,
|
1099 |
+
"loss": 0.9526,
|
1100 |
+
"step": 9100
|
1101 |
+
},
|
1102 |
+
{
|
1103 |
+
"epoch": 1.76,
|
1104 |
+
"learning_rate": 3.239368866653839e-05,
|
1105 |
+
"loss": 0.9878,
|
1106 |
+
"step": 9150
|
1107 |
+
},
|
1108 |
+
{
|
1109 |
+
"epoch": 1.77,
|
1110 |
+
"learning_rate": 3.229747931498942e-05,
|
1111 |
+
"loss": 0.9627,
|
1112 |
+
"step": 9200
|
1113 |
+
},
|
1114 |
+
{
|
1115 |
+
"epoch": 1.78,
|
1116 |
+
"learning_rate": 3.220126996344045e-05,
|
1117 |
+
"loss": 0.9809,
|
1118 |
+
"step": 9250
|
1119 |
+
},
|
1120 |
+
{
|
1121 |
+
"epoch": 1.79,
|
1122 |
+
"learning_rate": 3.210506061189148e-05,
|
1123 |
+
"loss": 0.9427,
|
1124 |
+
"step": 9300
|
1125 |
+
},
|
1126 |
+
{
|
1127 |
+
"epoch": 1.8,
|
1128 |
+
"learning_rate": 3.200885126034251e-05,
|
1129 |
+
"loss": 0.907,
|
1130 |
+
"step": 9350
|
1131 |
+
},
|
1132 |
+
{
|
1133 |
+
"epoch": 1.81,
|
1134 |
+
"learning_rate": 3.191264190879354e-05,
|
1135 |
+
"loss": 0.9627,
|
1136 |
+
"step": 9400
|
1137 |
+
},
|
1138 |
+
{
|
1139 |
+
"epoch": 1.82,
|
1140 |
+
"learning_rate": 3.181643255724456e-05,
|
1141 |
+
"loss": 0.9693,
|
1142 |
+
"step": 9450
|
1143 |
+
},
|
1144 |
+
{
|
1145 |
+
"epoch": 1.83,
|
1146 |
+
"learning_rate": 3.172022320569559e-05,
|
1147 |
+
"loss": 0.9561,
|
1148 |
+
"step": 9500
|
1149 |
+
},
|
1150 |
+
{
|
1151 |
+
"epoch": 1.84,
|
1152 |
+
"learning_rate": 3.1624013854146624e-05,
|
1153 |
+
"loss": 0.9346,
|
1154 |
+
"step": 9550
|
1155 |
+
},
|
1156 |
+
{
|
1157 |
+
"epoch": 1.85,
|
1158 |
+
"learning_rate": 3.152780450259765e-05,
|
1159 |
+
"loss": 0.9597,
|
1160 |
+
"step": 9600
|
1161 |
+
},
|
1162 |
+
{
|
1163 |
+
"epoch": 1.86,
|
1164 |
+
"learning_rate": 3.143159515104868e-05,
|
1165 |
+
"loss": 0.9479,
|
1166 |
+
"step": 9650
|
1167 |
+
},
|
1168 |
+
{
|
1169 |
+
"epoch": 1.87,
|
1170 |
+
"learning_rate": 3.133538579949971e-05,
|
1171 |
+
"loss": 0.9935,
|
1172 |
+
"step": 9700
|
1173 |
+
},
|
1174 |
+
{
|
1175 |
+
"epoch": 1.88,
|
1176 |
+
"learning_rate": 3.123917644795074e-05,
|
1177 |
+
"loss": 0.9036,
|
1178 |
+
"step": 9750
|
1179 |
+
},
|
1180 |
+
{
|
1181 |
+
"epoch": 1.89,
|
1182 |
+
"learning_rate": 3.114296709640177e-05,
|
1183 |
+
"loss": 0.9218,
|
1184 |
+
"step": 9800
|
1185 |
+
},
|
1186 |
+
{
|
1187 |
+
"epoch": 1.9,
|
1188 |
+
"learning_rate": 3.10467577448528e-05,
|
1189 |
+
"loss": 0.9123,
|
1190 |
+
"step": 9850
|
1191 |
+
},
|
1192 |
+
{
|
1193 |
+
"epoch": 1.9,
|
1194 |
+
"learning_rate": 3.095054839330383e-05,
|
1195 |
+
"loss": 0.9674,
|
1196 |
+
"step": 9900
|
1197 |
+
},
|
1198 |
+
{
|
1199 |
+
"epoch": 1.91,
|
1200 |
+
"learning_rate": 3.085433904175486e-05,
|
1201 |
+
"loss": 0.9119,
|
1202 |
+
"step": 9950
|
1203 |
+
},
|
1204 |
+
{
|
1205 |
+
"epoch": 1.92,
|
1206 |
+
"learning_rate": 3.075812969020589e-05,
|
1207 |
+
"loss": 0.931,
|
1208 |
+
"step": 10000
|
1209 |
+
},
|
1210 |
+
{
|
1211 |
+
"epoch": 1.93,
|
1212 |
+
"learning_rate": 3.066192033865692e-05,
|
1213 |
+
"loss": 0.9491,
|
1214 |
+
"step": 10050
|
1215 |
+
},
|
1216 |
+
{
|
1217 |
+
"epoch": 1.94,
|
1218 |
+
"learning_rate": 3.056571098710795e-05,
|
1219 |
+
"loss": 0.9285,
|
1220 |
+
"step": 10100
|
1221 |
+
},
|
1222 |
+
{
|
1223 |
+
"epoch": 1.95,
|
1224 |
+
"learning_rate": 3.0469501635558978e-05,
|
1225 |
+
"loss": 0.9483,
|
1226 |
+
"step": 10150
|
1227 |
+
},
|
1228 |
+
{
|
1229 |
+
"epoch": 1.96,
|
1230 |
+
"learning_rate": 3.037329228401001e-05,
|
1231 |
+
"loss": 0.9874,
|
1232 |
+
"step": 10200
|
1233 |
+
},
|
1234 |
+
{
|
1235 |
+
"epoch": 1.97,
|
1236 |
+
"learning_rate": 3.0277082932461037e-05,
|
1237 |
+
"loss": 0.9781,
|
1238 |
+
"step": 10250
|
1239 |
+
},
|
1240 |
+
{
|
1241 |
+
"epoch": 1.98,
|
1242 |
+
"learning_rate": 3.0180873580912065e-05,
|
1243 |
+
"loss": 0.9851,
|
1244 |
+
"step": 10300
|
1245 |
+
},
|
1246 |
+
{
|
1247 |
+
"epoch": 1.99,
|
1248 |
+
"learning_rate": 3.0084664229363097e-05,
|
1249 |
+
"loss": 0.9219,
|
1250 |
+
"step": 10350
|
1251 |
+
},
|
1252 |
+
{
|
1253 |
+
"epoch": 2.0,
|
1254 |
+
"learning_rate": 2.9988454877814125e-05,
|
1255 |
+
"loss": 0.9572,
|
1256 |
+
"step": 10400
|
1257 |
+
},
|
1258 |
+
{
|
1259 |
+
"epoch": 2.01,
|
1260 |
+
"learning_rate": 2.9892245526265156e-05,
|
1261 |
+
"loss": 0.9187,
|
1262 |
+
"step": 10450
|
1263 |
+
},
|
1264 |
+
{
|
1265 |
+
"epoch": 2.02,
|
1266 |
+
"learning_rate": 2.9796036174716184e-05,
|
1267 |
+
"loss": 0.905,
|
1268 |
+
"step": 10500
|
1269 |
+
},
|
1270 |
+
{
|
1271 |
+
"epoch": 2.03,
|
1272 |
+
"learning_rate": 2.9699826823167216e-05,
|
1273 |
+
"loss": 0.9498,
|
1274 |
+
"step": 10550
|
1275 |
+
},
|
1276 |
+
{
|
1277 |
+
"epoch": 2.04,
|
1278 |
+
"learning_rate": 2.9603617471618244e-05,
|
1279 |
+
"loss": 0.9586,
|
1280 |
+
"step": 10600
|
1281 |
+
},
|
1282 |
+
{
|
1283 |
+
"epoch": 2.05,
|
1284 |
+
"learning_rate": 2.9507408120069275e-05,
|
1285 |
+
"loss": 0.9464,
|
1286 |
+
"step": 10650
|
1287 |
+
},
|
1288 |
+
{
|
1289 |
+
"epoch": 2.06,
|
1290 |
+
"learning_rate": 2.9411198768520303e-05,
|
1291 |
+
"loss": 0.9205,
|
1292 |
+
"step": 10700
|
1293 |
+
},
|
1294 |
+
{
|
1295 |
+
"epoch": 2.07,
|
1296 |
+
"learning_rate": 2.9314989416971334e-05,
|
1297 |
+
"loss": 0.9331,
|
1298 |
+
"step": 10750
|
1299 |
+
},
|
1300 |
+
{
|
1301 |
+
"epoch": 2.08,
|
1302 |
+
"learning_rate": 2.9218780065422362e-05,
|
1303 |
+
"loss": 0.943,
|
1304 |
+
"step": 10800
|
1305 |
+
},
|
1306 |
+
{
|
1307 |
+
"epoch": 2.09,
|
1308 |
+
"learning_rate": 2.9122570713873387e-05,
|
1309 |
+
"loss": 0.9544,
|
1310 |
+
"step": 10850
|
1311 |
+
},
|
1312 |
+
{
|
1313 |
+
"epoch": 2.1,
|
1314 |
+
"learning_rate": 2.9026361362324415e-05,
|
1315 |
+
"loss": 0.9148,
|
1316 |
+
"step": 10900
|
1317 |
+
},
|
1318 |
+
{
|
1319 |
+
"epoch": 2.11,
|
1320 |
+
"learning_rate": 2.8930152010775447e-05,
|
1321 |
+
"loss": 0.9569,
|
1322 |
+
"step": 10950
|
1323 |
+
},
|
1324 |
+
{
|
1325 |
+
"epoch": 2.12,
|
1326 |
+
"learning_rate": 2.8833942659226475e-05,
|
1327 |
+
"loss": 0.8757,
|
1328 |
+
"step": 11000
|
1329 |
+
},
|
1330 |
+
{
|
1331 |
+
"epoch": 2.13,
|
1332 |
+
"learning_rate": 2.8737733307677506e-05,
|
1333 |
+
"loss": 0.9618,
|
1334 |
+
"step": 11050
|
1335 |
+
},
|
1336 |
+
{
|
1337 |
+
"epoch": 2.14,
|
1338 |
+
"learning_rate": 2.8641523956128534e-05,
|
1339 |
+
"loss": 0.9181,
|
1340 |
+
"step": 11100
|
1341 |
+
},
|
1342 |
+
{
|
1343 |
+
"epoch": 2.15,
|
1344 |
+
"learning_rate": 2.8545314604579565e-05,
|
1345 |
+
"loss": 0.9251,
|
1346 |
+
"step": 11150
|
1347 |
+
},
|
1348 |
+
{
|
1349 |
+
"epoch": 2.16,
|
1350 |
+
"learning_rate": 2.8449105253030594e-05,
|
1351 |
+
"loss": 0.9052,
|
1352 |
+
"step": 11200
|
1353 |
+
},
|
1354 |
+
{
|
1355 |
+
"epoch": 2.16,
|
1356 |
+
