Training in progress, epoch 0, checkpoint
Browse files- checkpoint-305280/config.json +24 -0
- checkpoint-305280/optimizer.pt +3 -0
- checkpoint-305280/pytorch_model.bin +3 -0
- checkpoint-305280/rng_state.pth +3 -0
- checkpoint-305280/scheduler.pt +3 -0
- checkpoint-305280/special_tokens_map.json +5 -0
- checkpoint-305280/tokenizer.json +0 -0
- checkpoint-305280/tokenizer_config.json +12 -0
- checkpoint-305280/trainer_state.json +3633 -0
- checkpoint-305280/training_args.bin +3 -0
checkpoint-305280/config.json
ADDED
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"architectures": [
|
3 |
+
"RNNForLanguageModeling"
|
4 |
+
],
|
5 |
+
"bidirectional": false,
|
6 |
+
"dropout_p": 0.1,
|
7 |
+
"emb_init_range": 0.1,
|
8 |
+
"embedding_dim": 1024,
|
9 |
+
"embedding_kwargs": {},
|
10 |
+
"hidden_dim": 1024,
|
11 |
+
"lin_init_range": 0.03125,
|
12 |
+
"lm_in_features": 1024,
|
13 |
+
"model_type": "rnn",
|
14 |
+
"num_layers": 2,
|
15 |
+
"output_last_state": false,
|
16 |
+
"output_recurrent_outputs": false,
|
17 |
+
"recur_init_range": 0.03125,
|
18 |
+
"rnn_kwargs": {},
|
19 |
+
"rnn_type": "LSTM",
|
20 |
+
"tie_weights": true,
|
21 |
+
"torch_dtype": "float32",
|
22 |
+
"transformers_version": "4.33.3",
|
23 |
+
"vocab_size": 50002
|
24 |
+
}
|
checkpoint-305280/optimizer.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:986064cf31b1d00833f66eb2e81ee8692728a53f64681c548bafff811b961be5
|
3 |
+
size 544373707
|
checkpoint-305280/pytorch_model.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:2040bf1cc9f06760ba58e30c7333286c8b213019fb22e4777de751cf5b82251b
|
3 |
+
size 272184705
|
checkpoint-305280/rng_state.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:6a8a662307466c00fe5427571e2c77d34796c497ec739717cb1bc68d44456dae
|
3 |
+
size 14575
|
checkpoint-305280/scheduler.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:5b6d659e990f2800c0e98e0bff5f43853a535555c6808cc14c8e03de57e51963
|
3 |
+
size 627
|
checkpoint-305280/special_tokens_map.json
ADDED
@@ -0,0 +1,5 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"eos_token": "<eos>",
|
3 |
+
"pad_token": "<pad>",
|
4 |
+
"unk_token": "<unk>"
|
5 |
+
}
|
checkpoint-305280/tokenizer.json
ADDED
The diff for this file is too large to render.
See raw diff
|
|
checkpoint-305280/tokenizer_config.json
ADDED
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"clean_up_tokenization_spaces": true,
|
3 |
+
"eos_token": "<eos>",
|
4 |
+
"max_length": null,
|
5 |
+
"model_max_length": 512,
|
6 |
+
"pad_to_multiple_of": null,
|
7 |
+
"pad_token": "<pad>",
|
8 |
+
"pad_token_type_id": 0,
|
9 |
+
"padding_side": "right",
|
10 |
+
"tokenizer_class": "PreTrainedTokenizerFast",
|
11 |
+
"unk_token": "<unk>"
|
12 |
+
}
|
checkpoint-305280/trainer_state.json
ADDED
@@ -0,0 +1,3633 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"best_metric": 4.2855424880981445,
|
3 |
+
"best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/re-irr-sv-agr/lstm/4/checkpoints/checkpoint-305280",
|
4 |
+
"epoch": 0.025000606015738065,
|
5 |
+
"eval_steps": 10,
|
6 |
+
"global_step": 305280,
|
7 |
+
"is_hyper_param_search": false,
|
8 |
+
"is_local_process_zero": true,
|
9 |
+
"is_world_process_zero": true,
|
10 |
+
"log_history": [
|
11 |
+
{
|
12 |
+
"epoch": 0.0,
|
13 |
+
"learning_rate": 4.999998362119627e-05,
|
14 |
+
"loss": 10.82,
|
15 |
+
"step": 1
|
16 |
+
},
|
17 |
+
{
|
18 |
+
"epoch": 0.0,
|
19 |
+
"learning_rate": 4.999161405248948e-05,
|
20 |
+
"loss": 7.5473,
|
21 |
+
"step": 512
|
22 |
+
},
|
23 |
+
{
|
24 |
+
"epoch": 0.0,
|
25 |
+
"learning_rate": 4.998322810497896e-05,
|
26 |
+
"loss": 7.0522,
|
27 |
+
"step": 1024
|
28 |
+
},
|
29 |
+
{
|
30 |
+
"epoch": 0.0,
|
31 |
+
"learning_rate": 4.997484215746844e-05,
|
32 |
+
"loss": 6.9852,
|
33 |
+
"step": 1536
|
34 |
+
},
|
35 |
+
{
|
36 |
+
"epoch": 0.0,
|
37 |
+
"learning_rate": 4.996645620995792e-05,
|
38 |
+
"loss": 6.9385,
|
39 |
+
"step": 2048
|
40 |
+
},
|
41 |
+
{
|
42 |
+
"epoch": 0.0,
|
43 |
+
"learning_rate": 4.99580702624474e-05,
|
44 |
+
"loss": 6.8823,
|
45 |
+
"step": 2560
|
46 |
+
},
|
47 |
+
{
|
48 |
+
"epoch": 0.0,
|
49 |
+
"learning_rate": 4.994968431493688e-05,
|
50 |
+
"loss": 6.7148,
|
51 |
+
"step": 3072
|
52 |
+
},
|
53 |
+
{
|
54 |
+
"epoch": 0.0,
|
55 |
+
"learning_rate": 4.994131474623009e-05,
|
56 |
+
"loss": 6.6152,
|
57 |
+
"step": 3584
|
58 |
+
},
|
59 |
+
{
|
60 |
+
"epoch": 0.0,
|
61 |
+
"learning_rate": 4.993292879871958e-05,
|
62 |
+
"loss": 6.5142,
|
63 |
+
"step": 4096
|
64 |
+
},
|
65 |
+
{
|
66 |
+
"epoch": 0.0,
|
67 |
+
"learning_rate": 4.992454285120906e-05,
|
68 |
+
"loss": 6.4434,
|
69 |
+
"step": 4608
|
70 |
+
},
|
71 |
+
{
|
72 |
+
"epoch": 0.0,
|
73 |
+
"learning_rate": 4.991615690369854e-05,
|
74 |
+
"loss": 6.3732,
|
75 |
+
"step": 5120
|
76 |
+
},
|
77 |
+
{
|
78 |
+
"epoch": 0.0,
|
79 |
+
"learning_rate": 4.990777095618801e-05,
|
80 |
+
"loss": 6.3151,
|
81 |
+
"step": 5632
|
82 |
+
},
|
83 |
+
{
|
84 |
+
"epoch": 0.0,
|
85 |
+
"learning_rate": 4.989938500867749e-05,
|
86 |
+
"loss": 6.2482,
|
87 |
+
"step": 6144
|
88 |
+
},
|
89 |
+
{
|
90 |
+
"epoch": 0.0,
|
91 |
+
"learning_rate": 4.989099906116697e-05,
|
92 |
+
"loss": 6.1796,
|
93 |
+
"step": 6656
|
94 |
+
},
|
95 |
+
{
|
96 |
+
"epoch": 0.0,
|
97 |
+
"learning_rate": 4.988261311365645e-05,
|
98 |
+
"loss": 6.1237,
|
99 |
+
"step": 7168
|
100 |
+
},
|
101 |
+
{
|
102 |
+
"epoch": 0.0,
|
103 |
+
"learning_rate": 4.987422716614593e-05,
|
104 |
+
"loss": 6.0564,
|
105 |
+
"step": 7680
|
106 |
+
},
|
107 |
+
{
|
108 |
+
"epoch": 0.0,
|
109 |
+
"learning_rate": 4.986584121863541e-05,
|
110 |
+
"loss": 6.0163,
|
111 |
+
"step": 8192
|
112 |
+
},
|
113 |
+
{
|
114 |
+
"epoch": 0.0,
|
115 |
+
"learning_rate": 4.985745527112489e-05,
|
116 |
+
"loss": 5.975,
|
117 |
+
"step": 8704
|
118 |
+
},
|
119 |
+
{
|
120 |
+
"epoch": 0.0,
|
121 |
+
"learning_rate": 4.984906932361437e-05,
|
122 |
+
"loss": 5.9339,
|
123 |
+
"step": 9216
|
124 |
+
},
|
125 |
+
{
|
126 |
+
"epoch": 0.0,
|
127 |
+
"learning_rate": 4.984068337610385e-05,
|
128 |
+
"loss": 5.8988,
|
129 |
+
"step": 9728
|
130 |
+
},
|
131 |
+
{
|
132 |
+
"epoch": 0.0,
|
133 |
+
"learning_rate": 4.983229742859333e-05,
|
134 |
+
"loss": 5.8686,
|
135 |
+
"step": 10240
|
136 |
+
},
|
137 |
+
{
|
138 |
+
"epoch": 0.0,
|
139 |
+
"learning_rate": 4.982391148108281e-05,
|
140 |
+
"loss": 5.8293,
|
141 |
+
"step": 10752
|
142 |
+
},
|
143 |
+
{
|
144 |
+
"epoch": 0.0,
|
145 |
+
"learning_rate": 4.981552553357229e-05,
|
146 |
+
"loss": 5.805,
|
147 |
+
"step": 11264
|
148 |
+
},
|
149 |
+
{
|
150 |
+
"epoch": 0.0,
|
151 |
+
"learning_rate": 4.980713958606178e-05,
|
152 |
+
"loss": 5.7631,
|
153 |
+
"step": 11776
|
154 |
+
},
|
155 |
+
{
|
156 |
+
"epoch": 0.0,
|
157 |
+
"learning_rate": 4.9798770017354986e-05,
|
158 |
+
"loss": 5.7504,
|
159 |
+
"step": 12288
|
160 |
+
},
|
161 |
+
{
|
162 |
+
"epoch": 0.0,
|
163 |
+
"learning_rate": 4.9790384069844466e-05,
|
164 |
+
"loss": 5.7189,
|
165 |
+
"step": 12800
|
166 |
+
},
|
167 |
+
{
|
168 |
+
"epoch": 0.0,
|
169 |
+
"learning_rate": 4.9781998122333946e-05,
|
170 |
+
"loss": 5.7002,
|
171 |
+
"step": 13312
|
172 |
+
},
|
173 |
+
{
|
174 |
+
"epoch": 0.0,
|
175 |
+
"learning_rate": 4.9773612174823426e-05,
|
176 |
+
"loss": 5.6721,
|
177 |
+
"step": 13824
|
178 |
+
},
|
179 |
+
{
|
180 |
+
"epoch": 0.0,
|
181 |
+
"learning_rate": 4.9765242606116635e-05,
|
182 |
+
"loss": 5.6404,
|
183 |
+
"step": 14336
|
184 |
+
},
|
185 |
+
{
|
186 |
+
"epoch": 0.0,
|
187 |
+
"learning_rate": 4.9756856658606115e-05,
|
188 |
+
"loss": 5.6248,
|
189 |
+
"step": 14848
|
190 |
+
},
|
191 |
+
{
|
192 |
+
"epoch": 0.01,
|
193 |
+
"learning_rate": 4.9748470711095595e-05,
|
194 |
+
"loss": 5.6,
|
195 |
+
"step": 15360
|
196 |
+
},
|
197 |
+
{
|
198 |
+
"epoch": 0.01,
|
199 |
+
"learning_rate": 4.974008476358507e-05,
|
200 |
+
"loss": 5.5899,
|
201 |
+
"step": 15872
|
202 |
+
},
|
203 |
+
{
|
204 |
+
"epoch": 0.01,
|
205 |
+
"learning_rate": 4.973169881607455e-05,
|
206 |
+
"loss": 5.5641,
|
207 |
+
"step": 16384
|
208 |
+
},
|
209 |
+
{
|
210 |
+
"epoch": 0.01,
|
211 |
+
"learning_rate": 4.9723329247367764e-05,
|
212 |
+
"loss": 5.5525,
|
213 |
+
"step": 16896
|
214 |
+
},
|
215 |
+
{
|
216 |
+
"epoch": 0.01,
|
217 |
+
"learning_rate": 4.9714943299857244e-05,
|
218 |
+
"loss": 5.542,
|
219 |
+
"step": 17408
|
220 |
+
},
|
221 |
+
{
|
222 |
+
"epoch": 0.01,
|
223 |
+
"learning_rate": 4.970657373115046e-05,
|
224 |
+
"loss": 5.5196,
|
225 |
+
"step": 17920
|
226 |
+
},
|
227 |
+
{
|
228 |
+
"epoch": 0.01,
|
229 |
+
"learning_rate": 4.969818778363994e-05,
|
230 |
+
"loss": 5.5083,
|
231 |
+
"step": 18432
|
232 |
+
},
|
233 |
+
{
|
234 |
+
"epoch": 0.01,
|
235 |
+
"learning_rate": 4.968980183612942e-05,
|
236 |
+
"loss": 5.4655,
|
237 |
+
"step": 18944
|
238 |
+
},
|
239 |
+
{
|
240 |
+
"epoch": 0.01,
|
241 |
+
"learning_rate": 4.96814158886189e-05,
|
242 |
+
"loss": 5.462,
|
243 |
+
"step": 19456
|
244 |
+
},
|
245 |
+
{
|
246 |
+
"epoch": 0.01,
|
247 |
+
"learning_rate": 4.967302994110837e-05,
|
248 |
+
"loss": 5.4422,
|
249 |
+
"step": 19968
|
250 |
+
},
|
251 |
+
{
|
252 |
+
"epoch": 0.01,
|
253 |
+
"learning_rate": 4.966464399359785e-05,
|
254 |
+
"loss": 5.4265,
|
255 |
+
"step": 20480
|
256 |
+
},
|
257 |
+
{
|
258 |
+
"epoch": 0.01,
|
259 |
+
"learning_rate": 4.965625804608733e-05,
|
260 |
+
"loss": 5.4222,
|
261 |
+
"step": 20992
|
262 |
+
},
|
263 |
+
{
|
264 |
+
"epoch": 0.01,
|
265 |
+
"learning_rate": 4.964787209857681e-05,
|
266 |
+
"loss": 5.4158,
|
267 |
+
"step": 21504
|
268 |
+
},
|
269 |
+
{
|
270 |
+
"epoch": 0.01,
|
271 |
+
"learning_rate": 4.963950252987002e-05,
|
272 |
+
"loss": 5.3867,
|
273 |
+
"step": 22016
|
274 |
+
},
|
275 |
+
{
|
276 |
+
"epoch": 0.01,
|
277 |
+
"learning_rate": 4.96311165823595e-05,
|
278 |
+
"loss": 5.3863,
|
279 |
+
"step": 22528
|
280 |
+
},
|
281 |
+
{
|
282 |
+
"epoch": 0.01,
|
283 |
+
"learning_rate": 4.962273063484898e-05,
|
284 |
+
"loss": 5.3819,
|
285 |
+
"step": 23040
|
286 |
+
},
|
287 |
+
{
|
288 |
+
"epoch": 0.01,
|
289 |
+
"learning_rate": 4.961434468733847e-05,
|
290 |
+
"loss": 5.3691,
|
291 |
+
"step": 23552
|
292 |
+
},
|
293 |
+
{
|
294 |
+
"epoch": 0.01,
|
295 |
+
"learning_rate": 4.960595873982795e-05,
|
296 |
+
"loss": 5.3632,
|
297 |
+
"step": 24064
|
298 |
+
},
|
299 |
+
{
|
300 |
+
"epoch": 0.01,
|
301 |
+
"learning_rate": 4.959757279231743e-05,
|
302 |
+
"loss": 5.3471,
|
303 |
+
"step": 24576
|
304 |
+
},
|
305 |
+
{
|
306 |
+
"epoch": 0.01,
|
307 |
+
"learning_rate": 4.958918684480691e-05,
|
308 |
+
"loss": 5.3251,
|
309 |
+
"step": 25088
|
310 |
+
},
|
311 |
+
{
|
312 |
+
"epoch": 0.01,
|
313 |
+
"learning_rate": 4.958080089729639e-05,
|
314 |
+
"loss": 5.3285,
|
315 |
+
"step": 25600
|
316 |
+
},
|
317 |
+
{
|
318 |
+
"epoch": 0.01,
|
319 |
+
"learning_rate": 4.957241494978587e-05,
|
320 |
+
"loss": 5.3084,
|
321 |
+
"step": 26112
|
322 |
+
},
|
323 |
+
{
|
324 |
+
"epoch": 0.01,
|
325 |
+
"learning_rate": 4.956402900227535e-05,
|
326 |
+
"loss": 5.3114,
|
327 |
+
"step": 26624
|
328 |
+
},
|
329 |
+
{
|
330 |
+
"epoch": 0.01,
|
331 |
+
"learning_rate": 4.955564305476483e-05,
|
332 |
+
"loss": 5.2796,
|
333 |
+
"step": 27136
|
334 |
+
},
|
335 |
+
{
|
336 |
+
"epoch": 0.01,
|
337 |
+
"learning_rate": 4.954725710725431e-05,
|
338 |
+
"loss": 5.2933,
|
339 |
+
"step": 27648
|
340 |
+
},
|
341 |
+
{
|
342 |
+
"epoch": 0.01,
|
343 |
+
"learning_rate": 4.953888753854752e-05,
|
344 |
+
"loss": 5.2666,
|
345 |
+
"step": 28160
|
346 |
+
},
|
347 |
+
{
|
348 |
+
"epoch": 0.01,
|
349 |
+
"learning_rate": 4.9530517969840727e-05,
|
350 |
+
"loss": 5.2844,
|
351 |
+
"step": 28672
|
352 |
+
},
|
353 |
+
{
|
354 |
+
"epoch": 0.01,
|
355 |
+
"learning_rate": 4.9522132022330207e-05,
|
356 |
+
"loss": 5.256,
|
357 |
+
"step": 29184
|
358 |
+
},
|
359 |
+
{
|
360 |
+
"epoch": 0.01,
|
361 |
+
"learning_rate": 4.9513746074819686e-05,
|
362 |
+
"loss": 5.2434,
|
363 |
+
"step": 29696
|
364 |
+
},
|
365 |
+
{
|
366 |
+
"epoch": 0.01,
|
367 |
+
"learning_rate": 4.9505360127309166e-05,
|
368 |
+
"loss": 5.2383,
|
369 |
+
"step": 30208
|
370 |
+
},
|
371 |
+
{
|
372 |
+
"epoch": 0.01,
|
373 |
+
"learning_rate": 4.949699055860238e-05,
|
374 |
+
"loss": 5.2334,
|
375 |
+
"step": 30720
|
376 |
+
},
|
377 |
+
{
|
378 |
+
"epoch": 0.01,
|
379 |
+
"learning_rate": 4.948860461109186e-05,
|
380 |
+
"loss": 5.2186,
|
381 |
+
"step": 31232
|
382 |
+
},
|
383 |
+
{
|
384 |
+
"epoch": 0.01,
|
385 |
+
"learning_rate": 4.948021866358134e-05,
|
386 |
+
"loss": 5.2216,
|
387 |
+
"step": 31744
|
388 |
+
},
|
389 |
+
{
|
390 |
+
"epoch": 0.01,
|
391 |
+
"learning_rate": 4.947183271607082e-05,
|
392 |
+
"loss": 5.1928,
|
393 |
+
"step": 32256
|
394 |
+
},
|
395 |
+
{
|
396 |
+
"epoch": 0.01,
|
397 |
+
"learning_rate": 4.94634467685603e-05,
|
398 |
+
"loss": 5.1905,
|
399 |
+
"step": 32768
|
400 |
+
},
|
401 |
+
{
|
402 |
+
"epoch": 0.01,
|
403 |
+
"learning_rate": 4.945507719985351e-05,
|
404 |
+
"loss": 5.2017,
|
405 |
+
"step": 33280
|
406 |
+
},
|
407 |
+
{
|
408 |
+
"epoch": 0.01,
|
409 |
+
"learning_rate": 4.944669125234299e-05,
|
410 |
+
"loss": 5.1932,
|
411 |
+
"step": 33792
|
412 |
+
},
|
413 |
+
{
|
414 |
+
"epoch": 0.01,
|
415 |
+
"learning_rate": 4.943830530483247e-05,
|
416 |
+
"loss": 5.1767,
|
417 |
+
"step": 34304
|
418 |
+
},
|
419 |
+
{
|
420 |
+
"epoch": 0.01,
|
421 |
+
"learning_rate": 4.942991935732195e-05,
|
422 |
+
"loss": 5.1594,
|
423 |
+
"step": 34816
|
424 |
+
},
|
425 |
+
{
|
426 |
+
"epoch": 0.01,
|
427 |
+
"learning_rate": 4.942153340981143e-05,
|
428 |
+
"loss": 5.1598,
|
429 |
+
"step": 35328
|
430 |
+
},
|
431 |
+
{
|
432 |
+
"epoch": 0.01,
|
433 |
+
"learning_rate": 4.9413147462300904e-05,
|
434 |
+
"loss": 5.1606,
|
435 |
+
"step": 35840
|
436 |
+
},
|
437 |
+
{
|
438 |
+
"epoch": 0.01,
|
439 |
+
"learning_rate": 4.9404761514790384e-05,
|
440 |
+
"loss": 5.153,
|
441 |
+
"step": 36352
|
442 |
+
},
|
443 |
+
{
|
444 |
+
"epoch": 0.01,
|
445 |
+
"learning_rate": 4.939637556727987e-05,
|
446 |
+
"loss": 5.1494,
|
447 |
+
"step": 36864
|
448 |
+
},
|
449 |
+
{
|
450 |
+
"epoch": 0.01,
|
451 |
+
"learning_rate": 4.938800599857309e-05,
|
452 |
+
"loss": 5.1453,
|
453 |
+
"step": 37376
|
454 |
+
},
|
455 |
+
{
|
456 |
+
"epoch": 0.01,
|
457 |
+
"learning_rate": 4.937962005106256e-05,
|
458 |
+
"loss": 5.145,
|
459 |
+
"step": 37888
|
460 |
+
},
|
461 |
+
{
|
462 |
+
"epoch": 0.01,
|
463 |
+
"learning_rate": 4.937123410355204e-05,
|
464 |
+
"loss": 5.1365,
|
465 |
+
"step": 38400
|
466 |
+
},
|
467 |
+
{
|
468 |
+
"epoch": 0.01,
|
469 |
+
"learning_rate": 4.9362864534845256e-05,
|
470 |
+
"loss": 5.1112,
|
471 |
+
"step": 38912
|
472 |
+
},
|
473 |
+
{
|
474 |
+
"epoch": 0.01,
|
475 |
+
"learning_rate": 4.935447858733473e-05,
|
476 |
+
"loss": 5.1063,
|
477 |
+
"step": 39424
|
478 |
+
},
|
479 |
+
{
|
480 |
+
"epoch": 0.01,
|
481 |
+
"learning_rate": 4.934609263982421e-05,
|
482 |
+
"loss": 5.104,
|
483 |
+
"step": 39936
|
484 |
+
},
|
485 |
+
{
|
486 |
+
"epoch": 0.01,
|
487 |
+
"learning_rate": 4.933770669231369e-05,
|
488 |
+
"loss": 5.0925,
|
489 |
+
"step": 40448
|
490 |
+
},
|
491 |
+
{
|
492 |
+
"epoch": 0.01,
|
493 |
+
"learning_rate": 4.932932074480317e-05,
|
494 |
+
"loss": 5.1014,
|
495 |
+
"step": 40960
|
496 |
+
},
|
497 |
+
{
|
498 |
+
"epoch": 0.01,
|
499 |
+
"learning_rate": 4.932093479729265e-05,
|
500 |
+
"loss": 5.0837,
|
501 |
+
"step": 41472
|
502 |
+
},
|
503 |
+
{
|
504 |
+
"epoch": 0.01,
|
505 |
+
"learning_rate": 4.931256522858586e-05,
|
506 |
+
"loss": 5.0995,
|
507 |
+
"step": 41984
|
508 |
+
},
|
509 |
+
{
|
510 |
+
"epoch": 0.01,
|
511 |
+
"learning_rate": 4.930417928107534e-05,
|
512 |
+
"loss": 5.0745,
|
513 |
+
"step": 42496
|
514 |
+
},
|
515 |
+
{
|
516 |
+
"epoch": 0.01,
|
517 |
+
"learning_rate": 4.9295793333564825e-05,
|
518 |
+
"loss": 5.0608,
|
519 |
+
"step": 43008
|
520 |
+
},
|
521 |
+
{
|
522 |
+
"epoch": 0.01,
|
523 |
+
"learning_rate": 4.9287407386054305e-05,
|
524 |
+
"loss": 5.0643,
|
525 |
+
"step": 43520
|
526 |
+
},
|
527 |
+
{
|
528 |
+
"epoch": 0.01,
|
529 |
+
"learning_rate": 4.9279021438543785e-05,
|
530 |
+
"loss": 5.0646,
|
531 |
+
"step": 44032
|
532 |
+
},
|
533 |
+
{
|
534 |
+
"epoch": 0.01,
|
535 |
+
"learning_rate": 4.9270635491033264e-05,
|
536 |
+
"loss": 5.0604,
|
537 |
+
"step": 44544
|
538 |
+
},
|
539 |
+
{
|
540 |
+
"epoch": 0.01,
|
541 |
+
"learning_rate": 4.9262265922326474e-05,
|
542 |
+
"loss": 5.0487,
|
543 |
+
"step": 45056
|
544 |
+
},
|
545 |
+
{
|
546 |
+
"epoch": 0.01,
|
547 |
+
"learning_rate": 4.9253879974815954e-05,
|
548 |
+
"loss": 5.0434,
|
549 |
+
"step": 45568
|
550 |
+
},
|
551 |
+
{
|
552 |
+
"epoch": 0.02,
|
553 |
+
"learning_rate": 4.9245494027305433e-05,
|
554 |
+
"loss": 5.0357,
|
555 |
+
"step": 46080
|
556 |
+
},
|
557 |
+
{
|
558 |
+
"epoch": 0.02,
|
559 |
+
"learning_rate": 4.9237108079794913e-05,
|
560 |
+
"loss": 5.0342,
|
561 |
+
"step": 46592
|
562 |
+
},
|
563 |
+
{
|
564 |
+
"epoch": 0.02,
|
565 |
+
"learning_rate": 4.922872213228439e-05,
|
566 |
+
"loss": 5.0238,
|
567 |
+
"step": 47104
|
568 |
+
},
|
569 |
+
{
|
570 |
+
"epoch": 0.02,
|
571 |
+
"learning_rate": 4.92203525635776e-05,
|
572 |
+
"loss": 5.0192,
|
573 |
+
"step": 47616
|
574 |
+
},
|
575 |
+
{
|
576 |
+
"epoch": 0.02,
|
577 |
+
"learning_rate": 4.921196661606708e-05,
|
578 |
+
"loss": 5.0186,
|
579 |
+
"step": 48128
|
580 |
+
},
|
581 |
+
{
|
582 |
+
"epoch": 0.02,
|
583 |
+
"learning_rate": 4.920358066855656e-05,
|
584 |
+
"loss": 5.0238,
|
585 |
+
"step": 48640
|
586 |
+
},
|
587 |
+
{
|
588 |
+
"epoch": 0.02,
|
589 |
+
"learning_rate": 4.919519472104604e-05,
|
590 |
+
"loss": 5.0035,
|
591 |
+
"step": 49152
|
592 |
+
},
|
593 |
+
{
|
594 |
+
"epoch": 0.02,
|
595 |
+
"learning_rate": 4.918680877353552e-05,
|
596 |
+
"loss": 4.9966,
|
597 |
+
"step": 49664
|
598 |
+
},
|
599 |
+
{
|
600 |
+
"epoch": 0.02,
|
601 |
+
"learning_rate": 4.917842282602501e-05,
|
602 |
+
"loss": 4.9899,
|
603 |
+
"step": 50176
|
604 |
+
},
|
605 |
+
{
|
606 |
+
"epoch": 0.02,
|
607 |
+
"learning_rate": 4.917003687851449e-05,
|
608 |
+
"loss": 4.988,
|
609 |
+
"step": 50688
|
610 |
+
},
|
611 |
+
{
|
612 |
+
"epoch": 0.02,
|
613 |
+
"learning_rate": 4.91616673098077e-05,
|
614 |
+
"loss": 4.9917,
|
615 |
+
"step": 51200
|
616 |
+
},
|
617 |
+
{
|
618 |
+
"epoch": 0.02,
|
619 |
+
"learning_rate": 4.915328136229718e-05,
|
620 |
+
"loss": 4.9787,
|
621 |
+
"step": 51712
|
622 |
+
},
|
623 |
+
{
|
624 |
+
"epoch": 0.02,
|
625 |
+
"learning_rate": 4.914489541478666e-05,
|
626 |
+
"loss": 4.9719,
|
627 |
+
"step": 52224
|
628 |
+
},
|
629 |
+
{
|
630 |
+
"epoch": 0.02,
|
631 |
+
"learning_rate": 4.913650946727614e-05,
|
632 |
+
"loss": 4.9635,
|
633 |
+
"step": 52736
|
634 |
+
},
|
635 |
+
{
|
636 |
+
"epoch": 0.02,
|
637 |
+
"learning_rate": 4.912812351976562e-05,
|
638 |
+
"loss": 4.9665,
|
639 |
+
"step": 53248
|
640 |
+
},
|
641 |
+
{
|
642 |
+
"epoch": 0.02,
|
643 |
+
"learning_rate": 4.911975395105883e-05,
|
644 |
+
"loss": 4.9394,
|
645 |
+
"step": 53760
|
646 |
+
},
|
647 |
+
{
|
648 |
+
"epoch": 0.02,
|
649 |
+
"learning_rate": 4.911136800354831e-05,
|
650 |
+
"loss": 4.9543,
|
651 |
+
"step": 54272
|
652 |
+
},
|
653 |
+
{
|
654 |
+
"epoch": 0.02,
|
655 |
+
"learning_rate": 4.910298205603779e-05,
|
656 |
+
"loss": 4.941,
|
657 |
+
"step": 54784
|
658 |
+
},
|
659 |
+
{
|
660 |
+
"epoch": 0.02,
|
661 |
+
"learning_rate": 4.909459610852727e-05,
|
662 |
+
"loss": 4.9473,
|
663 |
+
"step": 55296
|
664 |
+
},
|
665 |
+
{
|
666 |
+
"epoch": 0.02,
|
667 |
+
"learning_rate": 4.908621016101675e-05,
|
668 |
+
"loss": 4.9354,
|
669 |
+
"step": 55808
|
670 |
+
},
|
671 |
+
{
|
672 |
+
"epoch": 0.02,
|
673 |
+
"learning_rate": 4.907782421350623e-05,
|
674 |
+
"loss": 4.9431,
|
675 |
+
"step": 56320
|
676 |
+
},
|
677 |
+
{
|
678 |
+
"epoch": 0.02,
|
679 |
+
"learning_rate": 4.906945464479944e-05,
|
680 |
+
"loss": 4.9294,
|
681 |
+
"step": 56832
|
682 |
+
},
|
683 |
+
{
|
684 |
+
"epoch": 0.02,
|
685 |
+
"learning_rate": 4.906106869728892e-05,
|
686 |
+
"loss": 4.9222,
|
687 |
+
"step": 57344
|
688 |
+
},
|
689 |
+
{
|
690 |
+
"epoch": 0.02,
|
691 |
+
"learning_rate": 4.9052682749778396e-05,
|
692 |
+
"loss": 4.9307,
|
693 |
+
"step": 57856
|
694 |
+
},
|
695 |
+
{
|
696 |
+
"epoch": 0.02,
|
697 |
+
"learning_rate": 4.9044296802267876e-05,
|
698 |
+
"loss": 4.9087,
|
699 |
+
"step": 58368
|
700 |
+
},
|
701 |
+
{
|
702 |
+
"epoch": 0.02,
|
703 |
+
"learning_rate": 4.9035910854757356e-05,
|
704 |
+
"loss": 4.925,
|
705 |
+
"step": 58880
|
706 |
+
},
|
707 |
+
{
|
708 |
+
"epoch": 0.02,
|
709 |
+
"learning_rate": 4.9027541286050565e-05,
|
710 |
+
"loss": 4.9147,
|
711 |
+
"step": 59392
|
712 |
+
},
|
713 |
+
{
|
714 |
+
"epoch": 0.02,
|
715 |
+
"learning_rate": 4.901917171734378e-05,
|
716 |
+
"loss": 4.9125,
|
717 |
+
"step": 59904
|
718 |
+
},
|
719 |
+
{
|
720 |
+
"epoch": 0.02,
|
721 |
+
"learning_rate": 4.901078576983326e-05,
|
722 |
+
"loss": 4.9,
|
723 |
+
"step": 60416
|
724 |
+
},
|
725 |
+
{
|
726 |
+
"epoch": 0.02,
|
727 |
+
"learning_rate": 4.900239982232274e-05,
|
728 |
+
"loss": 4.8989,
|
729 |
+
"step": 60928
|
730 |
+
},
|
731 |
+
{
|
732 |
+
"epoch": 0.02,
|
733 |
+
"learning_rate": 4.8994013874812214e-05,
|
734 |
+
"loss": 4.9007,
|
735 |
+
"step": 61440
|
736 |
+
},
|
737 |
+
{
|
738 |
+
"epoch": 0.02,
|
739 |
+
"learning_rate": 4.898564430610543e-05,
|
740 |
+
"loss": 4.8932,
|
741 |
+
"step": 61952
|
742 |
+
},
|
743 |
+
{
|
744 |
+
"epoch": 0.02,
|
745 |
+
"learning_rate": 4.897725835859491e-05,
|
746 |
+
"loss": 4.892,
|
747 |
+
"step": 62464
|
748 |
+
},
|
749 |
+
{
|
750 |
+
"epoch": 0.02,
|
751 |
+
"learning_rate": 4.8968872411084396e-05,
|
752 |
+
"loss": 4.881,
|
753 |
+
"step": 62976
|
754 |
+
},
|
755 |
+
{
|
756 |
+
"epoch": 0.02,
|
757 |
+
"learning_rate": 4.896048646357387e-05,
|
758 |
+
"loss": 4.8836,
|
759 |
+
"step": 63488
|
760 |
+
},
|
761 |
+
{
|
762 |
+
"epoch": 0.02,
|
763 |
+
"learning_rate": 4.895210051606335e-05,
|
764 |
+
"loss": 4.8642,
|
765 |
+
"step": 64000
|
766 |
+
},
|
767 |
+
{
|
768 |
+
"epoch": 0.02,
|
769 |
+
"learning_rate": 4.894371456855283e-05,
|
770 |
+
"loss": 4.8687,
|
771 |
+
"step": 64512
|
772 |
+
},
|
773 |
+
{
|
774 |
+
"epoch": 0.02,
|
775 |
+
"learning_rate": 4.893532862104231e-05,
|
776 |
+
"loss": 4.8521,
|
777 |
+
"step": 65024
|
778 |
+
},
|
779 |
+
{
|
780 |
+
"epoch": 0.02,
|
781 |
+
"learning_rate": 4.892694267353179e-05,
|
782 |
+
"loss": 4.8586,
|
783 |
+
"step": 65536
|
784 |
+
},
|
785 |
+
{
|
786 |
+
"epoch": 0.02,
|
787 |
+
"learning_rate": 4.8918573104825e-05,
|
788 |
+
"loss": 4.8453,
|
789 |
+
"step": 66048
|
790 |
+
},
|
791 |
+
{
|
792 |
+
"epoch": 0.02,
|
793 |
+
"learning_rate": 4.891018715731448e-05,
|
794 |
+
"loss": 4.8663,
|
795 |
+
"step": 66560
|
796 |
+
},
|
797 |
+
{
|
798 |
+
"epoch": 0.02,
|
799 |
+
"learning_rate": 4.890180120980396e-05,
|
800 |
+
"loss": 4.8714,
|
801 |
+
"step": 67072
|
802 |
+
},
|
803 |
+
{
|
804 |
+
"epoch": 0.02,
|
805 |
+
"learning_rate": 4.889341526229344e-05,
|
806 |
+
"loss": 4.8554,
|
807 |
+
"step": 67584
|
808 |
+
},
|
809 |
+
{
|
810 |
+
"epoch": 0.02,
|
811 |
+
"learning_rate": 4.888502931478292e-05,
|
812 |
+
"loss": 4.8347,
|
813 |
+
"step": 68096
|
814 |
+
},
|
815 |
+
{
|
816 |
+
"epoch": 0.02,
|
817 |
+
"learning_rate": 4.8876659746076134e-05,
|
818 |
+
"loss": 4.8453,
|
819 |
+
"step": 68608
|
820 |
+
},
|
821 |
+
{
|
822 |
+
"epoch": 0.02,
|
823 |
+
"learning_rate": 4.8868273798565614e-05,
|
824 |
+
"loss": 4.8437,
|
825 |
+
"step": 69120
|
826 |
+
},
|
827 |
+
{
|
828 |
+
"epoch": 0.02,
|
829 |
+
"learning_rate": 4.8859887851055094e-05,
|
830 |
+
"loss": 4.8376,
|
831 |
+
"step": 69632
|
832 |
+
},
|
833 |
+
{
|
834 |
+
"epoch": 0.02,
|
835 |
+
"learning_rate": 4.8851501903544574e-05,
|
836 |
+
"loss": 4.8323,
|
837 |
+
"step": 70144
|
838 |
+
},
|
839 |
+
{
|
840 |
+
"epoch": 0.02,
|
841 |
+
"learning_rate": 4.8843115956034054e-05,
|
842 |
+
"loss": 4.8346,
|
843 |
+
"step": 70656
|
844 |
+
},
|
845 |
+
{
|
846 |
+
"epoch": 0.02,
|
847 |
+
"learning_rate": 4.8834730008523534e-05,
|
848 |
+
"loss": 4.8301,
|
849 |
+
"step": 71168
|
850 |
+
},
|
851 |
+
{
|
852 |
+
"epoch": 0.02,
|
853 |
+
"learning_rate": 4.882637681862047e-05,
|
854 |
+
"loss": 4.8193,
|
855 |
+
"step": 71680
|
856 |
+
},
|
857 |
+
{
|
858 |
+