"learning_rate": 2.8352895901481625e-05,
|
1357 |
+
"loss": 0.9723,
|
1358 |
+
"step": 11250
|
1359 |
+
},
|
1360 |
+
{
|
1361 |
+
"epoch": 2.17,
|
1362 |
+
"learning_rate": 2.8256686549932653e-05,
|
1363 |
+
"loss": 0.9938,
|
1364 |
+
"step": 11300
|
1365 |
+
},
|
1366 |
+
{
|
1367 |
+
"epoch": 2.18,
|
1368 |
+
"learning_rate": 2.8160477198383684e-05,
|
1369 |
+
"loss": 0.9029,
|
1370 |
+
"step": 11350
|
1371 |
+
},
|
1372 |
+
{
|
1373 |
+
"epoch": 2.19,
|
1374 |
+
"learning_rate": 2.8064267846834712e-05,
|
1375 |
+
"loss": 0.952,
|
1376 |
+
"step": 11400
|
1377 |
+
},
|
1378 |
+
{
|
1379 |
+
"epoch": 2.2,
|
1380 |
+
"learning_rate": 2.7968058495285744e-05,
|
1381 |
+
"loss": 0.9147,
|
1382 |
+
"step": 11450
|
1383 |
+
},
|
1384 |
+
{
|
1385 |
+
"epoch": 2.21,
|
1386 |
+
"learning_rate": 2.7871849143736772e-05,
|
1387 |
+
"loss": 0.9595,
|
1388 |
+
"step": 11500
|
1389 |
+
},
|
1390 |
+
{
|
1391 |
+
"epoch": 2.22,
|
1392 |
+
"learning_rate": 2.7775639792187803e-05,
|
1393 |
+
"loss": 0.8966,
|
1394 |
+
"step": 11550
|
1395 |
+
},
|
1396 |
+
{
|
1397 |
+
"epoch": 2.23,
|
1398 |
+
"learning_rate": 2.767943044063883e-05,
|
1399 |
+
"loss": 0.9401,
|
1400 |
+
"step": 11600
|
1401 |
+
},
|
1402 |
+
{
|
1403 |
+
"epoch": 2.24,
|
1404 |
+
"learning_rate": 2.7583221089089863e-05,
|
1405 |
+
"loss": 0.9376,
|
1406 |
+
"step": 11650
|
1407 |
+
},
|
1408 |
+
{
|
1409 |
+
"epoch": 2.25,
|
1410 |
+
"learning_rate": 2.748701173754089e-05,
|
1411 |
+
"loss": 0.901,
|
1412 |
+
"step": 11700
|
1413 |
+
},
|
1414 |
+
{
|
1415 |
+
"epoch": 2.26,
|
1416 |
+
"learning_rate": 2.739080238599192e-05,
|
1417 |
+
"loss": 0.9362,
|
1418 |
+
"step": 11750
|
1419 |
+
},
|
1420 |
+
{
|
1421 |
+
"epoch": 2.27,
|
1422 |
+
"learning_rate": 2.729459303444295e-05,
|
1423 |
+
"loss": 0.9836,
|
1424 |
+
"step": 11800
|
1425 |
+
},
|
1426 |
+
{
|
1427 |
+
"epoch": 2.28,
|
1428 |
+
"learning_rate": 2.7198383682893978e-05,
|
1429 |
+
"loss": 0.963,
|
1430 |
+
"step": 11850
|
1431 |
+
},
|
1432 |
+
{
|
1433 |
+
"epoch": 2.29,
|
1434 |
+
"learning_rate": 2.710217433134501e-05,
|
1435 |
+
"loss": 0.8923,
|
1436 |
+
"step": 11900
|
1437 |
+
},
|
1438 |
+
{
|
1439 |
+
"epoch": 2.3,
|
1440 |
+
"learning_rate": 2.7005964979796038e-05,
|
1441 |
+
"loss": 0.9313,
|
1442 |
+
"step": 11950
|
1443 |
+
},
|
1444 |
+
{
|
1445 |
+
"epoch": 2.31,
|
1446 |
+
"learning_rate": 2.690975562824707e-05,
|
1447 |
+
"loss": 0.9261,
|
1448 |
+
"step": 12000
|
1449 |
+
},
|
1450 |
+
{
|
1451 |
+
"epoch": 2.32,
|
1452 |
+
"learning_rate": 2.6813546276698097e-05,
|
1453 |
+
"loss": 0.9517,
|
1454 |
+
"step": 12050
|
1455 |
+
},
|
1456 |
+
{
|
1457 |
+
"epoch": 2.33,
|
1458 |
+
"learning_rate": 2.671733692514913e-05,
|
1459 |
+
"loss": 0.9302,
|
1460 |
+
"step": 12100
|
1461 |
+
},
|
1462 |
+
{
|
1463 |
+
"epoch": 2.34,
|
1464 |
+
"learning_rate": 2.6621127573600157e-05,
|
1465 |
+
"loss": 0.9249,
|
1466 |
+
"step": 12150
|
1467 |
+
},
|
1468 |
+
{
|
1469 |
+
"epoch": 2.35,
|
1470 |
+
"learning_rate": 2.6524918222051188e-05,
|
1471 |
+
"loss": 0.8492,
|
1472 |
+
"step": 12200
|
1473 |
+
},
|
1474 |
+
{
|
1475 |
+
"epoch": 2.36,
|
1476 |
+
"learning_rate": 2.6428708870502216e-05,
|
1477 |
+
"loss": 0.9433,
|
1478 |
+
"step": 12250
|
1479 |
+
},
|
1480 |
+
{
|
1481 |
+
"epoch": 2.37,
|
1482 |
+
"learning_rate": 2.633249951895324e-05,
|
1483 |
+
"loss": 0.9526,
|
1484 |
+
"step": 12300
|
1485 |
+
},
|
1486 |
+
{
|
1487 |
+
"epoch": 2.38,
|
1488 |
+
"learning_rate": 2.6236290167404272e-05,
|
1489 |
+
"loss": 0.9241,
|
1490 |
+
"step": 12350
|
1491 |
+
},
|
1492 |
+
{
|
1493 |
+
"epoch": 2.39,
|
1494 |
+
"learning_rate": 2.61400808158553e-05,
|
1495 |
+
"loss": 0.9073,
|
1496 |
+
"step": 12400
|
1497 |
+
},
|
1498 |
+
{
|
1499 |
+
"epoch": 2.4,
|
1500 |
+
"learning_rate": 2.6043871464306328e-05,
|
1501 |
+
"loss": 0.9214,
|
1502 |
+
"step": 12450
|
1503 |
+
},
|
1504 |
+
{
|
1505 |
+
"epoch": 2.41,
|
1506 |
+
"learning_rate": 2.594766211275736e-05,
|
1507 |
+
"loss": 0.9242,
|
1508 |
+
"step": 12500
|
1509 |
+
},
|
1510 |
+
{
|
1511 |
+
"epoch": 2.41,
|
1512 |
+
"learning_rate": 2.5851452761208388e-05,
|
1513 |
+
"loss": 0.9421,
|
1514 |
+
"step": 12550
|
1515 |
+
},
|
1516 |
+
{
|
1517 |
+
"epoch": 2.42,
|
1518 |
+
"learning_rate": 2.575524340965942e-05,
|
1519 |
+
"loss": 0.9111,
|
1520 |
+
"step": 12600
|
1521 |
+
},
|
1522 |
+
{
|
1523 |
+
"epoch": 2.43,
|
1524 |
+
"learning_rate": 2.5659034058110447e-05,
|
1525 |
+
"loss": 0.9089,
|
1526 |
+
"step": 12650
|
1527 |
+
},
|
1528 |
+
{
|
1529 |
+
"epoch": 2.44,
|
1530 |
+
"learning_rate": 2.556282470656148e-05,
|
1531 |
+
"loss": 0.8959,
|
1532 |
+
"step": 12700
|
1533 |
+
},
|
1534 |
+
{
|
1535 |
+
"epoch": 2.45,
|
1536 |
+
"learning_rate": 2.5466615355012507e-05,
|
1537 |
+
"loss": 0.8655,
|
1538 |
+
"step": 12750
|
1539 |
+
},
|
1540 |
+
{
|
1541 |
+
"epoch": 2.46,
|
1542 |
+
"learning_rate": 2.5370406003463538e-05,
|
1543 |
+
"loss": 0.9657,
|
1544 |
+
"step": 12800
|
1545 |
+
},
|
1546 |
+
{
|
1547 |
+
"epoch": 2.47,
|
1548 |
+
"learning_rate": 2.5274196651914566e-05,
|
1549 |
+
"loss": 0.9018,
|
1550 |
+
"step": 12850
|
1551 |
+
},
|
1552 |
+
{
|
1553 |
+
"epoch": 2.48,
|
1554 |
+
"learning_rate": 2.5177987300365597e-05,
|
1555 |
+
"loss": 0.9141,
|
1556 |
+
"step": 12900
|
1557 |
+
},
|
1558 |
+
{
|
1559 |
+
"epoch": 2.49,
|
1560 |
+
"learning_rate": 2.5081777948816626e-05,
|
1561 |
+
"loss": 0.9265,
|
1562 |
+
"step": 12950
|
1563 |
+
},
|
1564 |
+
{
|
1565 |
+
"epoch": 2.5,
|
1566 |
+
"learning_rate": 2.4985568597267657e-05,
|
1567 |
+
"loss": 0.9464,
|
1568 |
+
"step": 13000
|
1569 |
+
},
|
1570 |
+
{
|
1571 |
+
"epoch": 2.51,
|
1572 |
+
"learning_rate": 2.4889359245718685e-05,
|
1573 |
+
"loss": 0.9249,
|
1574 |
+
"step": 13050
|
1575 |
+
},
|
1576 |
+
{
|
1577 |
+
"epoch": 2.52,
|
1578 |
+
"learning_rate": 2.4793149894169716e-05,
|
1579 |
+
"loss": 0.9138,
|
1580 |
+
"step": 13100
|
1581 |
+
},
|
1582 |
+
{
|
1583 |
+
"epoch": 2.53,
|
1584 |
+
"learning_rate": 2.4696940542620744e-05,
|
1585 |
+
"loss": 0.9607,
|
1586 |
+
"step": 13150
|
1587 |
+
},
|
1588 |
+
{
|
1589 |
+
"epoch": 2.54,
|
1590 |
+
"learning_rate": 2.4600731191071776e-05,
|
1591 |
+
"loss": 0.9267,
|
1592 |
+
"step": 13200
|
1593 |
+
},
|
1594 |
+
{
|
1595 |
+
"epoch": 2.55,
|
1596 |
+
"learning_rate": 2.4504521839522804e-05,
|
1597 |
+
"loss": 0.945,
|
1598 |
+
"step": 13250
|
1599 |
+
},
|
1600 |
+
{
|
1601 |
+
"epoch": 2.56,
|
1602 |
+
"learning_rate": 2.4408312487973832e-05,
|
1603 |
+
"loss": 0.9202,
|
1604 |
+
"step": 13300
|
1605 |
+
},
|
1606 |
+
{
|
1607 |
+
"epoch": 2.57,
|
1608 |
+
"learning_rate": 2.4312103136424863e-05,
|
1609 |
+
"loss": 0.9134,
|
1610 |
+
"step": 13350