"epoch": 0.02,
|
859 |
+
"learning_rate": 4.881799087110995e-05,
|
860 |
+
"loss": 4.8205,
|
861 |
+
"step": 72192
|
862 |
+
},
|
863 |
+
{
|
864 |
+
"epoch": 0.02,
|
865 |
+
"learning_rate": 4.880960492359943e-05,
|
866 |
+
"loss": 4.8198,
|
867 |
+
"step": 72704
|
868 |
+
},
|
869 |
+
{
|
870 |
+
"epoch": 0.02,
|
871 |
+
"learning_rate": 4.880121897608891e-05,
|
872 |
+
"loss": 4.8089,
|
873 |
+
"step": 73216
|
874 |
+
},
|
875 |
+
{
|
876 |
+
"epoch": 0.02,
|
877 |
+
"learning_rate": 4.879283302857839e-05,
|
878 |
+
"loss": 4.8085,
|
879 |
+
"step": 73728
|
880 |
+
},
|
881 |
+
{
|
882 |
+
"epoch": 0.02,
|
883 |
+
"learning_rate": 4.878444708106787e-05,
|
884 |
+
"loss": 4.817,
|
885 |
+
"step": 74240
|
886 |
+
},
|
887 |
+
{
|
888 |
+
"epoch": 0.02,
|
889 |
+
"learning_rate": 4.877606113355735e-05,
|
890 |
+
"loss": 4.804,
|
891 |
+
"step": 74752
|
892 |
+
},
|
893 |
+
{
|
894 |
+
"epoch": 0.02,
|
895 |
+
"learning_rate": 4.876767518604683e-05,
|
896 |
+
"loss": 4.8023,
|
897 |
+
"step": 75264
|
898 |
+
},
|
899 |
+
{
|
900 |
+
"epoch": 0.02,
|
901 |
+
"learning_rate": 4.875928923853632e-05,
|
902 |
+
"loss": 4.8159,
|
903 |
+
"step": 75776
|
904 |
+
},
|
905 |
+
{
|
906 |
+
"epoch": 0.02,
|
907 |
+
"learning_rate": 4.87509032910258e-05,
|
908 |
+
"loss": 4.7916,
|
909 |
+
"step": 76288
|
910 |
+
},
|
911 |
+
{
|
912 |
+
"epoch": 0.03,
|
913 |
+
"eval_loss": 4.78051233291626,
|
914 |
+
"eval_runtime": 327.7198,
|
915 |
+
"eval_samples_per_second": 1164.382,
|
916 |
+
"eval_steps_per_second": 36.388,
|
917 |
+
"step": 76320
|
918 |
+
},
|
919 |
+
{
|
920 |
+
"epoch": 0.0,
|
921 |
+
"learning_rate": 4.874251734351528e-05,
|
922 |
+
"loss": 4.7776,
|
923 |
+
"step": 76800
|
924 |
+
},
|
925 |
+
{
|
926 |
+
"epoch": 0.0,
|
927 |
+
"learning_rate": 4.873413139600475e-05,
|
928 |
+
"loss": 4.7857,
|
929 |
+
"step": 77312
|
930 |
+
},
|
931 |
+
{
|
932 |
+
"epoch": 0.0,
|
933 |
+
"learning_rate": 4.872574544849423e-05,
|
934 |
+
"loss": 4.7992,
|
935 |
+
"step": 77824
|
936 |
+
},
|
937 |
+
{
|
938 |
+
"epoch": 0.0,
|
939 |
+
"learning_rate": 4.871735950098371e-05,
|
940 |
+
"loss": 4.7871,
|
941 |
+
"step": 78336
|
942 |
+
},
|
943 |
+
{
|
944 |
+
"epoch": 0.0,
|
945 |
+
"learning_rate": 4.870897355347319e-05,
|
946 |
+
"loss": 4.7927,
|
947 |
+
"step": 78848
|
948 |
+
},
|
949 |
+
{
|
950 |
+
"epoch": 0.0,
|
951 |
+
"learning_rate": 4.870058760596267e-05,
|
952 |
+
"loss": 4.768,
|
953 |
+
"step": 79360
|
954 |
+
},
|
955 |
+
{
|
956 |
+
"epoch": 0.0,
|
957 |
+
"learning_rate": 4.869220165845215e-05,
|
958 |
+
"loss": 4.7849,
|
959 |
+
"step": 79872
|
960 |
+
},
|
961 |
+
{
|
962 |
+
"epoch": 0.0,
|
963 |
+
"learning_rate": 4.868381571094163e-05,
|
964 |
+
"loss": 4.7561,
|
965 |
+
"step": 80384
|
966 |
+
},
|
967 |
+
{
|
968 |
+
"epoch": 0.0,
|
969 |
+
"learning_rate": 4.867542976343111e-05,
|
970 |
+
"loss": 4.7715,
|
971 |
+
"step": 80896
|
972 |
+
},
|
973 |
+
{
|
974 |
+
"epoch": 0.0,
|
975 |
+
"learning_rate": 4.866704381592059e-05,
|
976 |
+
"loss": 4.7592,
|
977 |
+
"step": 81408
|
978 |
+
},
|
979 |
+
{
|
980 |
+
"epoch": 0.0,
|
981 |
+
"learning_rate": 4.865865786841007e-05,
|
982 |
+
"loss": 4.7752,
|
983 |
+
"step": 81920
|
984 |
+
},
|
985 |
+
{
|
986 |
+
"epoch": 0.0,
|
987 |
+
"learning_rate": 4.865027192089955e-05,
|
988 |
+
"loss": 4.7705,
|
989 |
+
"step": 82432
|
990 |
+
},
|
991 |
+
{
|
992 |
+
"epoch": 0.0,
|
993 |
+
"learning_rate": 4.864190235219277e-05,
|
994 |
+
"loss": 4.757,
|
995 |
+
"step": 82944
|
996 |
+
},
|
997 |
+
{
|
998 |
+
"epoch": 0.0,
|
999 |
+
"learning_rate": 4.863351640468225e-05,
|
1000 |
+
"loss": 4.7516,
|
1001 |
+
"step": 83456
|
1002 |
+
},
|
1003 |
+
{
|
1004 |
+
"epoch": 0.0,
|
1005 |
+
"learning_rate": 4.862513045717173e-05,
|
1006 |
+
"loss": 4.7382,
|
1007 |
+
"step": 83968
|
1008 |
+
},
|
1009 |
+
{
|
1010 |
+
"epoch": 0.0,
|
1011 |
+
"learning_rate": 4.861674450966121e-05,
|
1012 |
+
"loss": 4.7402,
|
1013 |
+
"step": 84480
|
1014 |
+
},
|
1015 |
+
{
|
1016 |
+
"epoch": 0.0,
|
1017 |
+
"learning_rate": 4.860835856215069e-05,
|
1018 |
+
"loss": 4.7511,
|
1019 |
+
"step": 84992
|
1020 |
+
},
|
1021 |
+
{
|
1022 |
+
"epoch": 0.0,
|
1023 |
+
"learning_rate": 4.8599988993443896e-05,
|
1024 |
+
"loss": 4.7458,
|
1025 |
+
"step": 85504
|
1026 |
+
},
|
1027 |
+
{
|
1028 |
+
"epoch": 0.0,
|
1029 |
+
"learning_rate": 4.8591603045933376e-05,
|
1030 |
+
"loss": 4.7392,
|
1031 |
+
"step": 86016
|
1032 |
+
},
|
1033 |
+
{
|
1034 |
+
"epoch": 0.0,
|
1035 |
+
"learning_rate": 4.8583217098422856e-05,
|
1036 |
+
"loss": 4.7553,
|
1037 |
+
"step": 86528
|
1038 |
+
},
|
1039 |
+
{
|
1040 |
+
"epoch": 0.0,
|
1041 |
+
"learning_rate": 4.8574831150912336e-05,
|
1042 |
+
"loss": 4.7487,
|
1043 |
+
"step": 87040
|
1044 |
+
},
|
1045 |
+
{
|
1046 |
+
"epoch": 0.0,
|
1047 |
+
"learning_rate": 4.8566445203401816e-05,
|
1048 |
+
"loss": 4.7406,
|
1049 |
+
"step": 87552
|
1050 |
+
},
|
1051 |
+
{
|
1052 |
+
"epoch": 0.0,
|
1053 |
+
"learning_rate": 4.855805925589129e-05,
|
1054 |
+
"loss": 4.7285,
|
1055 |
+
"step": 88064
|
1056 |
+
},
|
1057 |
+
{
|
1058 |
+
"epoch": 0.0,
|
1059 |
+
"learning_rate": 4.854967330838077e-05,
|
1060 |
+
"loss": 4.7484,
|
1061 |
+
"step": 88576
|
1062 |
+
},
|
1063 |
+
{
|
1064 |
+
"epoch": 0.0,
|
1065 |
+
"learning_rate": 4.8541287360870256e-05,
|
1066 |
+
"loss": 4.7236,
|
1067 |
+
"step": 89088
|
1068 |
+
},
|
1069 |
+
{
|
1070 |
+
"epoch": 0.0,
|
1071 |
+
"learning_rate": 4.8532901413359736e-05,
|
1072 |
+
"loss": 4.7327,
|
1073 |
+
"step": 89600
|
1074 |
+
},
|
1075 |
+
{
|
1076 |
+
"epoch": 0.0,
|
1077 |
+
"learning_rate": 4.8524515465849216e-05,
|
1078 |
+
"loss": 4.7289,
|
1079 |
+
"step": 90112
|
1080 |
+
},
|
1081 |
+
{
|
1082 |
+
"epoch": 0.0,
|
1083 |
+
"learning_rate": 4.8516129518338696e-05,
|
1084 |
+
"loss": 4.723,
|
1085 |
+
"step": 90624
|
1086 |
+
},
|
1087 |
+
{
|
1088 |
+
"epoch": 0.0,
|
1089 |
+
"learning_rate": 4.8507759949631905e-05,
|
1090 |
+
"loss": 4.7116,
|
1091 |
+
"step": 91136
|
1092 |
+
},
|
1093 |
+
{
|
1094 |
+
"epoch": 0.01,
|
1095 |
+
"learning_rate": 4.8499374002121385e-05,
|
1096 |
+
"loss": 4.7129,
|
1097 |
+
"step": 91648
|
1098 |
+
},
|
1099 |
+
{
|
1100 |
+
"epoch": 0.01,
|
1101 |
+
"learning_rate": 4.8490988054610865e-05,
|
1102 |
+
"loss": 4.7155,
|
1103 |
+
"step": 92160
|
1104 |
+
},
|
1105 |
+
{
|
1106 |
+
"epoch": 0.01,
|
1107 |
+
"learning_rate": 4.8482602107100345e-05,
|
1108 |
+
"loss": 4.7109,
|
1109 |
+
"step": 92672
|
1110 |
+
},
|
1111 |
+
{
|
1112 |
+
"epoch": 0.01,
|
1113 |
+
"learning_rate": 4.8474216159589824e-05,
|
1114 |
+
"loss": 4.7111,
|
1115 |
+
"step": 93184
|
1116 |
+
},
|
1117 |
+
{
|
1118 |
+
"epoch": 0.01,
|
1119 |
+
"learning_rate": 4.8465846590883034e-05,
|
1120 |
+
"loss": 4.7176,
|
1121 |
+
"step": 93696
|
1122 |
+
},
|
1123 |
+
{
|
1124 |
+
"epoch": 0.01,
|
1125 |
+
"learning_rate": 4.8457460643372514e-05,
|
1126 |
+
"loss": 4.7082,
|
1127 |
+
"step": 94208
|
1128 |
+
},
|
1129 |
+
{
|
1130 |
+
"epoch": 0.01,
|
1131 |
+
"learning_rate": 4.8449074695861993e-05,
|
1132 |
+
"loss": 4.707,
|
1133 |
+
"step": 94720
|
1134 |
+
},
|
1135 |
+
{
|
1136 |
+
"epoch": 0.01,
|
1137 |
+
"learning_rate": 4.8440688748351473e-05,
|
1138 |
+
"loss": 4.6852,
|
1139 |
+
"step": 95232
|
1140 |
+
},
|
1141 |
+
{
|
1142 |
+
"epoch": 0.01,
|
1143 |
+
"learning_rate": 4.843230280084095e-05,
|
1144 |
+
"loss": 4.6927,
|
1145 |
+
"step": 95744
|
1146 |
+
},
|
1147 |
+
{
|
1148 |
+
"epoch": 0.01,
|
1149 |
+
"learning_rate": 4.842391685333043e-05,
|
1150 |
+
"loss": 4.683,
|
1151 |
+
"step": 96256
|
1152 |
+
},
|
1153 |
+
{
|
1154 |
+
"epoch": 0.01,
|
1155 |
+
"learning_rate": 4.841554728462365e-05,
|
1156 |
+
"loss": 4.6794,
|
1157 |
+
"step": 96768
|
1158 |
+
},
|
1159 |
+
{
|
1160 |
+
"epoch": 0.01,
|
1161 |
+
"learning_rate": 4.840716133711313e-05,
|
1162 |
+
"loss": 4.6897,
|
1163 |
+
"step": 97280
|
1164 |
+
},
|
1165 |
+
{
|
1166 |
+
"epoch": 0.01,
|
1167 |
+
"learning_rate": 4.839877538960261e-05,
|
1168 |
+
"loss": 4.692,
|
1169 |
+
"step": 97792
|
1170 |
+
},
|
1171 |
+
{
|
1172 |
+
"epoch": 0.01,
|
1173 |
+
"learning_rate": 4.839038944209209e-05,
|
1174 |
+
"loss": 4.6743,
|
1175 |
+
"step": 98304
|
1176 |
+
},
|
1177 |
+
{
|
1178 |
+
"epoch": 0.01,
|
1179 |
+
"learning_rate": 4.838200349458157e-05,
|
1180 |
+
"loss": 4.6823,
|
1181 |
+
"step": 98816
|
1182 |
+
},
|
1183 |
+
{
|
1184 |
+
"epoch": 0.01,
|
1185 |
+
"learning_rate": 4.837361754707105e-05,
|
1186 |
+
"loss": 4.6783,
|
1187 |
+
"step": 99328
|
1188 |
+
},
|
1189 |
+
{
|
1190 |
+
"epoch": 0.01,
|
1191 |
+
"learning_rate": 4.836523159956053e-05,
|
1192 |
+
"loss": 4.6822,
|
1193 |
+
"step": 99840
|
1194 |
+
},
|
1195 |
+
{
|
1196 |
+
"epoch": 0.01,
|
1197 |
+
"learning_rate": 4.835687840965747e-05,
|
1198 |
+
"loss": 4.6852,
|
1199 |
+
"step": 100352
|
1200 |
+
},
|
1201 |
+
{
|
1202 |
+
"epoch": 0.01,
|
1203 |
+
"learning_rate": 4.834849246214695e-05,
|
1204 |
+
"loss": 4.676,
|
1205 |
+
"step": 100864
|
1206 |
+
},
|
1207 |
+
{
|
1208 |
+
"epoch": 0.01,
|
1209 |
+
"learning_rate": 4.834010651463643e-05,
|
1210 |
+
"loss": 4.6651,
|
1211 |
+
"step": 101376
|
1212 |
+
},
|
1213 |
+
{
|
1214 |
+
"epoch": 0.01,
|
1215 |
+
"learning_rate": 4.833172056712591e-05,
|
1216 |
+
"loss": 4.6721,
|
1217 |
+
"step": 101888
|
1218 |
+
},
|
1219 |
+
{
|
1220 |
+
"epoch": 0.01,
|
1221 |
+
"learning_rate": 4.832333461961539e-05,
|
1222 |
+
"loss": 4.6605,
|
1223 |
+
"step": 102400
|
1224 |
+
},
|
1225 |
+
{
|
1226 |
+
"epoch": 0.01,
|
1227 |
+
"learning_rate": 4.8314948672104874e-05,
|
1228 |
+
"loss": 4.6692,
|
1229 |
+
"step": 102912
|
1230 |
+
},
|
1231 |
+
{
|
1232 |
+
"epoch": 0.01,
|
1233 |
+
"learning_rate": 4.8306562724594354e-05,
|
1234 |
+
"loss": 4.6561,
|
1235 |
+
"step": 103424
|
1236 |
+
},
|
1237 |
+
{
|
1238 |
+
"epoch": 0.01,
|
1239 |
+
"learning_rate": 4.829819315588756e-05,
|
1240 |
+
"loss": 4.6656,
|
1241 |
+
"step": 103936
|
1242 |
+
},
|
1243 |
+
{
|
1244 |
+
"epoch": 0.01,
|
1245 |
+
"learning_rate": 4.828980720837704e-05,
|
1246 |
+
"loss": 4.6456,
|
1247 |
+
"step": 104448
|
1248 |
+
},
|
1249 |
+
{
|
1250 |
+
"epoch": 0.01,
|
1251 |
+
"learning_rate": 4.828142126086652e-05,
|
1252 |
+
"loss": 4.6701,
|
1253 |
+
"step": 104960
|
1254 |
+
},
|
1255 |
+
{
|
1256 |
+
"epoch": 0.01,
|
1257 |
+
"learning_rate": 4.8273035313356e-05,
|
1258 |
+
"loss": 4.6512,
|
1259 |
+
"step": 105472
|
1260 |
+
},
|
1261 |
+
{
|
1262 |
+
"epoch": 0.01,
|
1263 |
+
"learning_rate": 4.8264649365845476e-05,
|
1264 |
+
"loss": 4.6475,
|
1265 |
+
"step": 105984
|
1266 |
+
},
|
1267 |
+
{
|
1268 |
+
"epoch": 0.01,
|
1269 |
+
"learning_rate": 4.8256263418334956e-05,
|
1270 |
+
"loss": 4.6545,
|
1271 |
+
"step": 106496
|
1272 |
+
},
|
1273 |
+
{
|
1274 |
+
"epoch": 0.01,
|
1275 |
+
"learning_rate": 4.824789384962817e-05,
|
1276 |
+
"loss": 4.6456,
|
1277 |
+
"step": 107008
|
1278 |
+
},
|
1279 |
+
{
|
1280 |
+
"epoch": 0.01,
|
1281 |
+
"learning_rate": 4.823950790211765e-05,
|
1282 |
+
"loss": 4.6386,
|
1283 |
+
"step": 107520
|
1284 |
+
},
|
1285 |
+
{
|
1286 |
+
"epoch": 0.01,
|
1287 |
+
"learning_rate": 4.8231121954607125e-05,
|
1288 |
+
"loss": 4.6526,
|
1289 |
+
"step": 108032
|
1290 |
+
},
|
1291 |
+
{
|
1292 |
+
"epoch": 0.01,
|
1293 |
+
"learning_rate": 4.822273600709661e-05,
|
1294 |
+
"loss": 4.6237,
|
1295 |
+
"step": 108544
|
1296 |
+
},
|
1297 |
+
{
|
1298 |
+
"epoch": 0.01,
|
1299 |
+
"learning_rate": 4.821435005958609e-05,
|
1300 |
+
"loss": 4.6294,
|
1301 |
+
"step": 109056
|
1302 |
+
},
|
1303 |
+
{
|
1304 |
+
"epoch": 0.01,
|
1305 |
+
"learning_rate": 4.820596411207557e-05,
|
1306 |
+
"loss": 4.6415,
|
1307 |
+
"step": 109568
|
1308 |
+
},
|
1309 |
+
{
|
1310 |
+
"epoch": 0.01,
|
1311 |
+
"learning_rate": 4.819757816456505e-05,
|
1312 |
+
"loss": 4.641,
|
1313 |
+
"step": 110080
|
1314 |
+
},
|
1315 |
+
{
|
1316 |
+
"epoch": 0.01,
|
1317 |
+
"learning_rate": 4.818919221705453e-05,
|
1318 |
+
"loss": 4.6324,
|
1319 |
+
"step": 110592
|
1320 |
+
},
|
1321 |
+
{
|
1322 |
+
"epoch": 0.01,
|
1323 |
+
"learning_rate": 4.818080626954401e-05,
|
1324 |
+
"loss": 4.6229,
|
1325 |
+
"step": 111104
|
1326 |
+
},
|
1327 |
+
{
|
1328 |
+
"epoch": 0.01,