|
1611 |
+
},
|
1612 |
+
{
|
1613 |
+
"epoch": 2.58,
|
1614 |
+
"learning_rate": 2.421589378487589e-05,
|
1615 |
+
"loss": 0.9387,
|
1616 |
+
"step": 13400
|
1617 |
+
},
|
1618 |
+
{
|
1619 |
+
"epoch": 2.59,
|
1620 |
+
"learning_rate": 2.411968443332692e-05,
|
1621 |
+
"loss": 0.9385,
|
1622 |
+
"step": 13450
|
1623 |
+
},
|
1624 |
+
{
|
1625 |
+
"epoch": 2.6,
|
1626 |
+
"learning_rate": 2.402347508177795e-05,
|
1627 |
+
"loss": 0.9397,
|
1628 |
+
"step": 13500
|
1629 |
+
},
|
1630 |
+
{
|
1631 |
+
"epoch": 2.61,
|
1632 |
+
"learning_rate": 2.392726573022898e-05,
|
1633 |
+
"loss": 0.9437,
|
1634 |
+
"step": 13550
|
1635 |
+
},
|
1636 |
+
{
|
1637 |
+
"epoch": 2.62,
|
1638 |
+
"learning_rate": 2.3831056378680007e-05,
|
1639 |
+
"loss": 0.9482,
|
1640 |
+
"step": 13600
|
1641 |
+
},
|
1642 |
+
{
|
1643 |
+
"epoch": 2.63,
|
1644 |
+
"learning_rate": 2.3734847027131038e-05,
|
1645 |
+
"loss": 0.8986,
|
1646 |
+
"step": 13650
|
1647 |
+
},
|
1648 |
+
{
|
1649 |
+
"epoch": 2.64,
|
1650 |
+
"learning_rate": 2.3638637675582066e-05,
|
1651 |
+
"loss": 0.8702,
|
1652 |
+
"step": 13700
|
1653 |
+
},
|
1654 |
+
{
|
1655 |
+
"epoch": 2.65,
|
1656 |
+
"learning_rate": 2.3542428324033098e-05,
|
1657 |
+
"loss": 0.9375,
|
1658 |
+
"step": 13750
|
1659 |
+
},
|
1660 |
+
{
|
1661 |
+
"epoch": 2.66,
|
1662 |
+
"learning_rate": 2.3446218972484126e-05,
|
1663 |
+
"loss": 0.9304,
|
1664 |
+
"step": 13800
|
1665 |
+
},
|
1666 |
+
{
|
1667 |
+
"epoch": 2.66,
|
1668 |
+
"learning_rate": 2.3350009620935157e-05,
|
1669 |
+
"loss": 0.9104,
|
1670 |
+
"step": 13850
|
1671 |
+
},
|
1672 |
+
{
|
1673 |
+
"epoch": 2.67,
|
1674 |
+
"learning_rate": 2.3253800269386185e-05,
|
1675 |
+
"loss": 0.9192,
|
1676 |
+
"step": 13900
|
1677 |
+
},
|
1678 |
+
{
|
1679 |
+
"epoch": 2.68,
|
1680 |
+
"learning_rate": 2.3157590917837217e-05,
|
1681 |
+
"loss": 0.9053,
|
1682 |
+
"step": 13950
|
1683 |
+
},
|
1684 |
+
{
|
1685 |
+
"epoch": 2.69,
|
1686 |
+
"learning_rate": 2.3061381566288245e-05,
|
1687 |
+
"loss": 0.891,
|
1688 |
+
"step": 14000
|
1689 |
+
},
|
1690 |
+
{
|
1691 |
+
"epoch": 2.7,
|
1692 |
+
"learning_rate": 2.2965172214739276e-05,
|
1693 |
+
"loss": 0.89,
|
1694 |
+
"step": 14050
|
1695 |
+
},
|
1696 |
+
{
|
1697 |
+
"epoch": 2.71,
|
1698 |
+
"learning_rate": 2.28689628631903e-05,
|
1699 |
+
"loss": 0.9109,
|
1700 |
+
"step": 14100
|
1701 |
+
},
|
1702 |
+
{
|
1703 |
+
"epoch": 2.72,
|
1704 |
+
"learning_rate": 2.2772753511641332e-05,
|
1705 |
+
"loss": 0.9166,
|
1706 |
+
"step": 14150
|
1707 |
+
},
|
1708 |
+
{
|
1709 |
+
"epoch": 2.73,
|
1710 |
+
"learning_rate": 2.267654416009236e-05,
|
1711 |
+
"loss": 0.8821,
|
1712 |
+
"step": 14200
|
1713 |
+
},
|
1714 |
+
{
|
1715 |
+
"epoch": 2.74,
|
1716 |
+
"learning_rate": 2.258033480854339e-05,
|
1717 |
+
"loss": 0.9186,
|
1718 |
+
"step": 14250
|
1719 |
+
},
|
1720 |
+
{
|
1721 |
+
"epoch": 2.75,
|
1722 |
+
"learning_rate": 2.248412545699442e-05,
|
1723 |
+
"loss": 0.9296,
|
1724 |
+
"step": 14300
|
1725 |
+
},
|
1726 |
+
{
|
1727 |
+
"epoch": 2.76,
|
1728 |
+
"learning_rate": 2.238791610544545e-05,
|
1729 |
+
"loss": 0.8647,
|
1730 |
+
"step": 14350
|
1731 |
+
},
|
1732 |
+
{
|
1733 |
+
"epoch": 2.77,
|
1734 |
+
"learning_rate": 2.229170675389648e-05,
|
1735 |
+
"loss": 0.9332,
|
1736 |
+
"step": 14400
|
1737 |
+
},
|
1738 |
+
{
|
1739 |
+
"epoch": 2.78,
|
1740 |
+
"learning_rate": 2.219549740234751e-05,
|
1741 |
+
"loss": 0.9078,
|
1742 |
+
"step": 14450
|
1743 |
+
},
|
1744 |
+
{
|
1745 |
+
"epoch": 2.79,
|
1746 |
+
"learning_rate": 2.209928805079854e-05,
|
1747 |
+
"loss": 0.9714,
|
1748 |
+
"step": 14500
|
1749 |
+
},
|
1750 |
+
{
|
1751 |
+
"epoch": 2.8,
|
1752 |
+
"learning_rate": 2.200307869924957e-05,
|
1753 |
+
"loss": 0.891,
|
1754 |
+
"step": 14550
|
1755 |
+
},
|
1756 |
+
{
|
1757 |
+
"epoch": 2.81,
|
1758 |
+
"learning_rate": 2.1906869347700598e-05,
|
1759 |
+
"loss": 0.9242,
|
1760 |
+
"step": 14600
|
1761 |
+
},
|
1762 |
+
{
|
1763 |
+
"epoch": 2.82,
|
1764 |
+
"learning_rate": 2.181065999615163e-05,
|
1765 |
+
"loss": 0.8897,
|
1766 |
+
"step": 14650
|
1767 |
+
},
|
1768 |
+
{
|
1769 |
+
"epoch": 2.83,
|
1770 |
+
"learning_rate": 2.1714450644602657e-05,
|
1771 |
+
"loss": 0.9339,
|
1772 |
+
"step": 14700
|
1773 |
+
},
|
1774 |
+
{
|
1775 |
+
"epoch": 2.84,
|
1776 |
+
"learning_rate": 2.1618241293053686e-05,
|
1777 |
+
"loss": 0.9459,
|
1778 |
+
"step": 14750
|
1779 |
+
},
|
1780 |
+
{
|
1781 |
+
"epoch": 2.85,
|
1782 |
+
"learning_rate": 2.1522031941504714e-05,
|
1783 |
+
"loss": 0.9447,
|
1784 |
+
"step": 14800
|
1785 |
+
},
|
1786 |
+
{
|
1787 |
+
"epoch": 2.86,
|
1788 |
+
"learning_rate": 2.1425822589955745e-05,
|
1789 |
+
"loss": 0.8597,
|
1790 |
+
"step": 14850
|
1791 |
+
},
|
1792 |
+
{
|
1793 |
+
"epoch": 2.87,
|
1794 |
+
"learning_rate": 2.1329613238406773e-05,
|
1795 |
+
"loss": 0.9117,
|
1796 |
+
"step": 14900
|
1797 |
+
},
|
1798 |
+
{
|
1799 |
+
"epoch": 2.88,
|
1800 |
+
"learning_rate": 2.1233403886857804e-05,
|
1801 |
+
"loss": 0.9621,
|
1802 |
+
"step": 14950
|
1803 |
+
},
|
1804 |
+
{
|
1805 |
+
"epoch": 2.89,
|
1806 |
+
"learning_rate": 2.1137194535308832e-05,
|
1807 |
+
"loss": 0.8957,
|
1808 |
+
"step": 15000
|
1809 |
+
},
|
1810 |
+
{
|
1811 |
+
"epoch": 2.9,
|
1812 |
+
"learning_rate": 2.104098518375986e-05,
|
1813 |
+
"loss": 0.9096,
|
1814 |
+
"step": 15050
|
1815 |
+
},
|
1816 |
+
{
|
1817 |
+
"epoch": 2.91,
|
1818 |
+
"learning_rate": 2.0944775832210892e-05,
|
1819 |
+
"loss": 0.8775,
|
1820 |
+
"step": 15100
|
1821 |
+
},
|
1822 |
+
{
|
1823 |
+
"epoch": 2.92,
|
1824 |
+
"learning_rate": 2.084856648066192e-05,
|
1825 |
+
"loss": 0.9468,
|
1826 |
+
"step": 15150
|
1827 |
+
},
|
1828 |
+
{
|
1829 |
+
"epoch": 2.92,
|
1830 |
+
"learning_rate": 2.075235712911295e-05,
|
1831 |
+
"loss": 0.9181,
|
1832 |
+
"step": 15200
|
1833 |
+
},
|
1834 |
+
{
|
1835 |
+
"epoch": 2.93,
|
1836 |
+
"learning_rate": 2.065614777756398e-05,
|
1837 |
+
"loss": 0.9437,
|
1838 |
+
"step": 15250
|
1839 |
+
},
|
1840 |
+
{
|
1841 |
+
"epoch": 2.94,
|
1842 |
+
"learning_rate": 2.055993842601501e-05,
|
1843 |
+
"loss": 0.9625,
|
1844 |
+
"step": 15300
|
1845 |
+
},
|
1846 |
+
{
|
1847 |
+
"epoch": 2.95,
|
1848 |
+
"learning_rate": 2.046372907446604e-05,
|
1849 |
+
"loss": 0.9204,
|
1850 |
+
"step": 15350
|
1851 |
+
},
|
1852 |
+
{
|
1853 |
+
"epoch": 2.96,
|
1854 |
+
"learning_rate": 2.036751972291707e-05,
|
1855 |
+
"loss": 0.916,
|
1856 |
+
"step": 15400
|
1857 |
+
},
|
1858 |
+
{
|
1859 |
+
"epoch": 2.97,
|
1860 |
+
"learning_rate": 2.02713103713681e-05,
|
1861 |
+
"loss": 0.8747,
|
1862 |
+
"step": 15450
|
1863 |
+
},
|
1864 |
+
{
|
1865 |
+
"epoch": 2.98,
|
1866 |
+
"learning_rate": 2.017510101981913e-05,