|
1329 |
+
"learning_rate": 4.817242032203349e-05,
|
1330 |
+
"loss": 4.6226,
|
1331 |
+
"step": 111616
|
1332 |
+
},
|
1333 |
+
{
|
1334 |
+
"epoch": 0.01,
|
1335 |
+
"learning_rate": 4.816403437452297e-05,
|
1336 |
+
"loss": 4.6271,
|
1337 |
+
"step": 112128
|
1338 |
+
},
|
1339 |
+
{
|
1340 |
+
"epoch": 0.01,
|
1341 |
+
"learning_rate": 4.815566480581618e-05,
|
1342 |
+
"loss": 4.6328,
|
1343 |
+
"step": 112640
|
1344 |
+
},
|
1345 |
+
{
|
1346 |
+
"epoch": 0.01,
|
1347 |
+
"learning_rate": 4.814727885830566e-05,
|
1348 |
+
"loss": 4.6267,
|
1349 |
+
"step": 113152
|
1350 |
+
},
|
1351 |
+
{
|
1352 |
+
"epoch": 0.01,
|
1353 |
+
"learning_rate": 4.813889291079514e-05,
|
1354 |
+
"loss": 4.627,
|
1355 |
+
"step": 113664
|
1356 |
+
},
|
1357 |
+
{
|
1358 |
+
"epoch": 0.01,
|
1359 |
+
"learning_rate": 4.813052334208835e-05,
|
1360 |
+
"loss": 4.6331,
|
1361 |
+
"step": 114176
|
1362 |
+
},
|
1363 |
+
{
|
1364 |
+
"epoch": 0.01,
|
1365 |
+
"learning_rate": 4.812213739457783e-05,
|
1366 |
+
"loss": 4.63,
|
1367 |
+
"step": 114688
|
1368 |
+
},
|
1369 |
+
{
|
1370 |
+
"epoch": 0.01,
|
1371 |
+
"learning_rate": 4.811375144706731e-05,
|
1372 |
+
"loss": 4.613,
|
1373 |
+
"step": 115200
|
1374 |
+
},
|
1375 |
+
{
|
1376 |
+
"epoch": 0.01,
|
1377 |
+
"learning_rate": 4.8105365499556796e-05,
|
1378 |
+
"loss": 4.6108,
|
1379 |
+
"step": 115712
|
1380 |
+
},
|
1381 |
+
{
|
1382 |
+
"epoch": 0.01,
|
1383 |
+
"learning_rate": 4.8096979552046276e-05,
|
1384 |
+
"loss": 4.6192,
|
1385 |
+
"step": 116224
|
1386 |
+
},
|
1387 |
+
{
|
1388 |
+
"epoch": 0.01,
|
1389 |
+
"learning_rate": 4.8088609983339485e-05,
|
1390 |
+
"loss": 4.6036,
|
1391 |
+
"step": 116736
|
1392 |
+
},
|
1393 |
+
{
|
1394 |
+
"epoch": 0.01,
|
1395 |
+
"learning_rate": 4.8080224035828965e-05,
|
1396 |
+
"loss": 4.616,
|
1397 |
+
"step": 117248
|
1398 |
+
},
|
1399 |
+
{
|
1400 |
+
"epoch": 0.01,
|
1401 |
+
"learning_rate": 4.8071838088318445e-05,
|
1402 |
+
"loss": 4.6031,
|
1403 |
+
"step": 117760
|
1404 |
+
},
|
1405 |
+
{
|
1406 |
+
"epoch": 0.01,
|
1407 |
+
"learning_rate": 4.8063452140807925e-05,
|
1408 |
+
"loss": 4.6259,
|
1409 |
+
"step": 118272
|
1410 |
+
},
|
1411 |
+
{
|
1412 |
+
"epoch": 0.01,
|
1413 |
+
"learning_rate": 4.8055066193297405e-05,
|
1414 |
+
"loss": 4.6062,
|
1415 |
+
"step": 118784
|
1416 |
+
},
|
1417 |
+
{
|
1418 |
+
"epoch": 0.01,
|
1419 |
+
"learning_rate": 4.8046680245786885e-05,
|
1420 |
+
"loss": 4.591,
|
1421 |
+
"step": 119296
|
1422 |
+
},
|
1423 |
+
{
|
1424 |
+
"epoch": 0.01,
|
1425 |
+
"learning_rate": 4.8038294298276365e-05,
|
1426 |
+
"loss": 4.6016,
|
1427 |
+
"step": 119808
|
1428 |
+
},
|
1429 |
+
{
|
1430 |
+
"epoch": 0.01,
|
1431 |
+
"learning_rate": 4.802990835076584e-05,
|
1432 |
+
"loss": 4.6053,
|
1433 |
+
"step": 120320
|
1434 |
+
},
|
1435 |
+
{
|
1436 |
+
"epoch": 0.01,
|
1437 |
+
"learning_rate": 4.802152240325532e-05,
|
1438 |
+
"loss": 4.6077,
|
1439 |
+
"step": 120832
|
1440 |
+
},
|
1441 |
+
{
|
1442 |
+
"epoch": 0.01,
|
1443 |
+
"learning_rate": 4.8013152834548534e-05,
|
1444 |
+
"loss": 4.5984,
|
1445 |
+
"step": 121344
|
1446 |
+
},
|
1447 |
+
{
|
1448 |
+
"epoch": 0.01,
|
1449 |
+
"learning_rate": 4.8004766887038014e-05,
|
1450 |
+
"loss": 4.5972,
|
1451 |
+
"step": 121856
|
1452 |
+
},
|
1453 |
+
{
|
1454 |
+
"epoch": 0.02,
|
1455 |
+
"learning_rate": 4.7996380939527494e-05,
|
1456 |
+
"loss": 4.6018,
|
1457 |
+
"step": 122368
|
1458 |
+
},
|
1459 |
+
{
|
1460 |
+
"epoch": 0.02,
|
1461 |
+
"learning_rate": 4.7987994992016974e-05,
|
1462 |
+
"loss": 4.5901,
|
1463 |
+
"step": 122880
|
1464 |
+
},
|
1465 |
+
{
|
1466 |
+
"epoch": 0.02,
|
1467 |
+
"learning_rate": 4.797962542331019e-05,
|
1468 |
+
"loss": 4.5895,
|
1469 |
+
"step": 123392
|
1470 |
+
},
|
1471 |
+
{
|
1472 |
+
"epoch": 0.02,
|
1473 |
+
"learning_rate": 4.797123947579966e-05,
|
1474 |
+
"loss": 4.5898,
|
1475 |
+
"step": 123904
|
1476 |
+
},
|
1477 |
+
{
|
1478 |
+
"epoch": 0.02,
|
1479 |
+
"learning_rate": 4.796285352828914e-05,
|
1480 |
+
"loss": 4.5948,
|
1481 |
+
"step": 124416
|
1482 |
+
},
|
1483 |
+
{
|
1484 |
+
"epoch": 0.02,
|
1485 |
+
"learning_rate": 4.795446758077862e-05,
|
1486 |
+
"loss": 4.5942,
|
1487 |
+
"step": 124928
|
1488 |
+
},
|
1489 |
+
{
|
1490 |
+
"epoch": 0.02,
|
1491 |
+
"learning_rate": 4.79460816332681e-05,
|
1492 |
+
"loss": 4.5902,
|
1493 |
+
"step": 125440
|
1494 |
+
},
|
1495 |
+
{
|
1496 |
+
"epoch": 0.02,
|
1497 |
+
"learning_rate": 4.793769568575758e-05,
|
1498 |
+
"loss": 4.576,
|
1499 |
+
"step": 125952
|
1500 |
+
},
|
1501 |
+
{
|
1502 |
+
"epoch": 0.02,
|
1503 |
+
"learning_rate": 4.792930973824706e-05,
|
1504 |
+
"loss": 4.578,
|
1505 |
+
"step": 126464
|
1506 |
+
},
|
1507 |
+
{
|
1508 |
+
"epoch": 0.02,
|
1509 |
+
"learning_rate": 4.792092379073654e-05,
|
1510 |
+
"loss": 4.5817,
|
1511 |
+
"step": 126976
|
1512 |
+
},
|
1513 |
+
{
|
1514 |
+
"epoch": 0.02,
|
1515 |
+
"learning_rate": 4.791255422202975e-05,
|
1516 |
+
"loss": 4.5812,
|
1517 |
+
"step": 127488
|
1518 |
+
},
|
1519 |
+
{
|
1520 |
+
"epoch": 0.02,
|
1521 |
+
"learning_rate": 4.790416827451923e-05,
|
1522 |
+
"loss": 4.5773,
|
1523 |
+
"step": 128000
|
1524 |
+
},
|
1525 |
+
{
|
1526 |
+
"epoch": 0.02,
|
1527 |
+
"learning_rate": 4.789578232700871e-05,
|
1528 |
+
"loss": 4.5739,
|
1529 |
+
"step": 128512
|
1530 |
+
},
|
1531 |
+
{
|
1532 |
+
"epoch": 0.02,
|
1533 |
+
"learning_rate": 4.788741275830193e-05,
|
1534 |
+
"loss": 4.5662,
|
1535 |
+
"step": 129024
|
1536 |
+
},
|
1537 |
+
{
|
1538 |
+
"epoch": 0.02,
|
1539 |
+
"learning_rate": 4.787902681079141e-05,
|
1540 |
+
"loss": 4.5773,
|
1541 |
+
"step": 129536
|
1542 |
+
},
|
1543 |
+
{
|
1544 |
+
"epoch": 0.02,
|
1545 |
+
"learning_rate": 4.787064086328089e-05,
|
1546 |
+
"loss": 4.5465,
|
1547 |
+
"step": 130048
|
1548 |
+
},
|
1549 |
+
{
|
1550 |
+
"epoch": 0.02,
|
1551 |
+
"learning_rate": 4.786225491577037e-05,
|
1552 |
+
"loss": 4.5706,
|
1553 |
+
"step": 130560
|
1554 |
+
},
|
1555 |
+
{
|
1556 |
+
"epoch": 0.02,
|
1557 |
+
"learning_rate": 4.785386896825985e-05,
|
1558 |
+
"loss": 4.5619,
|
1559 |
+
"step": 131072
|
1560 |
+
},
|
1561 |
+
{
|
1562 |
+
"epoch": 0.02,
|
1563 |
+
"learning_rate": 4.784548302074933e-05,
|
1564 |
+
"loss": 4.5695,
|
1565 |
+
"step": 131584
|
1566 |
+
},
|
1567 |
+
{
|
1568 |
+
"epoch": 0.02,
|
1569 |
+
"learning_rate": 4.7837113452042536e-05,
|
1570 |
+
"loss": 4.5577,
|
1571 |
+
"step": 132096
|
1572 |
+
},
|
1573 |
+
{
|
1574 |
+
"epoch": 0.02,
|
1575 |
+
"learning_rate": 4.7828727504532016e-05,
|
1576 |
+
"loss": 4.5702,
|
1577 |
+
"step": 132608
|
1578 |
+
},
|
1579 |
+
{
|
1580 |
+
"epoch": 0.02,
|
1581 |
+
"learning_rate": 4.7820341557021496e-05,
|
1582 |
+
"loss": 4.5579,
|
1583 |
+
"step": 133120
|
1584 |
+
},
|
1585 |
+
{
|
1586 |
+
"epoch": 0.02,
|
1587 |
+
"learning_rate": 4.7811955609510976e-05,
|
1588 |
+
"loss": 4.5548,
|
1589 |
+
"step": 133632
|
1590 |
+
},
|
1591 |
+
{
|
1592 |
+
"epoch": 0.02,
|
1593 |
+
"learning_rate": 4.7803569662000456e-05,
|
1594 |
+
"loss": 4.5673,
|
1595 |
+
"step": 134144
|
1596 |
+
},
|
1597 |
+
{
|
1598 |
+
"epoch": 0.02,
|
1599 |
+
"learning_rate": 4.7795183714489936e-05,
|
1600 |
+
"loss": 4.5496,
|
1601 |
+
"step": 134656
|
1602 |
+
},
|
1603 |
+
{
|
1604 |
+
"epoch": 0.02,
|
1605 |
+
"learning_rate": 4.7786797766979416e-05,
|
1606 |
+
"loss": 4.5632,
|
1607 |
+
"step": 135168
|
1608 |
+
},
|
1609 |
+
{
|
1610 |
+
"epoch": 0.02,
|
1611 |
+
"learning_rate": 4.7778411819468896e-05,
|
1612 |
+
"loss": 4.5632,
|
1613 |
+
"step": 135680
|
1614 |
+
},
|
1615 |
+
{
|
1616 |
+
"epoch": 0.02,
|
1617 |
+
"learning_rate": 4.777005862956584e-05,
|
1618 |
+
"loss": 4.557,
|
1619 |
+
"step": 136192
|
1620 |
+
},
|
1621 |
+
{
|
1622 |
+
"epoch": 0.02,
|
1623 |
+
"learning_rate": 4.776167268205532e-05,
|
1624 |
+
"loss": 4.5481,
|
1625 |
+
"step": 136704
|
1626 |
+
},
|
1627 |
+
{
|
1628 |
+
"epoch": 0.02,
|
1629 |
+
"learning_rate": 4.77532867345448e-05,
|
1630 |
+
"loss": 4.5524,
|
1631 |
+
"step": 137216
|
1632 |
+
},
|
1633 |
+
{
|
1634 |
+
"epoch": 0.02,
|
1635 |
+
"learning_rate": 4.774490078703428e-05,
|
1636 |
+
"loss": 4.5562,
|
1637 |
+
"step": 137728
|
1638 |
+
},
|
1639 |
+
{
|
1640 |
+
"epoch": 0.02,
|
1641 |
+
"learning_rate": 4.773651483952376e-05,
|
1642 |
+
"loss": 4.5412,
|
1643 |
+
"step": 138240
|
1644 |
+
},
|
1645 |
+
{
|
1646 |
+
"epoch": 0.02,
|
1647 |
+
"learning_rate": 4.772814527081697e-05,
|
1648 |
+
"loss": 4.5578,
|
1649 |
+
"step": 138752
|
1650 |
+
},
|
1651 |
+
{
|
1652 |
+
"epoch": 0.02,
|
1653 |
+
"learning_rate": 4.771975932330645e-05,
|
1654 |
+
"loss": 4.5408,
|
1655 |
+
"step": 139264
|
1656 |
+
},
|
1657 |
+
{
|
1658 |
+
"epoch": 0.02,
|
1659 |
+
"learning_rate": 4.771137337579593e-05,
|
1660 |
+
"loss": 4.548,
|
1661 |
+
"step": 139776
|
1662 |
+
},
|
1663 |
+
{
|
1664 |
+
"epoch": 0.02,
|
1665 |
+
"learning_rate": 4.770298742828541e-05,
|
1666 |
+
"loss": 4.5406,
|
1667 |
+
"step": 140288
|
1668 |
+
},
|
1669 |
+
{
|
1670 |
+
"epoch": 0.02,
|
1671 |
+
"learning_rate": 4.769461785957862e-05,
|
1672 |
+
"loss": 4.5389,
|
1673 |
+
"step": 140800
|
1674 |
+
},
|
1675 |
+
{
|
1676 |
+
"epoch": 0.02,
|
1677 |
+
"learning_rate": 4.7686231912068105e-05,
|
1678 |
+
"loss": 4.5208,
|
1679 |
+
"step": 141312
|
1680 |
+
},
|
1681 |
+
{
|
1682 |
+
"epoch": 0.02,
|
1683 |
+
"learning_rate": 4.7677845964557585e-05,
|
1684 |
+
"loss": 4.5354,
|
1685 |
+
"step": 141824
|
1686 |
+
},
|
1687 |
+
{
|
1688 |
+
"epoch": 0.02,
|
1689 |
+
"learning_rate": 4.7669460017047065e-05,
|
1690 |
+
"loss": 4.5219,
|
1691 |
+
"step": 142336
|
1692 |
+
},
|
1693 |
+
{
|
1694 |
+
"epoch": 0.02,
|
1695 |
+
"learning_rate": 4.7661090448340274e-05,
|
1696 |
+
"loss": 4.5445,
|
1697 |
+
"step": 142848
|
1698 |
+
},
|
1699 |
+
{
|
1700 |
+
"epoch": 0.02,
|
1701 |
+
"learning_rate": 4.7652704500829754e-05,
|
1702 |
+
"loss": 4.5557,
|
1703 |
+
"step": 143360
|
1704 |
+
},
|
1705 |
+
{
|
1706 |
+
"epoch": 0.02,
|
1707 |
+
"learning_rate": 4.7644318553319234e-05,
|
1708 |
+
"loss": 4.5394,
|
1709 |
+
"step": 143872
|
1710 |
+
},
|
1711 |
+
{
|
1712 |
+
"epoch": 0.02,
|
1713 |
+
"learning_rate": 4.7635932605808714e-05,
|
1714 |
+
"loss": 4.5206,
|
1715 |
+
"step": 144384
|
1716 |
+
},
|
1717 |
+
{
|
1718 |
+
"epoch": 0.02,
|
1719 |
+
"learning_rate": 4.7627546658298194e-05,
|
1720 |
+
"loss": 4.535,
|
1721 |
+
"step": 144896
|
1722 |
+
},
|
1723 |
+
{
|
1724 |
+
"epoch": 0.02,
|
1725 |
+
"learning_rate": 4.7619160710787674e-05,
|
1726 |
+
"loss": 4.5288,
|
1727 |
+
"step": 145408
|
1728 |
+
},
|
1729 |
+
{
|
1730 |
+
"epoch": 0.02,
|
1731 |
+
"learning_rate": 4.761077476327715e-05,
|
1732 |
+
"loss": 4.532,
|
1733 |
+
"step": 145920
|
1734 |
+
},
|
1735 |
+
{
|
1736 |
+
"epoch": 0.02,
|
1737 |
+
"learning_rate": 4.760240519457036e-05,
|
1738 |
+
"loss": 4.5285,
|
1739 |
+
"step": 146432
|
1740 |
+
},
|
1741 |
+
{
|
1742 |
+
"epoch": 0.02,
|
1743 |
+
"learning_rate": 4.759401924705984e-05,
|
1744 |
+
"loss": 4.5286,
|
1745 |
+
"step": 146944
|
1746 |
+
},
|
1747 |
+
{
|
1748 |
+
"epoch": 0.02,
|
1749 |
+
"learning_rate": 4.758563329954932e-05,
|
1750 |
+
"loss": 4.5242,
|
1751 |
+
"step": 147456
|
1752 |
+
},
|
1753 |
+
{
|
1754 |
+
"epoch": 0.02,
|
1755 |
+
"learning_rate": 4.75772473520388e-05,
|
1756 |
+
"loss": 4.5249,
|
1757 |
+
"step": 147968
|
1758 |
+
},
|
1759 |
+
{
|
1760 |
+
"epoch": 0.02,
|
1761 |
+
"learning_rate": 4.756887778333202e-05,
|
1762 |
+
"loss": 4.5242,
|
1763 |
+
"step": 148480
|
1764 |
+
},
|
1765 |
+
{
|
1766 |
+
"epoch": 0.02,
|
1767 |
+
"learning_rate": 4.75604918358215e-05,
|
1768 |
+
"loss": 4.5268,
|
1769 |
+
"step": 148992
|
1770 |
+
},
|
1771 |
+
{
|
1772 |
+
"epoch": 0.02,
|
1773 |
+
"learning_rate": 4.755210588831097e-05,
|
1774 |
+
"loss": 4.5184,
|
1775 |
+
"step": 149504
|
1776 |
+
},
|
1777 |
+
{
|
1778 |
+
"epoch": 0.02,
|
1779 |
+
"learning_rate": 4.754371994080045e-05,
|
1780 |
+
"loss": 4.5138,
|
1781 |
+
"step": 150016
|
1782 |
+
},
|
1783 |
+
{
|
1784 |
+
"epoch": 0.02,
|
1785 |
+
"learning_rate": 4.753535037209367e-05,
|
1786 |
+
"loss": 4.528,
|
1787 |
+
"step": 150528
|
1788 |
+
},
|
1789 |
+
{
|
1790 |
+
"epoch": 0.02,
|
1791 |
+
"learning_rate": 4.752696442458315e-05,