|
1867 |
+
"loss": 0.9374,
|
1868 |
+
"step": 15500
|
1869 |
+
},
|
1870 |
+
{
|
1871 |
+
"epoch": 2.99,
|
1872 |
+
"learning_rate": 2.0078891668270154e-05,
|
1873 |
+
"loss": 0.8864,
|
1874 |
+
"step": 15550
|
1875 |
+
},
|
1876 |
+
{
|
1877 |
+
"epoch": 3.0,
|
1878 |
+
"learning_rate": 1.9982682316721186e-05,
|
1879 |
+
"loss": 0.923,
|
1880 |
+
"step": 15600
|
1881 |
+
},
|
1882 |
+
{
|
1883 |
+
"epoch": 3.01,
|
1884 |
+
"learning_rate": 1.9886472965172214e-05,
|
1885 |
+
"loss": 0.8445,
|
1886 |
+
"step": 15650
|
1887 |
+
},
|
1888 |
+
{
|
1889 |
+
"epoch": 3.02,
|
1890 |
+
"learning_rate": 1.9790263613623245e-05,
|
1891 |
+
"loss": 0.8908,
|
1892 |
+
"step": 15700
|
1893 |
+
},
|
1894 |
+
{
|
1895 |
+
"epoch": 3.03,
|
1896 |
+
"learning_rate": 1.9694054262074273e-05,
|
1897 |
+
"loss": 0.8816,
|
1898 |
+
"step": 15750
|
1899 |
+
},
|
1900 |
+
{
|
1901 |
+
"epoch": 3.04,
|
1902 |
+
"learning_rate": 1.9597844910525305e-05,
|
1903 |
+
"loss": 0.8911,
|
1904 |
+
"step": 15800
|
1905 |
+
},
|
1906 |
+
{
|
1907 |
+
"epoch": 3.05,
|
1908 |
+
"learning_rate": 1.9501635558976333e-05,
|
1909 |
+
"loss": 0.9215,
|
1910 |
+
"step": 15850
|
1911 |
+
},
|
1912 |
+
{
|
1913 |
+
"epoch": 3.06,
|
1914 |
+
"learning_rate": 1.9405426207427364e-05,
|
1915 |
+
"loss": 0.8848,
|
1916 |
+
"step": 15900
|
1917 |
+
},
|
1918 |
+
{
|
1919 |
+
"epoch": 3.07,
|
1920 |
+
"learning_rate": 1.9309216855878392e-05,
|
1921 |
+
"loss": 0.8707,
|
1922 |
+
"step": 15950
|
1923 |
+
},
|
1924 |
+
{
|
1925 |
+
"epoch": 3.08,
|
1926 |
+
"learning_rate": 1.9213007504329424e-05,
|
1927 |
+
"loss": 0.8616,
|
1928 |
+
"step": 16000
|
1929 |
+
},
|
1930 |
+
{
|
1931 |
+
"epoch": 3.09,
|
1932 |
+
"learning_rate": 1.911679815278045e-05,
|
1933 |
+
"loss": 0.8581,
|
1934 |
+
"step": 16050
|
1935 |
+
},
|
1936 |
+
{
|
1937 |
+
"epoch": 3.1,
|
1938 |
+
"learning_rate": 1.9020588801231483e-05,
|
1939 |
+
"loss": 0.8828,
|
1940 |
+
"step": 16100
|
1941 |
+
},
|
1942 |
+
{
|
1943 |
+
"epoch": 3.11,
|
1944 |
+
"learning_rate": 1.892437944968251e-05,
|
1945 |
+
"loss": 0.9461,
|
1946 |
+
"step": 16150
|
1947 |
+
},
|
1948 |
+
{
|
1949 |
+
"epoch": 3.12,
|
1950 |
+
"learning_rate": 1.882817009813354e-05,
|
1951 |
+
"loss": 0.9004,
|
1952 |
+
"step": 16200
|
1953 |
+
},
|
1954 |
+
{
|
1955 |
+
"epoch": 3.13,
|
1956 |
+
"learning_rate": 1.8731960746584567e-05,
|
1957 |
+
"loss": 0.8839,
|
1958 |
+
"step": 16250
|
1959 |
+
},
|
1960 |
+
{
|
1961 |
+
"epoch": 3.14,
|
1962 |
+
"learning_rate": 1.86357513950356e-05,
|
1963 |
+
"loss": 0.8692,
|
1964 |
+
"step": 16300
|
1965 |
+
},
|
1966 |
+
{
|
1967 |
+
"epoch": 3.15,
|
1968 |
+
"learning_rate": 1.8539542043486627e-05,
|
1969 |
+
"loss": 0.918,
|
1970 |
+
"step": 16350
|
1971 |
+
},
|
1972 |
+
{
|
1973 |
+
"epoch": 3.16,
|
1974 |
+
"learning_rate": 1.8443332691937658e-05,
|
1975 |
+
"loss": 0.8885,
|
1976 |
+
"step": 16400
|
1977 |
+
},
|
1978 |
+
{
|
1979 |
+
"epoch": 3.17,
|
1980 |
+
"learning_rate": 1.8347123340388686e-05,
|
1981 |
+
"loss": 0.8852,
|
1982 |
+
"step": 16450
|
1983 |
+
},
|
1984 |
+
{
|
1985 |
+
"epoch": 3.17,
|
1986 |
+
"learning_rate": 1.8250913988839714e-05,
|
1987 |
+
"loss": 0.9233,
|
1988 |
+
"step": 16500
|
1989 |
+
},
|
1990 |
+
{
|
1991 |
+
"epoch": 3.18,
|
1992 |
+
"learning_rate": 1.8154704637290746e-05,
|
1993 |
+
"loss": 0.9075,
|
1994 |
+
"step": 16550
|
1995 |
+
},
|
1996 |
+
{
|
1997 |
+
"epoch": 3.19,
|
1998 |
+
"learning_rate": 1.8058495285741774e-05,
|
1999 |
+
"loss": 0.919,
|
2000 |
+
"step": 16600
|
2001 |
+
},
|
2002 |
+
{
|
2003 |
+
"epoch": 3.2,
|
2004 |
+
"learning_rate": 1.7962285934192805e-05,
|
2005 |
+
"loss": 0.9143,
|
2006 |
+
"step": 16650
|
2007 |
+
},
|
2008 |
+
{
|
2009 |
+
"epoch": 3.21,
|
2010 |
+
"learning_rate": 1.7866076582643833e-05,
|
2011 |
+
"loss": 0.9093,
|
2012 |
+
"step": 16700
|
2013 |
+
},
|
2014 |
+
{
|
2015 |
+
"epoch": 3.22,
|
2016 |
+
"learning_rate": 1.7769867231094864e-05,
|
2017 |
+
"loss": 0.8987,
|
2018 |
+
"step": 16750
|
2019 |
+
},
|
2020 |
+
{
|
2021 |
+
"epoch": 3.23,
|
2022 |
+
"learning_rate": 1.7673657879545892e-05,
|
2023 |
+
"loss": 0.9162,
|
2024 |
+
"step": 16800
|
2025 |
+
},
|
2026 |
+
{
|
2027 |
+
"epoch": 3.24,
|
2028 |
+
"learning_rate": 1.7577448527996924e-05,
|
2029 |
+
"loss": 0.8884,
|
2030 |
+
"step": 16850
|
2031 |
+
},
|
2032 |
+
{
|
2033 |
+
"epoch": 3.25,
|
2034 |
+
"learning_rate": 1.7481239176447952e-05,
|
2035 |
+
"loss": 0.8765,
|
2036 |
+
"step": 16900
|
2037 |
+
},
|
2038 |
+
{
|
2039 |
+
"epoch": 3.26,
|
2040 |
+
"learning_rate": 1.7385029824898983e-05,
|
2041 |
+
"loss": 0.8825,
|
2042 |
+
"step": 16950
|
2043 |
+
},
|
2044 |
+
{
|
2045 |
+
"epoch": 3.27,
|
2046 |
+
"learning_rate": 1.7288820473350008e-05,
|
2047 |
+
"loss": 0.8595,
|
2048 |
+
"step": 17000
|
2049 |
+
},
|
2050 |
+
{
|
2051 |
+
"epoch": 3.28,
|
2052 |
+
"learning_rate": 1.719261112180104e-05,
|
2053 |
+
"loss": 0.8704,
|
2054 |
+
"step": 17050
|
2055 |
+
},
|
2056 |
+
{
|
2057 |
+
"epoch": 3.29,
|
2058 |
+
"learning_rate": 1.7096401770252067e-05,
|
2059 |
+
"loss": 0.8655,
|
2060 |
+
"step": 17100
|
2061 |
+
},
|
2062 |
+
{
|
2063 |
+
"epoch": 3.3,
|
2064 |
+
"learning_rate": 1.70001924187031e-05,
|
2065 |
+
"loss": 0.8501,
|
2066 |
+
"step": 17150
|
2067 |
+
},
|
2068 |
+
{
|
2069 |
+
"epoch": 3.31,
|
2070 |
+
"learning_rate": 1.6903983067154127e-05,
|
2071 |
+
"loss": 0.8723,
|
2072 |
+
"step": 17200
|
2073 |
+
},
|
2074 |
+
{
|
2075 |
+
"epoch": 3.32,
|
2076 |
+
"learning_rate": 1.680777371560516e-05,
|
2077 |
+
"loss": 0.9043,
|
2078 |
+
"step": 17250
|
2079 |
+
},
|
2080 |
+
{
|
2081 |
+
"epoch": 3.33,
|
2082 |
+
"learning_rate": 1.6711564364056186e-05,
|
2083 |
+
"loss": 0.9083,
|
2084 |
+
"step": 17300
|
2085 |
+
},
|
2086 |
+
{
|
2087 |
+
"epoch": 3.34,
|
2088 |
+
"learning_rate": 1.6615355012507218e-05,
|
2089 |
+
"loss": 0.8813,
|
2090 |
+
"step": 17350
|
2091 |
+
},
|
2092 |
+
{
|
2093 |
+
"epoch": 3.35,
|
2094 |
+
"learning_rate": 1.6519145660958246e-05,
|
2095 |
+
"loss": 0.901,
|
2096 |
+
"step": 17400
|
2097 |
+
},
|
2098 |
+
{
|
2099 |
+
"epoch": 3.36,
|
2100 |
+
"learning_rate": 1.6422936309409277e-05,
|
2101 |
+
"loss": 0.8891,
|
2102 |
+
"step": 17450
|
2103 |
+
},
|
2104 |
+
{
|
2105 |
+
"epoch": 3.37,
|
2106 |
+
"learning_rate": 1.6326726957860305e-05,
|
2107 |
+
"loss": 0.8703,
|
2108 |
+
"step": 17500
|
2109 |
+
},
|
2110 |
+
{
|
2111 |
+
"epoch": 3.38,
|
2112 |
+
"learning_rate": 1.6230517606311337e-05,
|
2113 |
+
"loss": 0.8793,
|
2114 |
+
"step": 17550
|
2115 |
+
},
|
2116 |
+
{
|
2117 |
+
"epoch": 3.39,
|
2118 |
+
"learning_rate": 1.6134308254762365e-05,
|
2119 |
+
"loss": 0.9044,
|
2120 |
+
"step": 17600
|
2121 |
+
},
|
2122 |
+
{
|
2123 |
+
"epoch": 3.4,
|
2124 |
+
"learning_rate": 1.6038098903213393e-05,
|
2125 |
+
"loss": 0.8915,
|
2126 |
+
"step": 17650
|
2127 |
+
},
|
2128 |
+
{
|
2129 |
+
"epoch": 3.41,
|
2130 |
+
"learning_rate": 1.594188955166442e-05,
|
2131 |
+
"loss": 0.8458,
|
2132 |
+
"step": 17700
|
2133 |
+
},
|
2134 |
+
{
|
2135 |
+
"epoch": 3.42,
|
2136 |
+
"learning_rate": 1.5845680200115452e-05,
|
2137 |
+
"loss": 0.9001,
|
2138 |
+
"step": 17750
|
2139 |
+
},
|
2140 |
+
{
|
2141 |
+
"epoch": 3.43,
|
2142 |
+
"learning_rate": 1.574947084856648e-05,
|
2143 |
+
"loss": 0.8719,
|
2144 |
+
"step": 17800
|
2145 |
+
},
|
2146 |
+
{
|
2147 |
+
"epoch": 3.43,
|
2148 |
+
"learning_rate": 1.565326149701751e-05,
|
2149 |
+
"loss": 0.8884,
|
2150 |
+
"step": 17850
|
2151 |
+
},
|
2152 |
+
{
|
2153 |
+
"epoch": 3.44,
|
2154 |
+
"learning_rate": 1.555705214546854e-05,
|
2155 |
+
"loss": 0.9189,
|
2156 |
+
"step": 17900
|
2157 |
+
},
|
2158 |
+
{
|
2159 |
+
"epoch": 3.45,
|
2160 |
+
"learning_rate": 1.5460842793919568e-05,
|
2161 |
+
"loss": 0.9295,
|
2162 |
+
"step": 17950
|
2163 |
+
},
|
2164 |
+
{
|
2165 |
+
"epoch": 3.46,
|
2166 |
+
"learning_rate": 1.53646334423706e-05,
|
2167 |
+
"loss": 0.8898,
|
2168 |
+
"step": 18000
|
2169 |
+
},
|
2170 |
+
{
|
2171 |
+
"epoch": 3.47,
|
2172 |
+
"learning_rate": 1.5268424090821627e-05,
|
2173 |
+
"loss": 0.8434,
|
2174 |
+
"step": 18050
|
2175 |
+
},
|
2176 |
+
{
|
2177 |
+
"epoch": 3.48,
|
2178 |
+
"learning_rate": 1.5172214739272659e-05,
|
2179 |
+
"loss": 0.8689,
|
2180 |
+
"step": 18100
|
2181 |
+
},
|
2182 |
+
{
|
2183 |
+
"epoch": 3.49,
|
2184 |
+
"learning_rate": 1.5076005387723688e-05,
|
2185 |
+
"loss": 0.8968,
|
2186 |
+
"step": 18150
|
2187 |
+
},
|
2188 |
+
{
|
2189 |
+
"epoch": 3.5,
|
2190 |
+
"learning_rate": 1.4979796036174718e-05,
|
2191 |
+
"loss": 0.9064,
|
2192 |
+
"step": 18200
|
2193 |
+
},
|
2194 |
+
{
|
2195 |
+
"epoch": 3.51,
|
2196 |
+
"learning_rate": 1.4883586684625748e-05,
|
2197 |
+
"loss": 0.8808,
|
2198 |
+
"step": 18250
|
2199 |
+
},
|
2200 |
+
{
|
2201 |
+
"epoch": 3.52,
|
2202 |
+
"learning_rate": 1.4787377333076778e-05,
|
2203 |
+
"loss": 0.8748,
|
2204 |
+
"step": 18300
|
2205 |
+
},
|
2206 |
+
{
|
2207 |
+
"epoch": 3.53,
|
2208 |
+
"learning_rate": 1.4691167981527806e-05,
|
2209 |
+
"loss": 0.8964,
|
2210 |
+
"step": 18350
|
2211 |
+
},
|
2212 |
+
{
|
2213 |
+
"epoch": 3.54,
|
2214 |
+
"learning_rate": 1.4594958629978835e-05,
|
2215 |
+
"loss": 0.9299,
|
2216 |
+
"step": 18400
|
2217 |
+
},
|
2218 |
+
{
|
2219 |
+
"epoch": 3.55,
|
2220 |
+
"learning_rate": 1.4498749278429863e-05,
|
2221 |
+
"loss": 0.8759,
|
2222 |
+
"step": 18450
|
2223 |
+
},
|
2224 |
+
{
|
2225 |
+
"epoch": 3.56,
|
2226 |
+
"learning_rate": 1.4402539926880893e-05,
|
2227 |
+
"loss": 0.8822,
|
2228 |
+
"step": 18500
|
2229 |
+
},
|
2230 |
+
{
|
2231 |
+
"epoch": 3.57,
|
2232 |
+
"learning_rate": 1.4306330575331923e-05,
|
2233 |
+
"loss": 0.8792,
|
2234 |
+
"step": 18550
|
2235 |
+
},
|
2236 |
+
{
|
2237 |
+
"epoch": 3.58,
|
2238 |
+
"learning_rate": 1.4210121223782952e-05,
|
2239 |
+
"loss": 0.8443,
|
2240 |
+
"step": 18600
|
2241 |
+
},
|
2242 |
+
{
|
2243 |
+
"epoch": 3.59,
|
2244 |
+
"learning_rate": 1.411391187223398e-05,
|
2245 |
+
"loss": 0.8951,
|
2246 |
+
"step": 18650
|
2247 |
+
},
|
2248 |
+
{
|
2249 |
+
"epoch": 3.6,
|
2250 |
+
"learning_rate": 1.401770252068501e-05,
|
2251 |
+
"loss": 0.9153,
|
2252 |
+
"step": 18700
|
2253 |
+
},
|
2254 |
+
{
|
2255 |
+
"epoch": 3.61,
|
2256 |
+
"learning_rate": 1.392149316913604e-05,
|
2257 |
+
"loss": 0.8899,
|
2258 |
+
"step": 18750
|
2259 |
+
},
|
2260 |
+
{
|
2261 |
+
"epoch": 3.62,
|
2262 |
+
"learning_rate": 1.382528381758707e-05,
|
2263 |
+
"loss": 0.9196,
|
2264 |
+
"step": 18800
|
2265 |
+
},
|
2266 |
+
{
|
2267 |
+
"epoch": 3.63,
|
2268 |
+
"learning_rate": 1.37290744660381e-05,
|
2269 |
+
"loss": 0.8981,
|
2270 |
+
"step": 18850
|
2271 |
+
},
|
2272 |
+
{
|
2273 |
+
"epoch": 3.64,
|
2274 |
+
"learning_rate": 1.363286511448913e-05,
|
2275 |
+
"loss": 0.8634,
|
2276 |
+
"step": 18900
|
2277 |
+
},
|
2278 |
+
{
|
2279 |
+
"epoch": 3.65,
|
2280 |
+
"learning_rate": 1.3536655762940159e-05,
|
2281 |
+
"loss": 0.8741,
|
2282 |
+
"step": 18950
|
2283 |
+
},
|
2284 |
+
{
|
2285 |
+
"epoch": 3.66,
|
2286 |
+
"learning_rate": 1.3440446411391189e-05,
|
2287 |
+
"loss": 0.8729,
|
2288 |
+
"step": 19000
|
2289 |
+
},
|
2290 |
+
{
|
2291 |
+
"epoch": 3.67,
|
2292 |
+
"learning_rate": 1.3344237059842218e-05,
|
2293 |
+
"loss": 0.9247,
|
2294 |
+
"step": 19050
|
2295 |
+
},
|
2296 |
+
{
|
2297 |
+
"epoch": 3.68,
|
2298 |
+
"learning_rate": 1.3248027708293248e-05,
|
2299 |
+
"loss": 0.9436,
|
2300 |
+
"step": 19100
|
2301 |
+
},
|
2302 |
+
{
|
2303 |
+
"epoch": 3.68,
|
2304 |
+
"learning_rate": 1.3151818356744274e-05,
|
2305 |
+
"loss": 0.8898,
|
2306 |
+
"step": 19150
|
2307 |
+
},
|
2308 |
+
{
|
2309 |
+
"epoch": 3.69,
|
2310 |
+
"learning_rate": 1.3055609005195304e-05,
|
2311 |
+
"loss": 0.9404,
|
2312 |
+
"step": 19200
|
2313 |
+
},
|
2314 |
+
{
|
2315 |
+
"epoch": 3.7,
|
2316 |
+
"learning_rate": 1.2959399653646334e-05,
|
2317 |
+
"loss": 0.8733,
|
2318 |
+
"step": 19250
|
2319 |
+
},
|
2320 |
+
{
|
2321 |
+
"epoch": 3.71,
|
2322 |
+
"learning_rate": 1.2863190302097364e-05,
|
2323 |
+
"loss": 0.8954,
|
2324 |
+
"step": 19300
|
2325 |
+
},
|
2326 |
+
{
|
2327 |
+
"epoch": 3.72,
|
2328 |
+
"learning_rate": 1.2766980950548393e-05,
|
2329 |
+
"loss": 0.8925,
|
2330 |
+
"step": 19350
|
2331 |
+
},
|
2332 |
+
{
|
2333 |
+
"epoch": 3.73,
|
2334 |
+
"learning_rate": 1.2670771598999423e-05,
|
2335 |
+
"loss": 0.8607,
|
2336 |
+
"step": 19400
|
2337 |
+
},
|
2338 |
+
{
|
2339 |
+
"epoch": 3.74,
|
2340 |
+
"learning_rate": 1.2574562247450453e-05,
|
2341 |
+
"loss": 0.8658,
|
2342 |
+
"step": 19450
|
2343 |
+
},
|
2344 |
+
{
|
2345 |
+
"epoch": 3.75,
|
2346 |
+
"learning_rate": 1.2478352895901482e-05,
|
2347 |
+
"loss": 0.9051,
|
2348 |
+
"step": 19500
|
2349 |
+
},
|
2350 |
+
{
|
2351 |
+
"epoch": 3.76,
|
2352 |
+
"learning_rate": 1.2382143544352512e-05,
|
2353 |
+
"loss": 0.8882,
|
2354 |
+
"step": 19550
|
2355 |
+
},
|
2356 |
+
{
|
2357 |
+
"epoch": 3.77,
|
2358 |
+
"learning_rate": 1.2285934192803542e-05,
|
2359 |
+
"loss": 0.8899,
|
2360 |
+
"step": 19600
|
2361 |
+
},
|
2362 |
+
{
|
2363 |
+
"epoch": 3.78,
|
2364 |
+
"learning_rate": 1.2189724841254572e-05,
|
2365 |
+
"loss": 0.9144,
|
2366 |
+
"step": 19650
|
2367 |
+
},
|
2368 |
+
{
|
2369 |
+
"epoch": 3.79,
|
2370 |
+
"learning_rate": 1.20935154897056e-05,
|
2371 |
+
"loss": 0.8804,
|
2372 |
+
"step": 19700
|
2373 |
+
},
|
2374 |
+
{
|
2375 |
+
"epoch": 3.8,
|
2376 |
+
"learning_rate": 1.199730613815663e-05,
|
2377 |
+