|
1792 |
+
"loss": 4.5219,
|
1793 |
+
"step": 151040
|
1794 |
+
},
|
1795 |
+
{
|
1796 |
+
"epoch": 0.02,
|
1797 |
+
"learning_rate": 4.751857847707262e-05,
|
1798 |
+
"loss": 4.5113,
|
1799 |
+
"step": 151552
|
1800 |
+
},
|
1801 |
+
{
|
1802 |
+
"epoch": 0.02,
|
1803 |
+
"learning_rate": 4.751020890836584e-05,
|
1804 |
+
"loss": 4.5309,
|
1805 |
+
"step": 152064
|
1806 |
+
},
|
1807 |
+
{
|
1808 |
+
"epoch": 0.02,
|
1809 |
+
"learning_rate": 4.750182296085532e-05,
|
1810 |
+
"loss": 4.5106,
|
1811 |
+
"step": 152576
|
1812 |
+
},
|
1813 |
+
{
|
1814 |
+
"epoch": 0.03,
|
1815 |
+
"eval_loss": 4.504174709320068,
|
1816 |
+
"eval_runtime": 293.4935,
|
1817 |
+
"eval_samples_per_second": 1300.168,
|
1818 |
+
"eval_steps_per_second": 40.631,
|
1819 |
+
"step": 152640
|
1820 |
+
},
|
1821 |
+
{
|
1822 |
+
"epoch": 1.0,
|
1823 |
+
"learning_rate": 4.74934370133448e-05,
|
1824 |
+
"loss": 4.5019,
|
1825 |
+
"step": 153088
|
1826 |
+
},
|
1827 |
+
{
|
1828 |
+
"epoch": 1.0,
|
1829 |
+
"learning_rate": 4.748505106583428e-05,
|
1830 |
+
"loss": 4.504,
|
1831 |
+
"step": 153600
|
1832 |
+
},
|
1833 |
+
{
|
1834 |
+
"epoch": 1.0,
|
1835 |
+
"learning_rate": 4.747666511832376e-05,
|
1836 |
+
"loss": 4.5213,
|
1837 |
+
"step": 154112
|
1838 |
+
},
|
1839 |
+
{
|
1840 |
+
"epoch": 1.0,
|
1841 |
+
"learning_rate": 4.746829554961697e-05,
|
1842 |
+
"loss": 4.5127,
|
1843 |
+
"step": 154624
|
1844 |
+
},
|
1845 |
+
{
|
1846 |
+
"epoch": 1.0,
|
1847 |
+
"learning_rate": 4.7459909602106446e-05,
|
1848 |
+
"loss": 4.5167,
|
1849 |
+
"step": 155136
|
1850 |
+
},
|
1851 |
+
{
|
1852 |
+
"epoch": 1.0,
|
1853 |
+
"learning_rate": 4.7451523654595926e-05,
|
1854 |
+
"loss": 4.4965,
|
1855 |
+
"step": 155648
|
1856 |
+
},
|
1857 |
+
{
|
1858 |
+
"epoch": 1.0,
|
1859 |
+
"learning_rate": 4.7443137707085406e-05,
|
1860 |
+
"loss": 4.515,
|
1861 |
+
"step": 156160
|
1862 |
+
},
|
1863 |
+
{
|
1864 |
+
"epoch": 1.0,
|
1865 |
+
"learning_rate": 4.7434751759574886e-05,
|
1866 |
+
"loss": 4.4864,
|
1867 |
+
"step": 156672
|
1868 |
+
},
|
1869 |
+
{
|
1870 |
+
"epoch": 1.0,
|
1871 |
+
"learning_rate": 4.7426365812064366e-05,
|
1872 |
+
"loss": 4.5106,
|
1873 |
+
"step": 157184
|
1874 |
+
},
|
1875 |
+
{
|
1876 |
+
"epoch": 1.0,
|
1877 |
+
"learning_rate": 4.7417996243357575e-05,
|
1878 |
+
"loss": 4.494,
|
1879 |
+
"step": 157696
|
1880 |
+
},
|
1881 |
+
{
|
1882 |
+
"epoch": 1.0,
|
1883 |
+
"learning_rate": 4.7409610295847055e-05,
|
1884 |
+
"loss": 4.5109,
|
1885 |
+
"step": 158208
|
1886 |
+
},
|
1887 |
+
{
|
1888 |
+
"epoch": 1.0,
|
1889 |
+
"learning_rate": 4.7401224348336535e-05,
|
1890 |
+
"loss": 4.5093,
|
1891 |
+
"step": 158720
|
1892 |
+
},
|
1893 |
+
{
|
1894 |
+
"epoch": 1.0,
|
1895 |
+
"learning_rate": 4.7392838400826015e-05,
|
1896 |
+
"loss": 4.4959,
|
1897 |
+
"step": 159232
|
1898 |
+
},
|
1899 |
+
{
|
1900 |
+
"epoch": 1.0,
|
1901 |
+
"learning_rate": 4.7384452453315495e-05,
|
1902 |
+
"loss": 4.4926,
|
1903 |
+
"step": 159744
|
1904 |
+
},
|
1905 |
+
{
|
1906 |
+
"epoch": 1.0,
|
1907 |
+
"learning_rate": 4.7376066505804975e-05,
|
1908 |
+
"loss": 4.4861,
|
1909 |
+
"step": 160256
|
1910 |
+
},
|
1911 |
+
{
|
1912 |
+
"epoch": 1.0,
|
1913 |
+
"learning_rate": 4.736768055829446e-05,
|
1914 |
+
"loss": 4.4848,
|
1915 |
+
"step": 160768
|
1916 |
+
},
|
1917 |
+
{
|
1918 |
+
"epoch": 1.0,
|
1919 |
+
"learning_rate": 4.735929461078394e-05,
|
1920 |
+
"loss": 4.4913,
|
1921 |
+
"step": 161280
|
1922 |
+
},
|
1923 |
+
{
|
1924 |
+
"epoch": 1.0,
|
1925 |
+
"learning_rate": 4.735090866327342e-05,
|
1926 |
+
"loss": 4.4923,
|
1927 |
+
"step": 161792
|
1928 |
+
},
|
1929 |
+
{
|
1930 |
+
"epoch": 1.0,
|
1931 |
+
"learning_rate": 4.734253909456663e-05,
|
1932 |
+
"loss": 4.4909,
|
1933 |
+
"step": 162304
|
1934 |
+
},
|
1935 |
+
{
|
1936 |
+
"epoch": 1.0,
|
1937 |
+
"learning_rate": 4.733415314705611e-05,
|
1938 |
+
"loss": 4.502,
|
1939 |
+
"step": 162816
|
1940 |
+
},
|
1941 |
+
{
|
1942 |
+
"epoch": 1.0,
|
1943 |
+
"learning_rate": 4.732576719954559e-05,
|
1944 |
+
"loss": 4.5011,
|
1945 |
+
"step": 163328
|
1946 |
+
},
|
1947 |
+
{
|
1948 |
+
"epoch": 1.0,
|
1949 |
+
"learning_rate": 4.731738125203507e-05,
|
1950 |
+
"loss": 4.4945,
|
1951 |
+
"step": 163840
|
1952 |
+
},
|
1953 |
+
{
|
1954 |
+
"epoch": 1.0,
|
1955 |
+
"learning_rate": 4.730899530452455e-05,
|
1956 |
+
"loss": 4.4859,
|
1957 |
+
"step": 164352
|
1958 |
+
},
|
1959 |
+
{
|
1960 |
+
"epoch": 1.0,
|
1961 |
+
"learning_rate": 4.730062573581776e-05,
|
1962 |
+
"loss": 4.4972,
|
1963 |
+
"step": 164864
|
1964 |
+
},
|
1965 |
+
{
|
1966 |
+
"epoch": 1.0,
|
1967 |
+
"learning_rate": 4.729223978830724e-05,
|
1968 |
+
"loss": 4.4831,
|
1969 |
+
"step": 165376
|
1970 |
+
},
|
1971 |
+
{
|
1972 |
+
"epoch": 1.0,
|
1973 |
+
"learning_rate": 4.728385384079672e-05,
|
1974 |
+
"loss": 4.4908,
|
1975 |
+
"step": 165888
|
1976 |
+
},
|
1977 |
+
{
|
1978 |
+
"epoch": 1.0,
|
1979 |
+
"learning_rate": 4.72754678932862e-05,
|
1980 |
+
"loss": 4.4874,
|
1981 |
+
"step": 166400
|
1982 |
+
},
|
1983 |
+
{
|
1984 |
+
"epoch": 1.0,
|
1985 |
+
"learning_rate": 4.726708194577568e-05,
|
1986 |
+
"loss": 4.4826,
|
1987 |
+
"step": 166912
|
1988 |
+
},
|
1989 |
+
{
|
1990 |
+
"epoch": 1.0,
|
1991 |
+
"learning_rate": 4.725869599826516e-05,
|
1992 |
+
"loss": 4.4736,
|
1993 |
+
"step": 167424
|
1994 |
+
},
|
1995 |
+
{
|
1996 |
+
"epoch": 1.01,
|
1997 |
+
"learning_rate": 4.725031005075464e-05,
|
1998 |
+
"loss": 4.4761,
|
1999 |
+
"step": 167936
|
2000 |
+
},
|
2001 |
+
{
|
2002 |
+
"epoch": 1.01,
|
2003 |
+
"learning_rate": 4.7241940482047855e-05,
|
2004 |
+
"loss": 4.4784,
|
2005 |
+
"step": 168448
|
2006 |
+
},
|
2007 |
+
{
|
2008 |
+
"epoch": 1.01,
|
2009 |
+
"learning_rate": 4.7233554534537335e-05,
|
2010 |
+
"loss": 4.4804,
|
2011 |
+
"step": 168960
|
2012 |
+
},
|
2013 |
+
{
|
2014 |
+
"epoch": 1.01,
|
2015 |
+
"learning_rate": 4.722516858702681e-05,
|
2016 |
+
"loss": 4.4839,
|
2017 |
+
"step": 169472
|
2018 |
+
},
|
2019 |
+
{
|
2020 |
+
"epoch": 1.01,
|
2021 |
+
"learning_rate": 4.721678263951629e-05,
|
2022 |
+
"loss": 4.4796,
|
2023 |
+
"step": 169984
|
2024 |
+
},
|
2025 |
+
{
|
2026 |
+
"epoch": 1.01,
|
2027 |
+
"learning_rate": 4.720839669200577e-05,
|
2028 |
+
"loss": 4.4771,
|
2029 |
+
"step": 170496
|
2030 |
+
},
|
2031 |
+
{
|
2032 |
+
"epoch": 1.01,
|
2033 |
+
"learning_rate": 4.7200027123298984e-05,
|
2034 |
+
"loss": 4.4758,
|
2035 |
+
"step": 171008
|
2036 |
+
},
|
2037 |
+
{
|
2038 |
+
"epoch": 1.01,
|
2039 |
+
"learning_rate": 4.719164117578846e-05,
|
2040 |
+
"loss": 4.4578,
|
2041 |
+
"step": 171520
|
2042 |
+
},
|
2043 |
+
{
|
2044 |
+
"epoch": 1.01,
|
2045 |
+
"learning_rate": 4.718325522827794e-05,
|
2046 |
+
"loss": 4.4687,
|
2047 |
+
"step": 172032
|
2048 |
+
},
|
2049 |
+
{
|
2050 |
+
"epoch": 1.01,
|
2051 |
+
"learning_rate": 4.717486928076742e-05,
|
2052 |
+
"loss": 4.4602,
|
2053 |
+
"step": 172544
|
2054 |
+
},
|
2055 |
+
{
|
2056 |
+
"epoch": 1.01,
|
2057 |
+
"learning_rate": 4.71664833332569e-05,
|
2058 |
+
"loss": 4.4589,
|
2059 |
+
"step": 173056
|
2060 |
+
},
|
2061 |
+
{
|
2062 |
+
"epoch": 1.01,
|
2063 |
+
"learning_rate": 4.7158097385746384e-05,
|
2064 |
+
"loss": 4.464,
|
2065 |
+
"step": 173568
|
2066 |
+
},
|
2067 |
+
{
|
2068 |
+
"epoch": 1.01,
|
2069 |
+
"learning_rate": 4.714972781703959e-05,
|
2070 |
+
"loss": 4.4711,
|
2071 |
+
"step": 174080
|
2072 |
+
},
|
2073 |
+
{
|
2074 |
+
"epoch": 1.01,
|
2075 |
+
"learning_rate": 4.714134186952907e-05,
|
2076 |
+
"loss": 4.4581,
|
2077 |
+
"step": 174592
|
2078 |
+
},
|
2079 |
+
{
|
2080 |
+
"epoch": 1.01,
|
2081 |
+
"learning_rate": 4.713295592201855e-05,
|
2082 |
+
"loss": 4.4585,
|
2083 |
+
"step": 175104
|
2084 |
+
},
|
2085 |
+
{
|
2086 |
+
"epoch": 1.01,
|
2087 |
+
"learning_rate": 4.712456997450803e-05,
|
2088 |
+
"loss": 4.4662,
|
2089 |
+
"step": 175616
|
2090 |
+
},
|
2091 |
+
{
|
2092 |
+
"epoch": 1.01,
|
2093 |
+
"learning_rate": 4.711618402699751e-05,
|
2094 |
+
"loss": 4.46,
|
2095 |
+
"step": 176128
|
2096 |
+
},
|
2097 |
+
{
|
2098 |
+
"epoch": 1.01,
|
2099 |
+
"learning_rate": 4.710781445829072e-05,
|
2100 |
+
"loss": 4.4736,
|
2101 |
+
"step": 176640
|
2102 |
+
},
|
2103 |
+
{
|
2104 |
+
"epoch": 1.01,
|
2105 |
+
"learning_rate": 4.70994285107802e-05,
|
2106 |
+
"loss": 4.4581,
|
2107 |
+
"step": 177152
|
2108 |
+
},
|
2109 |
+
{
|
2110 |
+
"epoch": 1.01,
|
2111 |
+
"learning_rate": 4.709104256326968e-05,
|
2112 |
+
"loss": 4.4478,
|
2113 |
+
"step": 177664
|
2114 |
+
},
|
2115 |
+
{
|
2116 |
+
"epoch": 1.01,
|
2117 |
+
"learning_rate": 4.708265661575916e-05,
|
2118 |
+
"loss": 4.4604,
|
2119 |
+
"step": 178176
|
2120 |
+
},
|
2121 |
+
{
|
2122 |
+
"epoch": 1.01,
|
2123 |
+
"learning_rate": 4.707427066824864e-05,
|
2124 |
+
"loss": 4.4505,
|
2125 |
+
"step": 178688
|
2126 |
+
},
|
2127 |
+
{
|
2128 |
+
"epoch": 1.01,
|
2129 |
+
"learning_rate": 4.706588472073812e-05,
|
2130 |
+
"loss": 4.4595,
|
2131 |
+
"step": 179200
|
2132 |
+
},
|
2133 |
+
{
|
2134 |
+
"epoch": 1.01,
|
2135 |
+
"learning_rate": 4.70574987732276e-05,
|
2136 |
+
"loss": 4.4471,
|
2137 |
+
"step": 179712
|
2138 |
+
},
|
2139 |
+
{
|
2140 |
+
"epoch": 1.01,
|
2141 |
+
"learning_rate": 4.704911282571708e-05,
|
2142 |
+
"loss": 4.4516,
|
2143 |
+
"step": 180224
|
2144 |
+
},
|
2145 |
+
{
|
2146 |
+
"epoch": 1.01,
|
2147 |
+
"learning_rate": 4.70407432570103e-05,
|
2148 |
+
"loss": 4.4429,
|
2149 |
+
"step": 180736
|
2150 |
+
},
|
2151 |
+
{
|
2152 |
+
"epoch": 1.01,
|
2153 |
+
"learning_rate": 4.703235730949978e-05,
|
2154 |
+
"loss": 4.4565,
|
2155 |
+
"step": 181248
|
2156 |
+
},
|
2157 |
+
{
|
2158 |
+
"epoch": 1.01,
|
2159 |
+
"learning_rate": 4.702397136198926e-05,
|
2160 |
+
"loss": 4.4457,
|
2161 |
+
"step": 181760
|
2162 |
+
},
|
2163 |
+
{
|
2164 |
+
"epoch": 1.01,
|
2165 |
+
"learning_rate": 4.701558541447874e-05,
|
2166 |
+
"loss": 4.4454,
|
2167 |
+
"step": 182272
|
2168 |
+
},
|
2169 |
+
{
|
2170 |
+
"epoch": 1.01,
|
2171 |
+
"learning_rate": 4.700719946696822e-05,
|
2172 |
+
"loss": 4.4522,
|
2173 |
+
"step": 182784
|
2174 |
+
},
|
2175 |
+
{
|
2176 |
+
"epoch": 1.01,
|
2177 |
+
"learning_rate": 4.69988135194577e-05,
|
2178 |
+
"loss": 4.4395,
|
2179 |
+
"step": 183296
|
2180 |
+
},
|
2181 |
+
{
|
2182 |
+
"epoch": 1.01,
|
2183 |
+
"learning_rate": 4.6990443950750906e-05,
|
2184 |
+
"loss": 4.4345,
|
2185 |
+
"step": 183808
|
2186 |
+
},
|
2187 |
+
{
|
2188 |
+
"epoch": 1.01,
|
2189 |
+
"learning_rate": 4.6982058003240386e-05,
|
2190 |
+
"loss": 4.4566,
|
2191 |
+
"step": 184320
|
2192 |
+
},
|
2193 |
+
{
|
2194 |
+
"epoch": 1.01,
|
2195 |
+
"learning_rate": 4.6973672055729866e-05,
|
2196 |
+
"loss": 4.4232,
|
2197 |
+
"step": 184832
|
2198 |
+
},
|
2199 |
+
{
|
2200 |
+
"epoch": 1.01,
|
2201 |
+
"learning_rate": 4.6965302487023075e-05,
|
2202 |
+
"loss": 4.4286,
|
2203 |
+
"step": 185344
|
2204 |
+
},
|
2205 |
+
{
|
2206 |
+
"epoch": 1.01,
|
2207 |
+
"learning_rate": 4.6956916539512555e-05,
|
2208 |
+
"loss": 4.4376,
|
2209 |
+
"step": 185856
|
2210 |
+
},
|
2211 |
+
{
|
2212 |
+
"epoch": 1.01,
|
2213 |
+
"learning_rate": 4.6948530592002035e-05,
|
2214 |
+
"loss": 4.449,
|
2215 |
+
"step": 186368
|
2216 |
+
},
|
2217 |
+
{
|
2218 |
+
"epoch": 1.01,
|
2219 |
+
"learning_rate": 4.6940144644491515e-05,
|
2220 |
+
"loss": 4.4353,
|
2221 |
+
"step": 186880
|
2222 |
+
},
|
2223 |
+
{
|
2224 |
+
"epoch": 1.01,
|
2225 |
+
"learning_rate": 4.6931758696980995e-05,
|
2226 |
+
"loss": 4.4264,
|
2227 |
+
"step": 187392
|
2228 |
+
},
|
2229 |
+
{
|
2230 |
+
"epoch": 1.01,
|
2231 |
+
"learning_rate": 4.6923372749470475e-05,
|
2232 |
+
"loss": 4.4313,
|
2233 |
+
"step": 187904
|
2234 |
+
},
|
2235 |
+
{
|
2236 |
+
"epoch": 1.01,
|
2237 |
+
"learning_rate": 4.6914986801959955e-05,
|
2238 |
+
"loss": 4.4303,
|
2239 |
+
"step": 188416
|
2240 |
+
},
|
2241 |
+
{
|
2242 |
+
"epoch": 1.01,
|
2243 |
+
"learning_rate": 4.6906600854449435e-05,
|
2244 |
+
"loss": 4.4409,
|
2245 |
+
"step": 188928
|
2246 |
+
},
|
2247 |
+
{
|
2248 |
+
"epoch": 1.01,
|
2249 |
+
"learning_rate": 4.6898231285742644e-05,
|
2250 |
+
"loss": 4.4369,
|
2251 |
+
"step": 189440
|
2252 |
+
},
|
2253 |
+
{
|
2254 |
+
"epoch": 1.01,