"loss": 0.8631,
|
2378 |
+
"step": 19750
|
2379 |
+
},
|
2380 |
+
{
|
2381 |
+
"epoch": 3.81,
|
2382 |
+
"learning_rate": 1.190109678660766e-05,
|
2383 |
+
"loss": 0.8948,
|
2384 |
+
"step": 19800
|
2385 |
+
},
|
2386 |
+
{
|
2387 |
+
"epoch": 3.82,
|
2388 |
+
"learning_rate": 1.1804887435058689e-05,
|
2389 |
+
"loss": 0.8623,
|
2390 |
+
"step": 19850
|
2391 |
+
},
|
2392 |
+
{
|
2393 |
+
"epoch": 3.83,
|
2394 |
+
"learning_rate": 1.1708678083509719e-05,
|
2395 |
+
"loss": 0.8337,
|
2396 |
+
"step": 19900
|
2397 |
+
},
|
2398 |
+
{
|
2399 |
+
"epoch": 3.84,
|
2400 |
+
"learning_rate": 1.1612468731960748e-05,
|
2401 |
+
"loss": 0.8464,
|
2402 |
+
"step": 19950
|
2403 |
+
},
|
2404 |
+
{
|
2405 |
+
"epoch": 3.85,
|
2406 |
+
"learning_rate": 1.1516259380411776e-05,
|
2407 |
+
"loss": 0.8713,
|
2408 |
+
"step": 20000
|
2409 |
+
},
|
2410 |
+
{
|
2411 |
+
"epoch": 3.86,
|
2412 |
+
"learning_rate": 1.1420050028862806e-05,
|
2413 |
+
"loss": 0.8843,
|
2414 |
+
"step": 20050
|
2415 |
+
},
|
2416 |
+
{
|
2417 |
+
"epoch": 3.87,
|
2418 |
+
"learning_rate": 1.1323840677313836e-05,
|
2419 |
+
"loss": 0.8483,
|
2420 |
+
"step": 20100
|
2421 |
+
},
|
2422 |
+
{
|
2423 |
+
"epoch": 3.88,
|
2424 |
+
"learning_rate": 1.1227631325764864e-05,
|
2425 |
+
"loss": 0.8579,
|
2426 |
+
"step": 20150
|
2427 |
+
},
|
2428 |
+
{
|
2429 |
+
"epoch": 3.89,
|
2430 |
+
"learning_rate": 1.1131421974215894e-05,
|
2431 |
+
"loss": 0.8882,
|
2432 |
+
"step": 20200
|
2433 |
+
},
|
2434 |
+
{
|
2435 |
+
"epoch": 3.9,
|
2436 |
+
"learning_rate": 1.1035212622666923e-05,
|
2437 |
+
"loss": 0.8448,
|
2438 |
+
"step": 20250
|
2439 |
+
},
|
2440 |
+
{
|
2441 |
+
"epoch": 3.91,
|
2442 |
+
"learning_rate": 1.0939003271117953e-05,
|
2443 |
+
"loss": 0.8808,
|
2444 |
+
"step": 20300
|
2445 |
+
},
|
2446 |
+
{
|
2447 |
+
"epoch": 3.92,
|
2448 |
+
"learning_rate": 1.0842793919568983e-05,
|
2449 |
+
"loss": 0.8838,
|
2450 |
+
"step": 20350
|
2451 |
+
},
|
2452 |
+
{
|
2453 |
+
"epoch": 3.93,
|
2454 |
+
"learning_rate": 1.074658456802001e-05,
|
2455 |
+
"loss": 0.8926,
|
2456 |
+
"step": 20400
|
2457 |
+
},
|
2458 |
+
{
|
2459 |
+
"epoch": 3.93,
|
2460 |
+
"learning_rate": 1.065037521647104e-05,
|
2461 |
+
"loss": 0.8919,
|
2462 |
+
"step": 20450
|
2463 |
+
},
|
2464 |
+
{
|
2465 |
+
"epoch": 3.94,
|
2466 |
+
"learning_rate": 1.055416586492207e-05,
|
2467 |
+
"loss": 0.8706,
|
2468 |
+
"step": 20500
|
2469 |
+
},
|
2470 |
+
{
|
2471 |
+
"epoch": 3.95,
|
2472 |
+
"learning_rate": 1.04579565133731e-05,
|
2473 |
+
"loss": 0.8564,
|
2474 |
+
"step": 20550
|
2475 |
+
},
|
2476 |
+
{
|
2477 |
+
"epoch": 3.96,
|
2478 |
+
"learning_rate": 1.036174716182413e-05,
|
2479 |
+
"loss": 0.8579,
|
2480 |
+
"step": 20600
|
2481 |
+
},
|
2482 |
+
{
|
2483 |
+
"epoch": 3.97,
|
2484 |
+
"learning_rate": 1.026553781027516e-05,
|
2485 |
+
"loss": 0.9184,
|
2486 |
+
"step": 20650
|
2487 |
+
},
|
2488 |
+
{
|
2489 |
+
"epoch": 3.98,
|
2490 |
+
"learning_rate": 1.016932845872619e-05,
|
2491 |
+
"loss": 0.8813,
|
2492 |
+
"step": 20700
|
2493 |
+
},
|
2494 |
+
{
|
2495 |
+
"epoch": 3.99,
|
2496 |
+
"learning_rate": 1.0073119107177219e-05,
|
2497 |
+
"loss": 0.8257,
|
2498 |
+
"step": 20750
|
2499 |
+
},
|
2500 |
+
{
|
2501 |
+
"epoch": 4.0,
|
2502 |
+
"learning_rate": 9.976909755628247e-06,
|
2503 |
+
"loss": 0.8352,
|
2504 |
+
"step": 20800
|
2505 |
+
},
|
2506 |
+
{
|
2507 |
+
"epoch": 4.01,
|
2508 |
+
"learning_rate": 9.880700404079277e-06,
|
2509 |
+
"loss": 0.8739,
|
2510 |
+
"step": 20850
|
2511 |
+
},
|
2512 |
+
{
|
2513 |
+
"epoch": 4.02,
|
2514 |
+
"learning_rate": 9.784491052530306e-06,
|
2515 |
+
"loss": 0.8762,
|
2516 |
+
"step": 20900
|
2517 |
+
},
|
2518 |
+
{
|
2519 |
+
"epoch": 4.03,
|
2520 |
+
"learning_rate": 9.688281700981336e-06,
|
2521 |
+
"loss": 0.8652,
|
2522 |
+
"step": 20950
|
2523 |
+
},
|
2524 |
+
{
|
2525 |
+
"epoch": 4.04,
|
2526 |
+
"learning_rate": 9.592072349432366e-06,
|
2527 |
+
"loss": 0.8875,
|
2528 |
+
"step": 21000
|
2529 |
+
},
|
2530 |
+
{
|
2531 |
+
"epoch": 4.05,
|
2532 |
+
"learning_rate": 9.495862997883396e-06,
|
2533 |
+
"loss": 0.8277,
|
2534 |
+
"step": 21050
|
2535 |
+
},
|
2536 |
+
{
|
2537 |
+
"epoch": 4.06,
|
2538 |
+
"learning_rate": 9.399653646334425e-06,
|
2539 |
+
"loss": 0.8568,
|
2540 |
+
"step": 21100
|
2541 |
+
},
|
2542 |
+
{
|
2543 |
+
"epoch": 4.07,
|
2544 |
+
"learning_rate": 9.303444294785453e-06,
|
2545 |
+
"loss": 0.8651,
|
2546 |
+
"step": 21150
|
2547 |
+
},
|
2548 |
+
{
|
2549 |
+
"epoch": 4.08,
|
2550 |
+
"learning_rate": 9.207234943236483e-06,
|
2551 |
+
"loss": 0.8883,
|
2552 |
+
"step": 21200
|
2553 |
+
},
|
2554 |
+
{
|
2555 |
+
"epoch": 4.09,
|
2556 |
+
"learning_rate": 9.111025591687513e-06,
|
2557 |
+
"loss": 0.9303,
|
2558 |
+
"step": 21250
|
2559 |
+
},
|
2560 |
+
{
|
2561 |
+
"epoch": 4.1,
|
2562 |
+
"learning_rate": 9.014816240138543e-06,
|
2563 |
+
"loss": 0.8416,
|
2564 |
+
"step": 21300
|
2565 |
+
},
|
2566 |
+
{
|
2567 |
+
"epoch": 4.11,
|
2568 |
+
"learning_rate": 8.918606888589572e-06,
|
2569 |
+
"loss": 0.8409,
|
2570 |
+
"step": 21350
|
2571 |
+
},
|
2572 |
+
{
|
2573 |
+
"epoch": 4.12,
|
2574 |
+
"learning_rate": 8.822397537040602e-06,
|
2575 |
+
"loss": 0.9157,
|
2576 |
+
"step": 21400
|
2577 |
+
},
|
2578 |
+
{
|
2579 |
+
"epoch": 4.13,
|
2580 |
+
"learning_rate": 8.72618818549163e-06,
|
2581 |
+
"loss": 0.8638,
|
2582 |
+
"step": 21450
|
2583 |
+
},
|
2584 |
+
{
|
2585 |
+
"epoch": 4.14,
|
2586 |
+
"learning_rate": 8.62997883394266e-06,
|
2587 |
+
"loss": 0.8443,
|
2588 |
+
"step": 21500
|
2589 |
+
},
|
2590 |
+
{
|
2591 |
+
"epoch": 4.15,
|
2592 |
+
"learning_rate": 8.53376948239369e-06,
|
2593 |
+
"loss": 0.8533,
|
2594 |
+
"step": 21550
|
2595 |
+
},
|
2596 |
+
{
|
2597 |
+
"epoch": 4.16,
|
2598 |
+
"learning_rate": 8.437560130844717e-06,
|
2599 |
+
"loss": 0.8888,
|
2600 |
+
"step": 21600
|
2601 |
+
},
|
2602 |
+
{
|
2603 |
+
"epoch": 4.17,
|
2604 |
+
"learning_rate": 8.341350779295747e-06,
|
2605 |
+
"loss": 0.8741,
|
2606 |
+
"step": 21650
|
2607 |
+
},
|
2608 |
+
{
|
2609 |
+
"epoch": 4.18,
|
2610 |
+
"learning_rate": 8.245141427746777e-06,
|
2611 |
+
"loss": 0.876,
|
2612 |
+
"step": 21700
|
2613 |
+
},
|
2614 |
+
{
|
2615 |
+
"epoch": 4.19,
|
2616 |
+
"learning_rate": 8.148932076197807e-06,
|
2617 |
+
"loss": 0.9204,
|
2618 |
+
"step": 21750
|
2619 |
+
},
|
2620 |
+
{
|
2621 |
+
"epoch": 4.19,
|
2622 |
+
"learning_rate": 8.052722724648836e-06,
|
2623 |
+
"loss": 0.9072,
|
2624 |
+
"step": 21800
|
2625 |
+
},
|
2626 |
+
{
|
2627 |
+
"epoch": 4.2,
|
2628 |
+
"learning_rate": 7.956513373099864e-06,
|
2629 |
+
"loss": 0.8982,
|
2630 |
+
"step": 21850
|
2631 |
+
},
|
2632 |
+
{
|
2633 |
+
"epoch": 4.21,
|
2634 |
+
"learning_rate": 7.860304021550894e-06,
|
2635 |
+
"loss": 0.881,
|
2636 |
+
"step": 21900
|
2637 |
+
},
|
2638 |
+
{
|
2639 |
+
"epoch": 4.22,
|
2640 |
+
"learning_rate": 7.764094670001924e-06,
|
2641 |
+
"loss": 0.8485,
|
2642 |
+
"step": 21950
|
2643 |
+
},
|
2644 |
+
{
|
2645 |
+
"epoch": 4.23,
|
2646 |
+
"learning_rate": 7.667885318452954e-06,
|
2647 |
+
"loss": 0.8694,
|
2648 |
+
"step": 22000
|
2649 |
+
},
|
2650 |
+
{
|
2651 |
+
"epoch": 4.24,
|
2652 |
+
"learning_rate": 7.571675966903983e-06,
|
2653 |
+
"loss": 0.8448,
|
2654 |
+
"step": 22050
|
2655 |
+
},
|
2656 |
+
{
|
2657 |
+
"epoch": 4.25,
|
2658 |
+
"learning_rate": 7.475466615355013e-06,
|
2659 |
+
"loss": 0.8995,
|
2660 |
+
"step": 22100
|
2661 |
+
},
|
2662 |
+
{
|
2663 |
+
"epoch": 4.26,
|
2664 |
+
"learning_rate": 7.379257263806043e-06,
|
2665 |
+
"loss": 0.9009,
|
2666 |
+
"step": 22150
|
2667 |
+
},
|
2668 |
+
{
|
2669 |
+
"epoch": 4.27,
|
2670 |
+
"learning_rate": 7.283047912257071e-06,
|
2671 |
+
"loss": 0.8966,
|
2672 |
+
"step": 22200
|
2673 |
+
},
|
2674 |
+
{
|
2675 |
+
"epoch": 4.28,
|
2676 |
+
"learning_rate": 7.1868385607081006e-06,
|
2677 |
+
"loss": 0.872,
|
2678 |
+
"step": 22250
|
2679 |
+
},
|
2680 |
+
{
|
2681 |
+
"epoch": 4.29,
|
2682 |
+
"learning_rate": 7.09062920915913e-06,
|
2683 |
+
"loss": 0.8402,
|
2684 |
+
"step": 22300
|
2685 |
+
},
|
2686 |
+
{
|
2687 |
+
"epoch": 4.3,
|
2688 |
+
"learning_rate": 6.99441985761016e-06,
|
2689 |
+
"loss": 0.8175,
|
2690 |
+
"step": 22350
|
2691 |
+
},
|
2692 |
+
{
|
2693 |
+
"epoch": 4.31,
|
2694 |
+
"learning_rate": 6.89821050606119e-06,
|
2695 |
+
"loss": 0.8563,
|
2696 |
+
"step": 22400
|
2697 |
+
},
|
2698 |
+
{
|
2699 |
+
"epoch": 4.32,
|
2700 |
+
"learning_rate": 6.8020011545122195e-06,
|
2701 |
+
"loss": 0.8987,
|
2702 |
+
"step": 22450
|
2703 |
+
},
|
2704 |
+
{
|
2705 |
+
"epoch": 4.33,
|
2706 |
+
"learning_rate": 6.705791802963249e-06,
|
2707 |
+
"loss": 0.8586,
|
2708 |
+
"step": 22500
|
2709 |
+
},
|
2710 |
+
{
|
2711 |
+
"epoch": 4.34,
|
2712 |
+
"learning_rate": 6.609582451414278e-06,
|
2713 |
+
"loss": 0.8356,
|
2714 |
+
"step": 22550
|
2715 |
+
},
|
2716 |
+
{
|
2717 |
+
"epoch": 4.35,
|
2718 |
+
"learning_rate": 6.513373099865307e-06,
|
2719 |
+
"loss": 0.883,
|
2720 |
+
"step": 22600
|
2721 |
+
},
|
2722 |
+
{
|
2723 |
+
"epoch": 4.36,
|
2724 |
+
"learning_rate": 6.417163748316337e-06,
|
2725 |
+
"loss": 0.8474,
|
2726 |
+
"step": 22650
|
2727 |
+
},
|
2728 |
+
{
|
2729 |
+
"epoch": 4.37,
|
2730 |
+
"learning_rate": 6.3209543967673656e-06,
|
2731 |
+
"loss": 0.8856,
|
2732 |
+
"step": 22700
|
2733 |
+
},
|
2734 |
+
{
|
2735 |
+
"epoch": 4.38,
|
2736 |
+
"learning_rate": 6.224745045218395e-06,
|
2737 |
+
"loss": 0.8719,
|
2738 |
+
"step": 22750
|
2739 |
+
},
|
2740 |
+
{
|
2741 |
+
"epoch": 4.39,
|
2742 |
+
"learning_rate": 6.128535693669425e-06,
|
2743 |
+
"loss": 0.8827,
|
2744 |
+
"step": 22800
|
2745 |
+
},
|
2746 |
+
{
|
2747 |
+
"epoch": 4.4,
|
2748 |
+
"learning_rate": 6.032326342120454e-06,
|
2749 |
+
"loss": 0.8108,
|
2750 |
+
"step": 22850
|
2751 |
+
},
|
2752 |
+
{
|
2753 |
+
"epoch": 4.41,
|
2754 |
+
"learning_rate": 5.936116990571484e-06,
|
2755 |
+
"loss": 0.8395,
|
2756 |
+
"step": 22900
|
2757 |
+
},
|
2758 |
+
{
|
2759 |
+
"epoch": 4.42,
|
2760 |
+
"learning_rate": 5.839907639022513e-06,
|
2761 |
+
"loss": 0.8933,
|
2762 |
+
"step": 22950
|
2763 |
+
},
|
2764 |
+
{
|
2765 |
+
"epoch": 4.43,
|
2766 |
+
"learning_rate": 5.743698287473543e-06,
|
2767 |
+
"loss": 0.8987,
|
2768 |
+
"step": 23000
|
2769 |
+
},
|
2770 |
+
{
|
2771 |
+
"epoch": 4.44,
|
2772 |
+
"learning_rate": 5.647488935924572e-06,
|
2773 |
+
"loss": 0.8256,
|
2774 |
+
"step": 23050
|
2775 |
+
},
|
2776 |
+
{
|
2777 |
+
"epoch": 4.44,
|
2778 |
+
"learning_rate": 5.551279584375602e-06,
|
2779 |
+
"loss": 0.827,
|
2780 |
+
"step": 23100
|
2781 |
+
},
|
2782 |
+
{
|
2783 |
+
"epoch": 4.45,
|
2784 |
+
"learning_rate": 5.455070232826631e-06,
|
2785 |
+
"loss": 0.867,
|
2786 |
+
"step": 23150
|
2787 |
+
},
|
2788 |
+
{
|
2789 |
+
"epoch": 4.46,
|
2790 |
+
"learning_rate": 5.35886088127766e-06,
|
2791 |
+
"loss": 0.8714,
|
2792 |
+
"step": 23200
|
2793 |
+
},
|
2794 |
+
{
|
2795 |
+
"epoch": 4.47,
|
2796 |
+
"learning_rate": 5.26265152972869e-06,
|
2797 |
+
"loss": 0.8998,
|
2798 |
+
"step": 23250
|
2799 |
+
},
|
2800 |
+
{
|
2801 |
+
"epoch": 4.48,
|
2802 |
+
"learning_rate": 5.16644217817972e-06,
|
2803 |
+
"loss": 0.8851,
|
2804 |
+
"step": 23300
|
2805 |
+
},
|
2806 |
+
{
|
2807 |
+
"epoch": 4.49,
|
2808 |
+
"learning_rate": 5.070232826630749e-06,
|
2809 |
+
"loss": 0.8945,
|
2810 |
+
"step": 23350
|
2811 |
+
},
|
2812 |
+
{
|
2813 |
+
"epoch": 4.5,
|
2814 |
+
"learning_rate": 4.974023475081778e-06,
|
2815 |
+
"loss": 0.8409,
|
2816 |
+
"step": 23400
|
2817 |
+
},
|
2818 |
+
{
|
2819 |
+
"epoch": 4.51,
|
2820 |
+
"learning_rate": 4.877814123532807e-06,
|
2821 |
+
"loss": 0.8124,
|
2822 |
+
"step": 23450
|
2823 |
+
},
|
2824 |
+
{
|
2825 |
+
"epoch": 4.52,
|
2826 |
+
"learning_rate": 4.781604771983837e-06,
|
2827 |
+
"loss": 0.8847,
|
2828 |
+
"step": 23500
|
2829 |
+
},
|
2830 |
+
{
|
2831 |
+
"epoch": 4.53,
|
2832 |
+
"learning_rate": 4.685395420434866e-06,
|
2833 |
+
"loss": 0.8554,
|
2834 |
+
"step": 23550
|
2835 |
+
},
|
2836 |
+
{
|
2837 |
+
"epoch": 4.54,
|
2838 |
+
"learning_rate": 4.5891860688858956e-06,
|
2839 |
+
"loss": 0.8627,
|
2840 |
+
"step": 23600
|
2841 |
+
},
|
2842 |
+
{
|
2843 |
+
"epoch": 4.55,
|
2844 |
+
"learning_rate": 4.492976717336925e-06,
|
2845 |
+
"loss": 0.8093,
|
2846 |
+
"step": 23650
|
2847 |
+
},
|
2848 |
+
{
|
2849 |
+
"epoch": 4.56,
|
2850 |
+
"learning_rate": 4.396767365787955e-06,
|
2851 |
+
"loss": 0.8283,
|
2852 |
+
"step": 23700
|
2853 |
+
},
|
2854 |
+
{
|
2855 |
+
"epoch": 4.57,
|
2856 |
+
"learning_rate": 4.300558014238984e-06,
|
2857 |
+
"loss": 0.8821,
|
2858 |
+
"step": 23750
|
2859 |
+
},
|
2860 |
+
{
|
2861 |
+
"epoch": 4.58,
|
2862 |
+
"learning_rate": 4.204348662690014e-06,
|
2863 |
+
"loss": 0.8918,
|
2864 |
+
"step": 23800
|
2865 |
+
},
|
2866 |
+
{
|
2867 |
+
"epoch": 4.59,
|
2868 |
+
"learning_rate": 4.108139311141043e-06,
|
2869 |
+
"loss": 0.8876,
|
2870 |
+
"step": 23850
|
2871 |
+
},
|
2872 |
+
{
|
2873 |
+
"epoch": 4.6,
|
2874 |
+
"learning_rate": 4.011929959592073e-06,
|
2875 |
+
"loss": 0.8835,
|
2876 |
+
"step": 23900
|
2877 |
+
},
|
2878 |
+
{
|
2879 |
+
"epoch": 4.61,
|
2880 |
+
"learning_rate": 3.915720608043102e-06,
|
2881 |
+
"loss": 0.8492,
|
2882 |
+
"step": 23950
|
2883 |
+
},
|
2884 |
+
{
|
2885 |
+
"epoch": 4.62,
|
2886 |
+
"learning_rate": 3.819511256494132e-06,
|
2887 |
+
"loss": 0.8681,
|
2888 |
+
"step": 24000
|
2889 |
+
},
|
2890 |
+
{
|
2891 |
+