|
2255 |
+
"learning_rate": 4.688986171703586e-05,
|
2256 |
+
"loss": 4.4348,
|
2257 |
+
"step": 189952
|
2258 |
+
},
|
2259 |
+
{
|
2260 |
+
"epoch": 1.01,
|
2261 |
+
"learning_rate": 4.688147576952534e-05,
|
2262 |
+
"loss": 4.4396,
|
2263 |
+
"step": 190464
|
2264 |
+
},
|
2265 |
+
{
|
2266 |
+
"epoch": 1.01,
|
2267 |
+
"learning_rate": 4.687308982201481e-05,
|
2268 |
+
"loss": 4.442,
|
2269 |
+
"step": 190976
|
2270 |
+
},
|
2271 |
+
{
|
2272 |
+
"epoch": 1.01,
|
2273 |
+
"learning_rate": 4.686470387450429e-05,
|
2274 |
+
"loss": 4.4226,
|
2275 |
+
"step": 191488
|
2276 |
+
},
|
2277 |
+
{
|
2278 |
+
"epoch": 1.01,
|
2279 |
+
"learning_rate": 4.685631792699377e-05,
|
2280 |
+
"loss": 4.4227,
|
2281 |
+
"step": 192000
|
2282 |
+
},
|
2283 |
+
{
|
2284 |
+
"epoch": 1.01,
|
2285 |
+
"learning_rate": 4.684793197948325e-05,
|
2286 |
+
"loss": 4.4332,
|
2287 |
+
"step": 192512
|
2288 |
+
},
|
2289 |
+
{
|
2290 |
+
"epoch": 1.01,
|
2291 |
+
"learning_rate": 4.683954603197274e-05,
|
2292 |
+
"loss": 4.4152,
|
2293 |
+
"step": 193024
|
2294 |
+
},
|
2295 |
+
{
|
2296 |
+
"epoch": 1.01,
|
2297 |
+
"learning_rate": 4.683116008446222e-05,
|
2298 |
+
"loss": 4.4318,
|
2299 |
+
"step": 193536
|
2300 |
+
},
|
2301 |
+
{
|
2302 |
+
"epoch": 1.01,
|
2303 |
+
"learning_rate": 4.68227741369517e-05,
|
2304 |
+
"loss": 4.4173,
|
2305 |
+
"step": 194048
|
2306 |
+
},
|
2307 |
+
{
|
2308 |
+
"epoch": 1.01,
|
2309 |
+
"learning_rate": 4.681438818944118e-05,
|
2310 |
+
"loss": 4.4375,
|
2311 |
+
"step": 194560
|
2312 |
+
},
|
2313 |
+
{
|
2314 |
+
"epoch": 1.01,
|
2315 |
+
"learning_rate": 4.680600224193066e-05,
|
2316 |
+
"loss": 4.4238,
|
2317 |
+
"step": 195072
|
2318 |
+
},
|
2319 |
+
{
|
2320 |
+
"epoch": 1.01,
|
2321 |
+
"learning_rate": 4.679763267322387e-05,
|
2322 |
+
"loss": 4.4116,
|
2323 |
+
"step": 195584
|
2324 |
+
},
|
2325 |
+
{
|
2326 |
+
"epoch": 1.01,
|
2327 |
+
"learning_rate": 4.678924672571335e-05,
|
2328 |
+
"loss": 4.416,
|
2329 |
+
"step": 196096
|
2330 |
+
},
|
2331 |
+
{
|
2332 |
+
"epoch": 1.01,
|
2333 |
+
"learning_rate": 4.678086077820283e-05,
|
2334 |
+
"loss": 4.4232,
|
2335 |
+
"step": 196608
|
2336 |
+
},
|
2337 |
+
{
|
2338 |
+
"epoch": 1.01,
|
2339 |
+
"learning_rate": 4.677247483069231e-05,
|
2340 |
+
"loss": 4.431,
|
2341 |
+
"step": 197120
|
2342 |
+
},
|
2343 |
+
{
|
2344 |
+
"epoch": 1.01,
|
2345 |
+
"learning_rate": 4.676408888318179e-05,
|
2346 |
+
"loss": 4.4172,
|
2347 |
+
"step": 197632
|
2348 |
+
},
|
2349 |
+
{
|
2350 |
+
"epoch": 1.01,
|
2351 |
+
"learning_rate": 4.6755719314475e-05,
|
2352 |
+
"loss": 4.4225,
|
2353 |
+
"step": 198144
|
2354 |
+
},
|
2355 |
+
{
|
2356 |
+
"epoch": 1.02,
|
2357 |
+
"learning_rate": 4.674733336696448e-05,
|
2358 |
+
"loss": 4.4125,
|
2359 |
+
"step": 198656
|
2360 |
+
},
|
2361 |
+
{
|
2362 |
+
"epoch": 1.02,
|
2363 |
+
"learning_rate": 4.673894741945396e-05,
|
2364 |
+
"loss": 4.4215,
|
2365 |
+
"step": 199168
|
2366 |
+
},
|
2367 |
+
{
|
2368 |
+
"epoch": 1.02,
|
2369 |
+
"learning_rate": 4.673057785074717e-05,
|
2370 |
+
"loss": 4.4132,
|
2371 |
+
"step": 199680
|
2372 |
+
},
|
2373 |
+
{
|
2374 |
+
"epoch": 1.02,
|
2375 |
+
"learning_rate": 4.672219190323665e-05,
|
2376 |
+
"loss": 4.4096,
|
2377 |
+
"step": 200192
|
2378 |
+
},
|
2379 |
+
{
|
2380 |
+
"epoch": 1.02,
|
2381 |
+
"learning_rate": 4.671380595572613e-05,
|
2382 |
+
"loss": 4.4203,
|
2383 |
+
"step": 200704
|
2384 |
+
},
|
2385 |
+
{
|
2386 |
+
"epoch": 1.02,
|
2387 |
+
"learning_rate": 4.670542000821561e-05,
|
2388 |
+
"loss": 4.4184,
|
2389 |
+
"step": 201216
|
2390 |
+
},
|
2391 |
+
{
|
2392 |
+
"epoch": 1.02,
|
2393 |
+
"learning_rate": 4.669703406070509e-05,
|
2394 |
+
"loss": 4.4177,
|
2395 |
+
"step": 201728
|
2396 |
+
},
|
2397 |
+
{
|
2398 |
+
"epoch": 1.02,
|
2399 |
+
"learning_rate": 4.668864811319457e-05,
|
2400 |
+
"loss": 4.4027,
|
2401 |
+
"step": 202240
|
2402 |
+
},
|
2403 |
+
{
|
2404 |
+
"epoch": 1.02,
|
2405 |
+
"learning_rate": 4.668026216568405e-05,
|
2406 |
+
"loss": 4.4058,
|
2407 |
+
"step": 202752
|
2408 |
+
},
|
2409 |
+
{
|
2410 |
+
"epoch": 1.02,
|
2411 |
+
"learning_rate": 4.667187621817353e-05,
|
2412 |
+
"loss": 4.4103,
|
2413 |
+
"step": 203264
|
2414 |
+
},
|
2415 |
+
{
|
2416 |
+
"epoch": 1.02,
|
2417 |
+
"learning_rate": 4.6663490270663006e-05,
|
2418 |
+
"loss": 4.4091,
|
2419 |
+
"step": 203776
|
2420 |
+
},
|
2421 |
+
{
|
2422 |
+
"epoch": 1.02,
|
2423 |
+
"learning_rate": 4.6655104323152486e-05,
|
2424 |
+
"loss": 4.4089,
|
2425 |
+
"step": 204288
|
2426 |
+
},
|
2427 |
+
{
|
2428 |
+
"epoch": 1.02,
|
2429 |
+
"learning_rate": 4.66467347544457e-05,
|
2430 |
+
"loss": 4.4003,
|
2431 |
+
"step": 204800
|
2432 |
+
},
|
2433 |
+
{
|
2434 |
+
"epoch": 1.02,
|
2435 |
+
"learning_rate": 4.6638348806935175e-05,
|
2436 |
+
"loss": 4.3996,
|
2437 |
+
"step": 205312
|
2438 |
+
},
|
2439 |
+
{
|
2440 |
+
"epoch": 1.02,
|
2441 |
+
"learning_rate": 4.662996285942466e-05,
|
2442 |
+
"loss": 4.4099,
|
2443 |
+
"step": 205824
|
2444 |
+
},
|
2445 |
+
{
|
2446 |
+
"epoch": 1.02,
|
2447 |
+
"learning_rate": 4.662157691191414e-05,
|
2448 |
+
"loss": 4.3817,
|
2449 |
+
"step": 206336
|
2450 |
+
},
|
2451 |
+
{
|
2452 |
+
"epoch": 1.02,
|
2453 |
+
"learning_rate": 4.661320734320736e-05,
|
2454 |
+
"loss": 4.4004,
|
2455 |
+
"step": 206848
|
2456 |
+
},
|
2457 |
+
{
|
2458 |
+
"epoch": 1.02,
|
2459 |
+
"learning_rate": 4.660482139569683e-05,
|
2460 |
+
"loss": 4.3916,
|
2461 |
+
"step": 207360
|
2462 |
+
},
|
2463 |
+
{
|
2464 |
+
"epoch": 1.02,
|
2465 |
+
"learning_rate": 4.659643544818631e-05,
|
2466 |
+
"loss": 4.4071,
|
2467 |
+
"step": 207872
|
2468 |
+
},
|
2469 |
+
{
|
2470 |
+
"epoch": 1.02,
|
2471 |
+
"learning_rate": 4.658804950067579e-05,
|
2472 |
+
"loss": 4.3957,
|
2473 |
+
"step": 208384
|
2474 |
+
},
|
2475 |
+
{
|
2476 |
+
"epoch": 1.02,
|
2477 |
+
"learning_rate": 4.6579679931969006e-05,
|
2478 |
+
"loss": 4.4037,
|
2479 |
+
"step": 208896
|
2480 |
+
},
|
2481 |
+
{
|
2482 |
+
"epoch": 1.02,
|
2483 |
+
"learning_rate": 4.657129398445848e-05,
|
2484 |
+
"loss": 4.3946,
|
2485 |
+
"step": 209408
|
2486 |
+
},
|
2487 |
+
{
|
2488 |
+
"epoch": 1.02,
|
2489 |
+
"learning_rate": 4.656290803694796e-05,
|
2490 |
+
"loss": 4.3927,
|
2491 |
+
"step": 209920
|
2492 |
+
},
|
2493 |
+
{
|
2494 |
+
"epoch": 1.02,
|
2495 |
+
"learning_rate": 4.655452208943744e-05,
|
2496 |
+
"loss": 4.4037,
|
2497 |
+
"step": 210432
|
2498 |
+
},
|
2499 |
+
{
|
2500 |
+
"epoch": 1.02,
|
2501 |
+
"learning_rate": 4.654613614192692e-05,
|
2502 |
+
"loss": 4.3914,
|
2503 |
+
"step": 210944
|
2504 |
+
},
|
2505 |
+
{
|
2506 |
+
"epoch": 1.02,
|
2507 |
+
"learning_rate": 4.65377501944164e-05,
|
2508 |
+
"loss": 4.404,
|
2509 |
+
"step": 211456
|
2510 |
+
},
|
2511 |
+
{
|
2512 |
+
"epoch": 1.02,
|
2513 |
+
"learning_rate": 4.652936424690588e-05,
|
2514 |
+
"loss": 4.3992,
|
2515 |
+
"step": 211968
|
2516 |
+
},
|
2517 |
+
{
|
2518 |
+
"epoch": 1.02,
|
2519 |
+
"learning_rate": 4.652097829939536e-05,
|
2520 |
+
"loss": 4.3985,
|
2521 |
+
"step": 212480
|
2522 |
+
},
|
2523 |
+
{
|
2524 |
+
"epoch": 1.02,
|
2525 |
+
"learning_rate": 4.6512608730688575e-05,
|
2526 |
+
"loss": 4.39,
|
2527 |
+
"step": 212992
|
2528 |
+
},
|
2529 |
+
{
|
2530 |
+
"epoch": 1.02,
|
2531 |
+
"learning_rate": 4.6504222783178055e-05,
|
2532 |
+
"loss": 4.3856,
|
2533 |
+
"step": 213504
|
2534 |
+
},
|
2535 |
+
{
|
2536 |
+
"epoch": 1.02,
|
2537 |
+
"learning_rate": 4.6495836835667535e-05,
|
2538 |
+
"loss": 4.4009,
|
2539 |
+
"step": 214016
|
2540 |
+
},
|
2541 |
+
{
|
2542 |
+
"epoch": 1.02,
|
2543 |
+
"learning_rate": 4.6487450888157015e-05,
|
2544 |
+
"loss": 4.3843,
|
2545 |
+
"step": 214528
|
2546 |
+
},
|
2547 |
+
{
|
2548 |
+
"epoch": 1.02,
|
2549 |
+
"learning_rate": 4.6479064940646495e-05,
|
2550 |
+
"loss": 4.3996,
|
2551 |
+
"step": 215040
|
2552 |
+
},
|
2553 |
+
{
|
2554 |
+
"epoch": 1.02,
|
2555 |
+
"learning_rate": 4.6470695371939704e-05,
|
2556 |
+
"loss": 4.3843,
|
2557 |
+
"step": 215552
|
2558 |
+
},
|
2559 |
+
{
|
2560 |
+
"epoch": 1.02,
|
2561 |
+
"learning_rate": 4.6462309424429184e-05,
|
2562 |
+
"loss": 4.3894,
|
2563 |
+
"step": 216064
|
2564 |
+
},
|
2565 |
+
{
|
2566 |
+
"epoch": 1.02,
|
2567 |
+
"learning_rate": 4.6453923476918664e-05,
|
2568 |
+
"loss": 4.389,
|
2569 |
+
"step": 216576
|
2570 |
+
},
|
2571 |
+
{
|
2572 |
+
"epoch": 1.02,
|
2573 |
+
"learning_rate": 4.6445537529408144e-05,
|
2574 |
+
"loss": 4.3852,
|
2575 |
+
"step": 217088
|
2576 |
+
},
|
2577 |
+
{
|
2578 |
+
"epoch": 1.02,
|
2579 |
+
"learning_rate": 4.6437151581897624e-05,
|
2580 |
+
"loss": 4.3666,
|
2581 |
+
"step": 217600
|
2582 |
+
},
|
2583 |
+
{
|
2584 |
+
"epoch": 1.02,
|
2585 |
+
"learning_rate": 4.6428765634387104e-05,
|
2586 |
+
"loss": 4.3846,
|
2587 |
+
"step": 218112
|
2588 |
+
},
|
2589 |
+
{
|
2590 |
+
"epoch": 1.02,
|
2591 |
+
"learning_rate": 4.6420379686876584e-05,
|
2592 |
+
"loss": 4.3722,
|
2593 |
+
"step": 218624
|
2594 |
+
},
|
2595 |
+
{
|
2596 |
+
"epoch": 1.02,
|
2597 |
+
"learning_rate": 4.64120101181698e-05,
|
2598 |
+
"loss": 4.3871,
|
2599 |
+
"step": 219136
|
2600 |
+
},
|
2601 |
+
{
|
2602 |
+
"epoch": 1.02,
|
2603 |
+
"learning_rate": 4.640362417065928e-05,
|
2604 |
+
"loss": 4.4007,
|
2605 |
+
"step": 219648
|
2606 |
+
},
|
2607 |
+
{
|
2608 |
+
"epoch": 1.02,
|
2609 |
+
"learning_rate": 4.639523822314876e-05,
|
2610 |
+
"loss": 4.3949,
|
2611 |
+
"step": 220160
|
2612 |
+
},
|
2613 |
+
{
|
2614 |
+
"epoch": 1.02,
|
2615 |
+
"learning_rate": 4.638685227563824e-05,
|
2616 |
+
"loss": 4.3641,
|
2617 |
+
"step": 220672
|
2618 |
+
},
|
2619 |
+
{
|
2620 |
+
"epoch": 1.02,
|
2621 |
+
"learning_rate": 4.637848270693145e-05,
|
2622 |
+
"loss": 4.3882,
|
2623 |
+
"step": 221184
|
2624 |
+
},
|
2625 |
+
{
|
2626 |
+
"epoch": 1.02,
|
2627 |
+
"learning_rate": 4.637009675942093e-05,
|
2628 |
+
"loss": 4.3827,
|
2629 |
+
"step": 221696
|
2630 |
+
},
|
2631 |
+
{
|
2632 |
+
"epoch": 1.02,
|
2633 |
+
"learning_rate": 4.636171081191041e-05,
|
2634 |
+
"loss": 4.3809,
|
2635 |
+
"step": 222208
|
2636 |
+
},
|
2637 |
+
{
|
2638 |
+
"epoch": 1.02,
|
2639 |
+
"learning_rate": 4.635332486439989e-05,
|
2640 |
+
"loss": 4.3706,
|
2641 |
+
"step": 222720
|
2642 |
+
},
|
2643 |
+
{
|
2644 |
+
"epoch": 1.02,
|
2645 |
+
"learning_rate": 4.634493891688936e-05,
|
2646 |
+
"loss": 4.3836,
|
2647 |
+
"step": 223232
|
2648 |
+
},
|
2649 |
+
{
|
2650 |
+
"epoch": 1.02,
|
2651 |
+
"learning_rate": 4.633655296937884e-05,
|
2652 |
+
"loss": 4.3772,
|
2653 |
+
"step": 223744
|
2654 |
+
},
|
2655 |
+
{
|
2656 |
+
"epoch": 1.02,
|
2657 |
+
"learning_rate": 4.632816702186832e-05,
|
2658 |
+
"loss": 4.3758,
|
2659 |
+
"step": 224256
|
2660 |
+
},
|
2661 |
+
{
|
2662 |
+
"epoch": 1.02,
|
2663 |
+
"learning_rate": 4.63197810743578e-05,
|
2664 |
+
"loss": 4.3782,
|
2665 |
+
"step": 224768
|
2666 |
+
},
|
2667 |
+
{
|
2668 |
+
"epoch": 1.02,
|
2669 |
+
"learning_rate": 4.631141150565102e-05,
|
2670 |
+
"loss": 4.3797,
|
2671 |
+
"step": 225280
|
2672 |
+
},
|
2673 |
+
{
|
2674 |
+
"epoch": 1.02,
|
2675 |
+
"learning_rate": 4.63030255581405e-05,
|
2676 |
+
"loss": 4.3755,
|
2677 |
+
"step": 225792
|
2678 |
+
},
|
2679 |
+
{
|
2680 |
+
"epoch": 1.02,
|
2681 |
+
"learning_rate": 4.629463961062998e-05,
|
2682 |
+
"loss": 4.3728,
|
2683 |
+
"step": 226304
|
2684 |
+
},
|
2685 |
+
{
|
2686 |
+
"epoch": 1.02,
|
2687 |
+
"learning_rate": 4.628625366311946e-05,
|
2688 |
+
"loss": 4.3783,
|
2689 |
+
"step": 226816
|
2690 |
+
},
|
2691 |
+
{
|
2692 |
+
"epoch": 1.02,
|
2693 |
+
"learning_rate": 4.627786771560894e-05,
|
2694 |
+
"loss": 4.3778,
|
2695 |
+
"step": 227328
|
2696 |
+
},
|
2697 |
+
{
|
2698 |
+
"epoch": 1.02,
|
2699 |
+
"learning_rate": 4.626948176809842e-05,
|
2700 |
+
"loss": 4.3668,
|
2701 |
+
"step": 227840
|
2702 |
+
},
|
2703 |
+
{
|
2704 |
+
"epoch": 1.02,
|
2705 |
+
"learning_rate": 4.62610958205879e-05,
|
2706 |
+
"loss": 4.3842,
|
2707 |
+
"step": 228352
|
2708 |
+
},
|
2709 |
+
{
|
2710 |
+
"epoch": 1.02,
|
2711 |
+
"learning_rate": 4.625270987307738e-05,
|
2712 |
+
"loss": 4.3687,
|
2713 |
+
"step": 228864
|
2714 |
+
},
|
2715 |
+
{
|
2716 |
+
"epoch": 1.03,
|
2717 |
+
"eval_loss": 4.369020462036133,