"epoch": 4.63,
|
2892 |
+
"learning_rate": 3.723301904945161e-06,
|
2893 |
+
"loss": 0.825,
|
2894 |
+
"step": 24050
|
2895 |
+
},
|
2896 |
+
{
|
2897 |
+
"epoch": 4.64,
|
2898 |
+
"learning_rate": 3.62709255339619e-06,
|
2899 |
+
"loss": 0.9186,
|
2900 |
+
"step": 24100
|
2901 |
+
},
|
2902 |
+
{
|
2903 |
+
"epoch": 4.65,
|
2904 |
+
"learning_rate": 3.5308832018472196e-06,
|
2905 |
+
"loss": 0.8303,
|
2906 |
+
"step": 24150
|
2907 |
+
},
|
2908 |
+
{
|
2909 |
+
"epoch": 4.66,
|
2910 |
+
"learning_rate": 3.4346738502982493e-06,
|
2911 |
+
"loss": 0.8478,
|
2912 |
+
"step": 24200
|
2913 |
+
},
|
2914 |
+
{
|
2915 |
+
"epoch": 4.67,
|
2916 |
+
"learning_rate": 3.338464498749279e-06,
|
2917 |
+
"loss": 0.9059,
|
2918 |
+
"step": 24250
|
2919 |
+
},
|
2920 |
+
{
|
2921 |
+
"epoch": 4.68,
|
2922 |
+
"learning_rate": 3.242255147200308e-06,
|
2923 |
+
"loss": 0.8887,
|
2924 |
+
"step": 24300
|
2925 |
+
},
|
2926 |
+
{
|
2927 |
+
"epoch": 4.69,
|
2928 |
+
"learning_rate": 3.1460457956513372e-06,
|
2929 |
+
"loss": 0.8936,
|
2930 |
+
"step": 24350
|
2931 |
+
},
|
2932 |
+
{
|
2933 |
+
"epoch": 4.7,
|
2934 |
+
"learning_rate": 3.049836444102367e-06,
|
2935 |
+
"loss": 0.8862,
|
2936 |
+
"step": 24400
|
2937 |
+
},
|
2938 |
+
{
|
2939 |
+
"epoch": 4.7,
|
2940 |
+
"learning_rate": 2.9536270925533963e-06,
|
2941 |
+
"loss": 0.8715,
|
2942 |
+
"step": 24450
|
2943 |
+
},
|
2944 |
+
{
|
2945 |
+
"epoch": 4.71,
|
2946 |
+
"learning_rate": 2.8574177410044256e-06,
|
2947 |
+
"loss": 0.8568,
|
2948 |
+
"step": 24500
|
2949 |
+
},
|
2950 |
+
{
|
2951 |
+
"epoch": 4.72,
|
2952 |
+
"learning_rate": 2.7612083894554553e-06,
|
2953 |
+
"loss": 0.8931,
|
2954 |
+
"step": 24550
|
2955 |
+
},
|
2956 |
+
{
|
2957 |
+
"epoch": 4.73,
|
2958 |
+
"learning_rate": 2.6649990379064846e-06,
|
2959 |
+
"loss": 0.901,
|
2960 |
+
"step": 24600
|
2961 |
+
},
|
2962 |
+
{
|
2963 |
+
"epoch": 4.74,
|
2964 |
+
"learning_rate": 2.5687896863575143e-06,
|
2965 |
+
"loss": 0.8796,
|
2966 |
+
"step": 24650
|
2967 |
+
},
|
2968 |
+
{
|
2969 |
+
"epoch": 4.75,
|
2970 |
+
"learning_rate": 2.4725803348085436e-06,
|
2971 |
+
"loss": 0.8116,
|
2972 |
+
"step": 24700
|
2973 |
+
},
|
2974 |
+
{
|
2975 |
+
"epoch": 4.76,
|
2976 |
+
"learning_rate": 2.376370983259573e-06,
|
2977 |
+
"loss": 0.8417,
|
2978 |
+
"step": 24750
|
2979 |
+
},
|
2980 |
+
{
|
2981 |
+
"epoch": 4.77,
|
2982 |
+
"learning_rate": 2.2801616317106022e-06,
|
2983 |
+
"loss": 0.8734,
|
2984 |
+
"step": 24800
|
2985 |
+
},
|
2986 |
+
{
|
2987 |
+
"epoch": 4.78,
|
2988 |
+
"learning_rate": 2.183952280161632e-06,
|
2989 |
+
"loss": 0.8653,
|
2990 |
+
"step": 24850
|
2991 |
+
},
|
2992 |
+
{
|
2993 |
+
"epoch": 4.79,
|
2994 |
+
"learning_rate": 2.0877429286126613e-06,
|
2995 |
+
"loss": 0.8758,
|
2996 |
+
"step": 24900
|
2997 |
+
},
|
2998 |
+
{
|
2999 |
+
"epoch": 4.8,
|
3000 |
+
"learning_rate": 1.9915335770636906e-06,
|
3001 |
+
"loss": 0.9223,
|
3002 |
+
"step": 24950
|
3003 |
+
},
|
3004 |
+
{
|
3005 |
+
"epoch": 4.81,
|
3006 |
+
"learning_rate": 1.89532422551472e-06,
|
3007 |
+
"loss": 0.8926,
|
3008 |
+
"step": 25000
|
3009 |
+
},
|
3010 |
+
{
|
3011 |
+
"epoch": 4.82,
|
3012 |
+
"learning_rate": 1.7991148739657494e-06,
|
3013 |
+
"loss": 0.9047,
|
3014 |
+
"step": 25050
|
3015 |
+
},
|
3016 |
+
{
|
3017 |
+
"epoch": 4.83,
|
3018 |
+
"learning_rate": 1.7029055224167791e-06,
|
3019 |
+
"loss": 0.853,
|
3020 |
+
"step": 25100
|
3021 |
+
},
|
3022 |
+
{
|
3023 |
+
"epoch": 4.84,
|
3024 |
+
"learning_rate": 1.6066961708678084e-06,
|
3025 |
+
"loss": 0.8797,
|
3026 |
+
"step": 25150
|
3027 |
+
},
|
3028 |
+
{
|
3029 |
+
"epoch": 4.85,
|
3030 |
+
"learning_rate": 1.5104868193188377e-06,
|
3031 |
+
"loss": 0.8212,
|
3032 |
+
"step": 25200
|
3033 |
+
},
|
3034 |
+
{
|
3035 |
+
"epoch": 4.86,
|
3036 |
+
"learning_rate": 1.4142774677698672e-06,
|
3037 |
+
"loss": 0.8769,
|
3038 |
+
"step": 25250
|
3039 |
+
},
|
3040 |
+
{
|
3041 |
+
"epoch": 4.87,
|
3042 |
+
"learning_rate": 1.3180681162208968e-06,
|
3043 |
+
"loss": 0.8531,
|
3044 |
+
"step": 25300
|
3045 |
+
},
|
3046 |
+
{
|
3047 |
+
"epoch": 4.88,
|
3048 |
+
"learning_rate": 1.221858764671926e-06,
|
3049 |
+
"loss": 0.8877,
|
3050 |
+
"step": 25350
|
3051 |
+
},
|
3052 |
+
{
|
3053 |
+
"epoch": 4.89,
|
3054 |
+
"learning_rate": 1.1256494131229556e-06,
|
3055 |
+
"loss": 0.8349,
|
3056 |
+
"step": 25400
|
3057 |
+
},
|
3058 |
+
{
|
3059 |
+
"epoch": 4.9,
|
3060 |
+
"learning_rate": 1.029440061573985e-06,
|
3061 |
+
"loss": 0.867,
|
3062 |
+
"step": 25450
|
3063 |
+
},
|
3064 |
+
{
|
3065 |
+
"epoch": 4.91,
|
3066 |
+
"learning_rate": 9.332307100250145e-07,
|
3067 |
+
"loss": 0.8496,
|
3068 |
+
"step": 25500
|
3069 |
+
},
|
3070 |
+
{
|
3071 |
+
"epoch": 4.92,
|
3072 |
+
"learning_rate": 8.37021358476044e-07,
|
3073 |
+
"loss": 0.8753,
|
3074 |
+
"step": 25550
|
3075 |
+
},
|
3076 |
+
{
|
3077 |
+
"epoch": 4.93,
|
3078 |
+
"learning_rate": 7.408120069270733e-07,
|
3079 |
+
"loss": 0.8526,
|
3080 |
+
"step": 25600
|
3081 |
+
},
|
3082 |
+
{
|
3083 |
+
"epoch": 4.94,
|
3084 |
+
"learning_rate": 6.446026553781028e-07,
|
3085 |
+
"loss": 0.8247,
|
3086 |
+
"step": 25650
|
3087 |
+
},
|
3088 |
+
{
|
3089 |
+
"epoch": 4.95,
|
3090 |
+
"learning_rate": 5.483933038291322e-07,
|
3091 |
+
"loss": 0.8471,
|
3092 |
+
"step": 25700
|
3093 |
+
},
|
3094 |
+
{
|
3095 |
+
"epoch": 4.95,
|
3096 |
+
"learning_rate": 4.521839522801616e-07,
|
3097 |
+
"loss": 0.8819,
|
3098 |
+
"step": 25750
|
3099 |
+
},
|
3100 |
+
{
|
3101 |
+
"epoch": 4.96,
|
3102 |
+
"learning_rate": 3.5597460073119107e-07,
|
3103 |
+
"loss": 0.8805,
|
3104 |
+
"step": 25800
|
3105 |
+
},
|
3106 |
+
{
|
3107 |
+
"epoch": 4.97,
|
3108 |
+
"learning_rate": 2.5976524918222053e-07,
|
3109 |
+
"loss": 0.8706,
|
3110 |
+
"step": 25850
|
3111 |
+
},
|
3112 |
+
{
|
3113 |
+
"epoch": 4.98,
|
3114 |
+
"learning_rate": 1.6355589763324997e-07,
|
3115 |
+
"loss": 0.9151,
|
3116 |
+
"step": 25900
|
3117 |
+
},
|
3118 |
+
{
|
3119 |
+
"epoch": 4.99,
|
3120 |
+
"learning_rate": 6.734654608427939e-08,
|
3121 |
+
"loss": 0.8774,
|
3122 |
+
"step": 25950
|
3123 |
+
},
|
3124 |
+
{
|
3125 |
+
"epoch": 5.0,
|
3126 |
+
"step": 25985,
|
3127 |
+
"train_runtime": 11347.9361,
|
3128 |
+
"train_samples_per_second": 2.29
|
3129 |
+
}
|
3130 |
+
],
|
3131 |
+
"max_steps": 25985,
|
3132 |
+
"num_train_epochs": 5,
|
3133 |
+
"total_flos": 77899961943982080,
|
3134 |
+
"trial_name": null,
|
3135 |
+
"trial_params": null
|
3136 |
+
}
|
training_args.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:b97617bbafb16ff82fc4b50376f357ee4b93fb6257e0e64fb9677ff5cdf17267
|
3 |
+
size 1967
|
vocab.json
ADDED
The diff for this file is too large to render.
See raw diff
|
|