|
2718 |
+
"eval_runtime": 293.4838,
|
2719 |
+
"eval_samples_per_second": 1300.211,
|
2720 |
+
"eval_steps_per_second": 40.633,
|
2721 |
+
"step": 228960
|
2722 |
+
},
|
2723 |
+
{
|
2724 |
+
"epoch": 0.0,
|
2725 |
+
"learning_rate": 4.6244340304370586e-05,
|
2726 |
+
"loss": 4.3576,
|
2727 |
+
"step": 229376
|
2728 |
+
},
|
2729 |
+
{
|
2730 |
+
"epoch": 0.0,
|
2731 |
+
"learning_rate": 4.6235954356860066e-05,
|
2732 |
+
"loss": 4.3609,
|
2733 |
+
"step": 229888
|
2734 |
+
},
|
2735 |
+
{
|
2736 |
+
"epoch": 0.0,
|
2737 |
+
"learning_rate": 4.6227568409349546e-05,
|
2738 |
+
"loss": 4.3805,
|
2739 |
+
"step": 230400
|
2740 |
+
},
|
2741 |
+
{
|
2742 |
+
"epoch": 0.0,
|
2743 |
+
"learning_rate": 4.6219182461839026e-05,
|
2744 |
+
"loss": 4.3682,
|
2745 |
+
"step": 230912
|
2746 |
+
},
|
2747 |
+
{
|
2748 |
+
"epoch": 0.0,
|
2749 |
+
"learning_rate": 4.6210796514328506e-05,
|
2750 |
+
"loss": 4.3807,
|
2751 |
+
"step": 231424
|
2752 |
+
},
|
2753 |
+
{
|
2754 |
+
"epoch": 0.0,
|
2755 |
+
"learning_rate": 4.6202410566817986e-05,
|
2756 |
+
"loss": 4.3553,
|
2757 |
+
"step": 231936
|
2758 |
+
},
|
2759 |
+
{
|
2760 |
+
"epoch": 0.0,
|
2761 |
+
"learning_rate": 4.6194024619307466e-05,
|
2762 |
+
"loss": 4.3746,
|
2763 |
+
"step": 232448
|
2764 |
+
},
|
2765 |
+
{
|
2766 |
+
"epoch": 0.0,
|
2767 |
+
"learning_rate": 4.6185638671796946e-05,
|
2768 |
+
"loss": 4.3446,
|
2769 |
+
"step": 232960
|
2770 |
+
},
|
2771 |
+
{
|
2772 |
+
"epoch": 0.0,
|
2773 |
+
"learning_rate": 4.617725272428643e-05,
|
2774 |
+
"loss": 4.3762,
|
2775 |
+
"step": 233472
|
2776 |
+
},
|
2777 |
+
{
|
2778 |
+
"epoch": 0.0,
|
2779 |
+
"learning_rate": 4.6168866776775906e-05,
|
2780 |
+
"loss": 4.3568,
|
2781 |
+
"step": 233984
|
2782 |
+
},
|
2783 |
+
{
|
2784 |
+
"epoch": 0.0,
|
2785 |
+
"learning_rate": 4.6160480829265386e-05,
|
2786 |
+
"loss": 4.3721,
|
2787 |
+
"step": 234496
|
2788 |
+
},
|
2789 |
+
{
|
2790 |
+
"epoch": 0.0,
|
2791 |
+
"learning_rate": 4.6152094881754866e-05,
|
2792 |
+
"loss": 4.3707,
|
2793 |
+
"step": 235008
|
2794 |
+
},
|
2795 |
+
{
|
2796 |
+
"epoch": 0.0,
|
2797 |
+
"learning_rate": 4.614372531304808e-05,
|
2798 |
+
"loss": 4.3542,
|
2799 |
+
"step": 235520
|
2800 |
+
},
|
2801 |
+
{
|
2802 |
+
"epoch": 0.0,
|
2803 |
+
"learning_rate": 4.6135339365537555e-05,
|
2804 |
+
"loss": 4.3602,
|
2805 |
+
"step": 236032
|
2806 |
+
},
|
2807 |
+
{
|
2808 |
+
"epoch": 0.0,
|
2809 |
+
"learning_rate": 4.6126953418027035e-05,
|
2810 |
+
"loss": 4.3531,
|
2811 |
+
"step": 236544
|
2812 |
+
},
|
2813 |
+
{
|
2814 |
+
"epoch": 0.0,
|
2815 |
+
"learning_rate": 4.6118567470516515e-05,
|
2816 |
+
"loss": 4.3465,
|
2817 |
+
"step": 237056
|
2818 |
+
},
|
2819 |
+
{
|
2820 |
+
"epoch": 0.0,
|
2821 |
+
"learning_rate": 4.6110181523005995e-05,
|
2822 |
+
"loss": 4.3617,
|
2823 |
+
"step": 237568
|
2824 |
+
},
|
2825 |
+
{
|
2826 |
+
"epoch": 0.0,
|
2827 |
+
"learning_rate": 4.6101795575495475e-05,
|
2828 |
+
"loss": 4.3574,
|
2829 |
+
"step": 238080
|
2830 |
+
},
|
2831 |
+
{
|
2832 |
+
"epoch": 0.0,
|
2833 |
+
"learning_rate": 4.6093409627984955e-05,
|
2834 |
+
"loss": 4.3582,
|
2835 |
+
"step": 238592
|
2836 |
+
},
|
2837 |
+
{
|
2838 |
+
"epoch": 0.0,
|
2839 |
+
"learning_rate": 4.6085023680474435e-05,
|
2840 |
+
"loss": 4.366,
|
2841 |
+
"step": 239104
|
2842 |
+
},
|
2843 |
+
{
|
2844 |
+
"epoch": 0.0,
|
2845 |
+
"learning_rate": 4.607665411176765e-05,
|
2846 |
+
"loss": 4.3698,
|
2847 |
+
"step": 239616
|
2848 |
+
},
|
2849 |
+
{
|
2850 |
+
"epoch": 0.0,
|
2851 |
+
"learning_rate": 4.606826816425713e-05,
|
2852 |
+
"loss": 4.3607,
|
2853 |
+
"step": 240128
|
2854 |
+
},
|
2855 |
+
{
|
2856 |
+
"epoch": 0.0,
|
2857 |
+
"learning_rate": 4.605988221674661e-05,
|
2858 |
+
"loss": 4.3538,
|
2859 |
+
"step": 240640
|
2860 |
+
},
|
2861 |
+
{
|
2862 |
+
"epoch": 0.0,
|
2863 |
+
"learning_rate": 4.605151264803982e-05,
|
2864 |
+
"loss": 4.3653,
|
2865 |
+
"step": 241152
|
2866 |
+
},
|
2867 |
+
{
|
2868 |
+
"epoch": 0.0,
|
2869 |
+
"learning_rate": 4.60431267005293e-05,
|
2870 |
+
"loss": 4.3514,
|
2871 |
+
"step": 241664
|
2872 |
+
},
|
2873 |
+
{
|
2874 |
+
"epoch": 0.0,
|
2875 |
+
"learning_rate": 4.603474075301878e-05,
|
2876 |
+
"loss": 4.3636,
|
2877 |
+
"step": 242176
|
2878 |
+
},
|
2879 |
+
{
|
2880 |
+
"epoch": 0.0,
|
2881 |
+
"learning_rate": 4.602635480550826e-05,
|
2882 |
+
"loss": 4.3564,
|
2883 |
+
"step": 242688
|
2884 |
+
},
|
2885 |
+
{
|
2886 |
+
"epoch": 0.0,
|
2887 |
+
"learning_rate": 4.601796885799774e-05,
|
2888 |
+
"loss": 4.349,
|
2889 |
+
"step": 243200
|
2890 |
+
},
|
2891 |
+
{
|
2892 |
+
"epoch": 0.0,
|
2893 |
+
"learning_rate": 4.600958291048722e-05,
|
2894 |
+
"loss": 4.3438,
|
2895 |
+
"step": 243712
|
2896 |
+
},
|
2897 |
+
{
|
2898 |
+
"epoch": 0.01,
|
2899 |
+
"learning_rate": 4.60011969629767e-05,
|
2900 |
+
"loss": 4.3505,
|
2901 |
+
"step": 244224
|
2902 |
+
},
|
2903 |
+
{
|
2904 |
+
"epoch": 0.01,
|
2905 |
+
"learning_rate": 4.599281101546618e-05,
|
2906 |
+
"loss": 4.3466,
|
2907 |
+
"step": 244736
|
2908 |
+
},
|
2909 |
+
{
|
2910 |
+
"epoch": 0.01,
|
2911 |
+
"learning_rate": 4.598444144675939e-05,
|
2912 |
+
"loss": 4.3545,
|
2913 |
+
"step": 245248
|
2914 |
+
},
|
2915 |
+
{
|
2916 |
+
"epoch": 0.01,
|
2917 |
+
"learning_rate": 4.597605549924887e-05,
|
2918 |
+
"loss": 4.3544,
|
2919 |
+
"step": 245760
|
2920 |
+
},
|
2921 |
+
{
|
2922 |
+
"epoch": 0.01,
|
2923 |
+
"learning_rate": 4.596766955173835e-05,
|
2924 |
+
"loss": 4.3518,
|
2925 |
+
"step": 246272
|
2926 |
+
},
|
2927 |
+
{
|
2928 |
+
"epoch": 0.01,
|
2929 |
+
"learning_rate": 4.5959283604227835e-05,
|
2930 |
+
"loss": 4.3519,
|
2931 |
+
"step": 246784
|
2932 |
+
},
|
2933 |
+
{
|
2934 |
+
"epoch": 0.01,
|
2935 |
+
"learning_rate": 4.5950914035521044e-05,
|
2936 |
+
"loss": 4.3489,
|
2937 |
+
"step": 247296
|
2938 |
+
},
|
2939 |
+
{
|
2940 |
+
"epoch": 0.01,
|
2941 |
+
"learning_rate": 4.5942528088010524e-05,
|
2942 |
+
"loss": 4.3332,
|
2943 |
+
"step": 247808
|
2944 |
+
},
|
2945 |
+
{
|
2946 |
+
"epoch": 0.01,
|
2947 |
+
"learning_rate": 4.5934142140500004e-05,
|
2948 |
+
"loss": 4.3418,
|
2949 |
+
"step": 248320
|
2950 |
+
},
|
2951 |
+
{
|
2952 |
+
"epoch": 0.01,
|
2953 |
+
"learning_rate": 4.5925756192989484e-05,
|
2954 |
+
"loss": 4.3342,
|
2955 |
+
"step": 248832
|
2956 |
+
},
|
2957 |
+
{
|
2958 |
+
"epoch": 0.01,
|
2959 |
+
"learning_rate": 4.5917370245478964e-05,
|
2960 |
+
"loss": 4.3396,
|
2961 |
+
"step": 249344
|
2962 |
+
},
|
2963 |
+
{
|
2964 |
+
"epoch": 0.01,
|
2965 |
+
"learning_rate": 4.590898429796844e-05,
|
2966 |
+
"loss": 4.3418,
|
2967 |
+
"step": 249856
|
2968 |
+
},
|
2969 |
+
{
|
2970 |
+
"epoch": 0.01,
|
2971 |
+
"learning_rate": 4.590059835045792e-05,
|
2972 |
+
"loss": 4.3453,
|
2973 |
+
"step": 250368
|
2974 |
+
},
|
2975 |
+
{
|
2976 |
+
"epoch": 0.01,
|
2977 |
+
"learning_rate": 4.589222878175113e-05,
|
2978 |
+
"loss": 4.3412,
|
2979 |
+
"step": 250880
|
2980 |
+
},
|
2981 |
+
{
|
2982 |
+
"epoch": 0.01,
|
2983 |
+
"learning_rate": 4.588384283424061e-05,
|
2984 |
+
"loss": 4.3325,
|
2985 |
+
"step": 251392
|
2986 |
+
},
|
2987 |
+
{
|
2988 |
+
"epoch": 0.01,
|
2989 |
+
"learning_rate": 4.587545688673009e-05,
|
2990 |
+
"loss": 4.345,
|
2991 |
+
"step": 251904
|
2992 |
+
},
|
2993 |
+
{
|
2994 |
+
"epoch": 0.01,
|
2995 |
+
"learning_rate": 4.586707093921957e-05,
|
2996 |
+
"loss": 4.3431,
|
2997 |
+
"step": 252416
|
2998 |
+
},
|
2999 |
+
{
|
3000 |
+
"epoch": 0.01,
|
3001 |
+
"learning_rate": 4.585868499170905e-05,
|
3002 |
+
"loss": 4.3468,
|
3003 |
+
"step": 252928
|
3004 |
+
},
|
3005 |
+
{
|
3006 |
+
"epoch": 0.01,
|
3007 |
+
"learning_rate": 4.585029904419853e-05,
|
3008 |
+
"loss": 4.3382,
|
3009 |
+
"step": 253440
|
3010 |
+
},
|
3011 |
+
{
|
3012 |
+
"epoch": 0.01,
|
3013 |
+
"learning_rate": 4.584192947549174e-05,
|
3014 |
+
"loss": 4.326,
|
3015 |
+
"step": 253952
|
3016 |
+
},
|
3017 |
+
{
|
3018 |
+
"epoch": 0.01,
|
3019 |
+
"learning_rate": 4.583354352798122e-05,
|
3020 |
+
"loss": 4.3409,
|
3021 |
+
"step": 254464
|
3022 |
+
},
|
3023 |
+
{
|
3024 |
+
"epoch": 0.01,
|
3025 |
+
"learning_rate": 4.582517395927444e-05,
|
3026 |
+
"loss": 4.3303,
|
3027 |
+
"step": 254976
|
3028 |
+
},
|
3029 |
+
{
|
3030 |
+
"epoch": 0.01,
|
3031 |
+
"learning_rate": 4.581678801176391e-05,
|
3032 |
+
"loss": 4.3389,
|
3033 |
+
"step": 255488
|
3034 |
+
},
|
3035 |
+
{
|
3036 |
+
"epoch": 0.01,
|
3037 |
+
"learning_rate": 4.580840206425339e-05,
|
3038 |
+
"loss": 4.3287,
|
3039 |
+
"step": 256000
|
3040 |
+
},
|
3041 |
+
{
|
3042 |
+
"epoch": 0.01,
|
3043 |
+
"learning_rate": 4.580001611674287e-05,
|
3044 |
+
"loss": 4.3275,
|
3045 |
+
"step": 256512
|
3046 |
+
},
|
3047 |
+
{
|
3048 |
+
"epoch": 0.01,
|
3049 |
+
"learning_rate": 4.579163016923235e-05,
|
3050 |
+
"loss": 4.3279,
|
3051 |
+
"step": 257024
|
3052 |
+
},
|
3053 |
+
{
|
3054 |
+
"epoch": 0.01,
|
3055 |
+
"learning_rate": 4.578324422172183e-05,
|
3056 |
+
"loss": 4.3339,
|
3057 |
+
"step": 257536
|
3058 |
+
},
|
3059 |
+
{
|
3060 |
+
"epoch": 0.01,
|
3061 |
+
"learning_rate": 4.577485827421131e-05,
|
3062 |
+
"loss": 4.3302,
|
3063 |
+
"step": 258048
|
3064 |
+
},
|
3065 |
+
{
|
3066 |
+
"epoch": 0.01,
|
3067 |
+
"learning_rate": 4.576647232670079e-05,
|
3068 |
+
"loss": 4.3315,
|
3069 |
+
"step": 258560
|
3070 |
+
},
|
3071 |
+
{
|
3072 |
+
"epoch": 0.01,
|
3073 |
+
"learning_rate": 4.575808637919027e-05,
|
3074 |
+
"loss": 4.3302,
|
3075 |
+
"step": 259072
|
3076 |
+
},
|
3077 |
+
{
|
3078 |
+
"epoch": 0.01,
|
3079 |
+
"learning_rate": 4.574970043167976e-05,
|
3080 |
+
"loss": 4.322,
|
3081 |
+
"step": 259584
|
3082 |
+
},
|
3083 |
+
{
|
3084 |
+
"epoch": 0.01,
|
3085 |
+
"learning_rate": 4.5741330862972966e-05,
|
3086 |
+
"loss": 4.321,
|
3087 |
+
"step": 260096
|
3088 |
+
},
|
3089 |
+
{
|
3090 |
+
"epoch": 0.01,
|
3091 |
+
"learning_rate": 4.5732944915462446e-05,
|
3092 |
+
"loss": 4.3418,
|
3093 |
+
"step": 260608
|
3094 |
+
},
|
3095 |
+
{
|
3096 |
+
"epoch": 0.01,
|
3097 |
+
"learning_rate": 4.5724558967951926e-05,
|
3098 |
+
"loss": 4.3103,
|
3099 |
+
"step": 261120
|
3100 |
+
},
|
3101 |
+
{
|
3102 |
+
"epoch": 0.01,
|
3103 |
+
"learning_rate": 4.5716173020441406e-05,
|
3104 |
+
"loss": 4.3121,
|
3105 |
+
"step": 261632
|
3106 |
+
},
|
3107 |
+
{
|
3108 |
+
"epoch": 0.01,
|
3109 |
+
"learning_rate": 4.5707787072930886e-05,
|
3110 |
+
"loss": 4.3189,
|
3111 |
+
"step": 262144
|
3112 |
+
},
|
3113 |
+
{
|
3114 |
+
"epoch": 0.01,
|
3115 |
+
"learning_rate": 4.5699401125420366e-05,
|
3116 |
+
"loss": 4.3357,
|
3117 |
+
"step": 262656
|
3118 |
+
},
|
3119 |
+
{
|
3120 |
+
"epoch": 0.01,
|
3121 |
+
"learning_rate": 4.5691015177909846e-05,
|
3122 |
+
"loss": 4.3204,
|
3123 |
+
"step": 263168
|
3124 |
+
},
|
3125 |
+
{
|
3126 |
+
"epoch": 0.01,
|
3127 |
+
"learning_rate": 4.5682629230399326e-05,
|
3128 |
+
"loss": 4.3188,
|
3129 |
+
"step": 263680
|
3130 |
+
},
|
3131 |
+
{
|
3132 |
+
"epoch": 0.01,
|
3133 |
+
"learning_rate": 4.5674259661692535e-05,
|
3134 |
+
"loss": 4.3111,
|
3135 |
+
"step": 264192
|
3136 |
+
},
|
3137 |
+
{
|
3138 |
+
"epoch": 0.01,
|
3139 |
+
"learning_rate": 4.5665873714182015e-05,
|
3140 |
+
"loss": 4.3174,
|
3141 |
+
"step": 264704
|
3142 |
+
},
|
3143 |
+
{
|
3144 |
+
"epoch": 0.01,
|
3145 |
+
"learning_rate": 4.565752052427896e-05,
|
3146 |
+
"loss": 4.3283,
|
3147 |
+
"step": 265216
|
3148 |
+
},
|
3149 |
+
{
|
3150 |
+
"epoch": 0.01,
|
3151 |
+
"learning_rate": 4.564913457676844e-05,
|
3152 |
+
"loss": 4.3217,
|
3153 |
+
"step": 265728
|
3154 |
+
},
|
3155 |
+
{
|
3156 |
+
"epoch": 0.01,
|
3157 |
+
"learning_rate": 4.564074862925792e-05,
|
3158 |
+
"loss": 4.3201,
|
3159 |
+
"step": 266240
|
3160 |
+
},
|
3161 |
+
{
|
3162 |
+
"epoch": 0.01,
|
3163 |
+
"learning_rate": 4.56323626817474e-05,
|
3164 |
+
"loss": 4.3314,
|
3165 |
+
"step": 266752
|
3166 |
+
},
|
3167 |
+
{
|
3168 |
+
"epoch": 0.01,
|
3169 |
+
"learning_rate": 4.562397673423688e-05,
|
3170 |
+
"loss": 4.3293,
|
3171 |
+
"step": 267264
|
3172 |
+
},
|
3173 |
+
{
|
3174 |
+
"epoch": 0.01,
|
3175 |
+
"learning_rate": 4.561559078672636e-05,
|
3176 |
+
"loss": 4.3104,
|
3177 |
+
"step": 267776
|
3178 |
+
},
|
3179 |
+
{
|
3180 |
+
"epoch": 0.01,
|
3181 |
+
"learning_rate": 4.560720483921584e-05,
|
3182 |
+
"loss": 4.3133,
|
3183 |
+
"step": 268288
|
3184 |
+
},
|
3185 |
+
{
|
3186 |
+
"epoch": 0.01,
|
3187 |
+
"learning_rate": 4.559881889170532e-05,
|
3188 |
+
"loss": 4.3262,
|
3189 |
+
"step": 268800
|
3190 |
+
},
|
3191 |
+
{
|
3192 |
+
"epoch": 0.01,
|
3193 |
+
"learning_rate": 4.55904329441948e-05,
|
3194 |
+
"loss": 4.3008,
|
3195 |
+
"step": 269312
|
3196 |
+
},
|
3197 |
+
{
|
3198 |
+
"epoch": 0.01,
|
3199 |
+
"learning_rate": 4.558204699668427e-05,
|
3200 |
+
"loss": 4.3211,
|
3201 |
+
"step": 269824
|
3202 |
+
},
|
3203 |
+
{
|
3204 |
+
"epoch": 0.01,
|
3205 |
+
"learning_rate": 4.557367742797749e-05,
|
3206 |
+
"loss": 4.3068,
|
3207 |
+
"step": 270336
|
3208 |
+
},
|
3209 |
+
{
|
3210 |
+
"epoch": 0.01,
|
3211 |
+
"learning_rate": 4.556529148046697e-05,
|
3212 |
+
"loss": 4.3291,
|
3213 |
+
"step": 270848
|
3214 |
+
},
|
3215 |
+
{
|
3216 |
+
"epoch": 0.01,
|
3217 |
+
"learning_rate": 4.555690553295645e-05,
|
3218 |
+
"loss": 4.3139,
|
3219 |
+
"step": 271360
|
3220 |
+
},
|
3221 |
+
{
|
3222 |
+
"epoch": 0.01,
|
3223 |
+
"learning_rate": 4.554851958544593e-05,
|
3224 |
+
"loss": 4.3049,
|
3225 |
+
"step": 271872
|
3226 |
+
},
|
3227 |
+
{
|
3228 |
+
"epoch": 0.01,
|
3229 |
+
"learning_rate": 4.554013363793541e-05,
|
3230 |
+
"loss": 4.3034,
|
3231 |
+
"step": 272384
|
3232 |
+
},
|
3233 |
+
{
|
3234 |
+
"epoch": 0.01,
|
3235 |
+
"learning_rate": 4.553174769042489e-05,
|
3236 |
+
"loss": 4.3183,
|
3237 |
+
"step": 272896
|
3238 |
+
},
|
3239 |
+
{
|
3240 |
+
"epoch": 0.01,
|
3241 |
+
"learning_rate": 4.552336174291437e-05,
|
3242 |
+
"loss": 4.3232,
|
3243 |
+
"step": 273408
|
3244 |
+
},
|
3245 |
+
{
|
3246 |
+
"epoch": 0.01,
|
3247 |
+
"learning_rate": 4.551499217420758e-05,
|
3248 |
+
"loss": 4.3123,
|
3249 |
+
"step": 273920
|
3250 |
+
},
|
3251 |
+
{
|
3252 |
+
"epoch": 0.01,
|
3253 |
+
"learning_rate": 4.550660622669706e-05,
|
3254 |
+
"loss": 4.3112,
|
3255 |
+
"step": 274432
|
3256 |
+
},
|
3257 |
+
{
|
3258 |
+
"epoch": 0.02,
|
3259 |
+
"learning_rate": 4.549822027918654e-05,
|
3260 |
+
"loss": 4.3046,
|
3261 |
+
"step": 274944
|
3262 |
+
},
|
3263 |
+
{
|
3264 |
+
"epoch": 0.02,
|
3265 |
+
"learning_rate": 4.548983433167602e-05,
|
3266 |
+
"loss": 4.3163,
|
3267 |
+
"step": 275456
|
3268 |
+
},
|
3269 |
+
{
|
3270 |
+
"epoch": 0.02,
|
3271 |
+
"learning_rate": 4.5481464762969226e-05,
|
3272 |
+
"loss": 4.3062,
|
3273 |
+
"step": 275968
|
3274 |
+
},
|
3275 |
+
{
|
3276 |
+
"epoch": 0.02,
|
3277 |
+
"learning_rate": 4.547309519426244e-05,
|
3278 |
+
"loss": 4.3058,
|
3279 |
+
"step": 276480
|
3280 |
+
},
|
3281 |
+
{
|
3282 |
+
"epoch": 0.02,
|
3283 |
+
"learning_rate": 4.546470924675192e-05,
|
3284 |
+
"loss": 4.3148,
|
3285 |
+
"step": 276992
|
3286 |
+
},
|
3287 |
+
{
|
3288 |
+
"epoch": 0.02,
|
3289 |
+
"learning_rate": 4.54563232992414e-05,
|
3290 |
+
"loss": 4.3134,
|
3291 |
+
"step": 277504
|
3292 |
+
},
|
3293 |
+
{
|
3294 |
+
"epoch": 0.02,
|
3295 |
+
"learning_rate": 4.544793735173088e-05,
|
3296 |
+
"loss": 4.3097,
|
3297 |
+
"step": 278016
|
3298 |
+
},
|
3299 |
+
{
|
3300 |
+
"epoch": 0.02,
|
3301 |
+
"learning_rate": 4.543955140422036e-05,
|
3302 |
+
"loss": 4.2963,
|
3303 |
+
"step": 278528
|
3304 |
+
},
|
3305 |
+
{
|
3306 |
+
"epoch": 0.02,
|
3307 |
+
"learning_rate": 4.543116545670984e-05,
|
3308 |
+
"loss": 4.3048,
|
3309 |
+
"step": 279040
|
3310 |
+
},
|
3311 |
+
{
|
3312 |
+
"epoch": 0.02,
|
3313 |
+
"learning_rate": 4.542277950919932e-05,
|
3314 |
+
"loss": 4.3037,
|
3315 |
+
"step": 279552
|
3316 |
+
},
|
3317 |
+
{
|
3318 |
+
"epoch": 0.02,
|
3319 |
+
"learning_rate": 4.54143935616888e-05,
|
3320 |
+
"loss": 4.3065,
|
3321 |
+
"step": 280064
|
3322 |
+
},
|
3323 |
+
{
|
3324 |
+
"epoch": 0.02,
|
3325 |
+
"learning_rate": 4.540602399298201e-05,
|
3326 |
+
"loss": 4.3049,
|
3327 |
+
"step": 280576
|
3328 |
+
},
|
3329 |
+
{
|
3330 |
+
"epoch": 0.02,
|
3331 |
+
"learning_rate": 4.539763804547149e-05,
|
3332 |
+
"loss": 4.2962,
|
3333 |
+
"step": 281088
|
3334 |
+
},
|
3335 |
+
{
|
3336 |
+
"epoch": 0.02,
|
3337 |
+
"learning_rate": 4.538925209796097e-05,
|
3338 |
+
"loss": 4.3005,
|
3339 |
+
"step": 281600
|
3340 |
+
},
|
3341 |
+
{
|
3342 |
+
"epoch": 0.02,
|
3343 |
+
"learning_rate": 4.538086615045045e-05,
|
3344 |
+
"loss": 4.3052,
|
3345 |
+
"step": 282112
|
3346 |
+
},
|
3347 |
+
{
|
3348 |
+
"epoch": 0.02,
|
3349 |
+
"learning_rate": 4.537248020293993e-05,
|
3350 |
+
"loss": 4.2811,
|
3351 |
+
"step": 282624
|
3352 |
+
},
|
3353 |
+
{
|
3354 |
+
"epoch": 0.02,
|
3355 |
+
"learning_rate": 4.536409425542941e-05,
|
3356 |
+
"loss": 4.3005,
|
3357 |
+
"step": 283136
|
3358 |
+
},
|
3359 |
+
{
|
3360 |
+
"epoch": 0.02,
|
3361 |
+
"learning_rate": 4.535572468672262e-05,
|
3362 |
+
"loss": 4.2857,
|
3363 |
+
"step": 283648
|
3364 |
+
},
|
3365 |
+
{
|
3366 |
+
"epoch": 0.02,
|
3367 |
+
"learning_rate": 4.53473387392121e-05,
|
3368 |
+
"loss": 4.3038,
|
3369 |
+
"step": 284160
|
3370 |
+
},
|
3371 |
+
{
|
3372 |
+
"epoch": 0.02,
|
3373 |
+
"learning_rate": 4.533895279170158e-05,
|
3374 |
+
"loss": 4.2959,
|
3375 |
+
"step": 284672
|
3376 |
+
},
|
3377 |
+
{
|
3378 |
+
"epoch": 0.02,
|
3379 |
+
"learning_rate": 4.533056684419107e-05,
|
3380 |
+
"loss": 4.2995,
|
3381 |
+
"step": 285184
|
3382 |
+
},
|
3383 |
+
{
|
3384 |
+
"epoch": 0.02,
|
3385 |
+
"learning_rate": 4.5322197275484276e-05,
|
3386 |
+
"loss": 4.2916,
|
3387 |
+
"step": 285696
|
3388 |
+
},
|
3389 |
+
{
|
3390 |
+
"epoch": 0.02,
|
3391 |
+
"learning_rate": 4.5313811327973756e-05,
|
3392 |
+
"loss": 4.2954,
|
3393 |
+
"step": 286208
|
3394 |
+
},
|
3395 |
+
{
|
3396 |
+
"epoch": 0.02,
|
3397 |
+
"learning_rate": 4.5305425380463236e-05,
|
3398 |
+
"loss": 4.3044,
|
3399 |
+
"step": 286720
|
3400 |
+
},
|
3401 |
+
{
|
3402 |
+
"epoch": 0.02,
|
3403 |
+
"learning_rate": 4.5297039432952716e-05,
|
3404 |
+
"loss": 4.288,
|
3405 |
+
"step": 287232
|
3406 |
+
},
|
3407 |
+
{
|
3408 |
+
"epoch": 0.02,
|
3409 |
+
"learning_rate": 4.5288669864245925e-05,
|
3410 |
+
"loss": 4.3001,
|
3411 |
+
"step": 287744
|
3412 |
+
},
|
3413 |
+
{
|
3414 |
+
"epoch": 0.02,
|
3415 |
+
"learning_rate": 4.5280283916735405e-05,
|
3416 |
+
"loss": 4.3,
|
3417 |
+
"step": 288256
|
3418 |
+
},
|
3419 |
+
{
|
3420 |
+
"epoch": 0.02,
|
3421 |
+
"learning_rate": 4.5271914348028614e-05,
|
3422 |
+
"loss": 4.2982,
|
3423 |
+
"step": 288768
|
3424 |
+
},
|
3425 |
+
{
|
3426 |
+
"epoch": 0.02,
|
3427 |
+
"learning_rate": 4.5263528400518094e-05,
|
3428 |
+
"loss": 4.2998,
|
3429 |
+
"step": 289280
|
3430 |
+
},
|
3431 |
+
{
|
3432 |
+
"epoch": 0.02,
|
3433 |
+
"learning_rate": 4.5255142453007574e-05,
|
3434 |
+
"loss": 4.283,
|
3435 |
+
"step": 289792
|
3436 |
+
},
|
3437 |
+
{
|
3438 |
+
"epoch": 0.02,
|
3439 |
+
"learning_rate": 4.5246756505497054e-05,
|
3440 |
+
"loss": 4.3028,
|
3441 |
+
"step": 290304
|
3442 |
+
},
|
3443 |
+
{
|
3444 |
+
"epoch": 0.02,
|
3445 |
+
"learning_rate": 4.5238370557986534e-05,
|
3446 |
+
"loss": 4.2891,
|
3447 |
+
"step": 290816
|
3448 |
+
},
|
3449 |
+
{
|
3450 |
+
"epoch": 0.02,
|
3451 |
+
"learning_rate": 4.522998461047602e-05,
|
3452 |
+
"loss": 4.2976,
|
3453 |
+
"step": 291328
|
3454 |
+
},
|
3455 |
+
{
|
3456 |
+
"epoch": 0.02,
|
3457 |
+
"learning_rate": 4.52215986629655e-05,
|
3458 |
+
"loss": 4.2863,
|
3459 |
+
"step": 291840
|
3460 |
+
},
|
3461 |
+
{
|
3462 |
+
"epoch": 0.02,
|
3463 |
+
"learning_rate": 4.521321271545498e-05,
|
3464 |
+
"loss": 4.2921,
|
3465 |
+
"step": 292352
|
3466 |
+
},
|
3467 |
+
{
|
3468 |
+
"epoch": 0.02,
|
3469 |
+
"learning_rate": 4.520484314674819e-05,
|
3470 |
+
"loss": 4.2905,
|
3471 |
+
"step": 292864
|
3472 |
+
},
|
3473 |
+
{
|
3474 |
+
"epoch": 0.02,
|
3475 |
+
"learning_rate": 4.519645719923767e-05,
|
3476 |
+
"loss": 4.2857,
|
3477 |
+
"step": 293376
|
3478 |
+
},
|
3479 |
+
{
|
3480 |
+
"epoch": 0.02,
|
3481 |
+
"learning_rate": 4.518807125172715e-05,
|
3482 |
+
"loss": 4.2732,
|
3483 |
+
"step": 293888
|
3484 |
+
},
|
3485 |
+
{
|
3486 |
+
"epoch": 0.02,
|
3487 |
+
"learning_rate": 4.517968530421663e-05,
|
3488 |
+
"loss": 4.2849,
|
3489 |
+
"step": 294400
|
3490 |
+
},
|
3491 |
+
{
|
3492 |
+
"epoch": 0.02,
|
3493 |
+
"learning_rate": 4.517129935670611e-05,
|
3494 |
+
"loss": 4.2796,
|
3495 |
+
"step": 294912
|
3496 |
+
},
|
3497 |
+
{
|
3498 |
+
"epoch": 0.02,
|
3499 |
+
"learning_rate": 4.516291340919558e-05,
|
3500 |
+
"loss": 4.2903,
|
3501 |
+
"step": 295424
|
3502 |
+
},
|
3503 |
+
{
|
3504 |
+
"epoch": 0.02,
|
3505 |
+
"learning_rate": 4.515452746168506e-05,
|
3506 |
+
"loss": 4.3007,
|
3507 |
+
"step": 295936
|
3508 |
+
},
|
3509 |
+
{
|
3510 |
+
"epoch": 0.02,
|
3511 |
+
"learning_rate": 4.514615789297828e-05,
|
3512 |
+
"loss": 4.2982,
|
3513 |
+
"step": 296448
|
3514 |
+
},
|
3515 |
+
{
|
3516 |
+
"epoch": 0.02,
|
3517 |
+
"learning_rate": 4.513777194546776e-05,
|
3518 |
+
"loss": 4.27,
|
3519 |
+
"step": 296960
|
3520 |
+
},
|
3521 |
+
{
|
3522 |
+
"epoch": 0.02,
|
3523 |
+
"learning_rate": 4.512938599795724e-05,
|
3524 |
+
"loss": 4.2952,
|
3525 |
+
"step": 297472
|
3526 |
+
},
|
3527 |
+
{
|
3528 |
+
"epoch": 0.02,
|
3529 |
+
"learning_rate": 4.512100005044672e-05,
|
3530 |
+
"loss": 4.286,
|
3531 |
+
"step": 297984
|
3532 |
+
},
|
3533 |
+
{
|
3534 |
+
"epoch": 0.02,
|
3535 |
+
"learning_rate": 4.51126141029362e-05,
|
3536 |
+
"loss": 4.2854,
|
3537 |
+
"step": 298496
|
3538 |
+
},
|
3539 |
+
{
|
3540 |
+
"epoch": 0.02,
|
3541 |
+
"learning_rate": 4.510422815542568e-05,
|
3542 |
+
"loss": 4.2795,
|
3543 |
+
"step": 299008
|
3544 |
+
},
|
3545 |
+
{
|
3546 |
+
"epoch": 0.02,
|
3547 |
+
"learning_rate": 4.509584220791516e-05,
|
3548 |
+
"loss": 4.2886,
|
3549 |
+
"step": 299520
|
3550 |
+
},
|
3551 |
+
{
|
3552 |
+
"epoch": 0.02,
|
3553 |
+
"learning_rate": 4.508745626040464e-05,
|
3554 |
+
"loss": 4.2796,
|
3555 |
+
"step": 300032
|
3556 |
+
},
|
3557 |
+
{
|
3558 |
+
"epoch": 0.02,
|
3559 |
+
"learning_rate": 4.507908669169785e-05,
|
3560 |
+
"loss": 4.2879,
|
3561 |
+
"step": 300544
|
3562 |
+
},
|
3563 |
+
{
|
3564 |
+
"epoch": 0.02,
|
3565 |
+
"learning_rate": 4.507070074418733e-05,
|
3566 |
+
"loss": 4.2809,
|
3567 |
+
"step": 301056
|
3568 |
+
},
|
3569 |
+
{
|
3570 |
+
"epoch": 0.02,
|
3571 |
+
"learning_rate": 4.5062331175480536e-05,
|
3572 |
+
"loss": 4.2842,
|
3573 |
+
"step": 301568
|
3574 |
+
},
|
3575 |
+
{
|
3576 |
+
"epoch": 0.02,
|
3577 |
+
"learning_rate": 4.5053945227970016e-05,
|
3578 |
+
"loss": 4.2868,
|
3579 |
+
"step": 302080
|
3580 |
+
},
|
3581 |
+
{
|
3582 |
+
"epoch": 0.02,
|
3583 |
+
"learning_rate": 4.5045559280459496e-05,
|
3584 |
+
"loss": 4.2788,
|
3585 |
+
"step": 302592
|
3586 |
+
},
|
3587 |
+
{
|
3588 |
+
"epoch": 0.02,
|
3589 |
+
"learning_rate": 4.5037173332948976e-05,
|
3590 |
+
"loss": 4.2823,
|
3591 |
+
"step": 303104
|
3592 |
+
},
|
3593 |
+
{
|
3594 |
+
"epoch": 0.02,
|
3595 |
+
"learning_rate": 4.502880376424219e-05,
|
3596 |
+
"loss": 4.2913,
|
3597 |
+
"step": 303616
|
3598 |
+
},
|
3599 |
+
{
|
3600 |
+
"epoch": 0.02,
|
3601 |
+
"learning_rate": 4.502041781673167e-05,
|
3602 |
+
"loss": 4.2731,
|
3603 |
+
"step": 304128
|
3604 |
+
},
|
3605 |
+
{
|
3606 |
+
"epoch": 0.02,
|
3607 |
+
"learning_rate": 4.501203186922115e-05,
|
3608 |
+
"loss": 4.2894,
|
3609 |
+
"step": 304640
|
3610 |
+
},
|
3611 |
+
{
|
3612 |
+
"epoch": 0.02,
|
3613 |
+
"learning_rate": 4.500364592171063e-05,
|
3614 |
+
"loss": 4.282,
|
3615 |
+
"step": 305152
|
3616 |
+
},
|
3617 |
+
{
|
3618 |
+
"epoch": 0.03,
|
3619 |
+
"eval_loss": 4.2855424880981445,
|
3620 |
+
"eval_runtime": 283.2935,
|
3621 |
+
"eval_samples_per_second": 1346.981,
|
3622 |
+
"eval_steps_per_second": 42.094,
|
3623 |
+
"step": 305280
|
3624 |
+
}
|
3625 |
+
],
|
3626 |
+
"logging_steps": 512,
|
3627 |
+
"max_steps": 3052726,
|
3628 |
+
"num_train_epochs": 9223372036854775807,
|
3629 |
+
"save_steps": 10,
|
3630 |
+
"total_flos": 1.2223819719679363e+17,
|
3631 |
+
"trial_name": null,
|
3632 |
+
"trial_params": null
|
3633 |
+
}
|
checkpoint-305280/training_args.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:5a536f20432ae4561f25e7b014a00f428a82b0d95b63c7ada115a3a9f369d5db
|
3 |
+
size 4219
|