mathislucka
commited on
Commit
•
a10d0bd
1
Parent(s):
2c15ccf
add model
Browse files- added_tokens.json +3 -0
- config.json +41 -0
- optimizer.pt +3 -0
- pytorch_model.bin +3 -0
- rng_state.pth +3 -0
- scaler.pt +3 -0
- scheduler.pt +3 -0
- special_tokens_map.json +9 -0
- spm.model +3 -0
- tokenizer.json +0 -0
- tokenizer_config.json +17 -0
- trainer_state.json +3529 -0
- training_args.bin +3 -0
added_tokens.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"[MASK]": 128000
|
| 3 |
+
}
|
config.json
ADDED
|
@@ -0,0 +1,41 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "microsoft/deberta-v3-large",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"DebertaV2ForSequenceClassification"
|
| 5 |
+
],
|
| 6 |
+
"attention_probs_dropout_prob": 0.1,
|
| 7 |
+
"hidden_act": "gelu",
|
| 8 |
+
"hidden_dropout_prob": 0.1,
|
| 9 |
+
"hidden_size": 1024,
|
| 10 |
+
"id2label": {
|
| 11 |
+
"0": "LABEL_0"
|
| 12 |
+
},
|
| 13 |
+
"initializer_range": 0.02,
|
| 14 |
+
"intermediate_size": 4096,
|
| 15 |
+
"label2id": {
|
| 16 |
+
"LABEL_0": 0
|
| 17 |
+
},
|
| 18 |
+
"layer_norm_eps": 1e-07,
|
| 19 |
+
"max_position_embeddings": 512,
|
| 20 |
+
"max_relative_positions": -1,
|
| 21 |
+
"model_type": "deberta-v2",
|
| 22 |
+
"norm_rel_ebd": "layer_norm",
|
| 23 |
+
"num_attention_heads": 16,
|
| 24 |
+
"num_hidden_layers": 24,
|
| 25 |
+
"pad_token_id": 0,
|
| 26 |
+
"pooler_dropout": 0,
|
| 27 |
+
"pooler_hidden_act": "gelu",
|
| 28 |
+
"pooler_hidden_size": 1024,
|
| 29 |
+
"pos_att_type": [
|
| 30 |
+
"p2c",
|
| 31 |
+
"c2p"
|
| 32 |
+
],
|
| 33 |
+
"position_biased_input": false,
|
| 34 |
+
"position_buckets": 256,
|
| 35 |
+
"relative_attention": true,
|
| 36 |
+
"share_att_key": true,
|
| 37 |
+
"torch_dtype": "float32",
|
| 38 |
+
"transformers_version": "4.25.1",
|
| 39 |
+
"type_vocab_size": 0,
|
| 40 |
+
"vocab_size": 128100
|
| 41 |
+
}
|
optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:de94e6ab2aaa08b6d7324d01aeb0659e2c857a6d85166598a880429ba9153881
|
| 3 |
+
size 3480732377
|
pytorch_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:088ad91836d1b2d9fcee4fc983ef6c8dde367e23cd13cd87f66742bfda6807d8
|
| 3 |
+
size 1740389291
|
rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1fd840ee95617fd38136faf9a75aad6982700347f2a8f636e47b9709dba824f0
|
| 3 |
+
size 14503
|
scaler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d3b4058dd8e03e9cf790b5323cb6c1d8cab49afd2d26a75b337f76bd92010b14
|
| 3 |
+
size 559
|
scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:81f3ecb1bd9130fd1b1f4863e998d4bd3fb3e15eb7691f64f31ac59bea55d2c3
|
| 3 |
+
size 623
|
special_tokens_map.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token": "[CLS]",
|
| 3 |
+
"cls_token": "[CLS]",
|
| 4 |
+
"eos_token": "[SEP]",
|
| 5 |
+
"mask_token": "[MASK]",
|
| 6 |
+
"pad_token": "[PAD]",
|
| 7 |
+
"sep_token": "[SEP]",
|
| 8 |
+
"unk_token": "[UNK]"
|
| 9 |
+
}
|
spm.model
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c679fbf93643d19aab7ee10c0b99e460bdbc02fedf34b92b05af343b4af586fd
|
| 3 |
+
size 2464616
|
tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1,17 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token": "[CLS]",
|
| 3 |
+
"cls_token": "[CLS]",
|
| 4 |
+
"do_lower_case": false,
|
| 5 |
+
"eos_token": "[SEP]",
|
| 6 |
+
"mask_token": "[MASK]",
|
| 7 |
+
"model_max_length": 1000000000000000019884624838656,
|
| 8 |
+
"name_or_path": "microsoft/deberta-v3-large",
|
| 9 |
+
"pad_token": "[PAD]",
|
| 10 |
+
"sep_token": "[SEP]",
|
| 11 |
+
"sp_model_kwargs": {},
|
| 12 |
+
"special_tokens_map_file": null,
|
| 13 |
+
"split_by_punct": false,
|
| 14 |
+
"tokenizer_class": "DebertaV2Tokenizer",
|
| 15 |
+
"unk_token": "[UNK]",
|
| 16 |
+
"vocab_type": "spm"
|
| 17 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,3529 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 1.2412053346633911,
|
| 3 |
+
"best_model_checkpoint": "./models/deberta-v3-large-all-v2/checkpoint-292273",
|
| 4 |
+
"epoch": 1.0,
|
| 5 |
+
"global_step": 292273,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.0,
|
| 12 |
+
"learning_rate": 6.988120695377267e-06,
|
| 13 |
+
"loss": 0.8116,
|
| 14 |
+
"step": 500
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.0,
|
| 18 |
+
"learning_rate": 6.976145589910803e-06,
|
| 19 |
+
"loss": 0.3656,
|
| 20 |
+
"step": 1000
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.01,
|
| 24 |
+
"learning_rate": 6.964170484444338e-06,
|
| 25 |
+
"loss": 0.3417,
|
| 26 |
+
"step": 1500
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.01,
|
| 30 |
+
"learning_rate": 6.952195378977873e-06,
|
| 31 |
+
"loss": 0.3577,
|
| 32 |
+
"step": 2000
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.01,
|
| 36 |
+
"learning_rate": 6.940244223722342e-06,
|
| 37 |
+
"loss": 0.3302,
|
| 38 |
+
"step": 2500
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.01,
|
| 42 |
+
"learning_rate": 6.928269118255877e-06,
|
| 43 |
+
"loss": 0.3478,
|
| 44 |
+
"step": 3000
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.01,
|
| 48 |
+
"learning_rate": 6.916317963000345e-06,
|
| 49 |
+
"loss": 0.3343,
|
| 50 |
+
"step": 3500
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.01,
|
| 54 |
+
"learning_rate": 6.904342857533881e-06,
|
| 55 |
+
"loss": 0.3814,
|
| 56 |
+
"step": 4000
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.02,
|
| 60 |
+
"learning_rate": 6.892367752067416e-06,
|
| 61 |
+
"loss": 0.2832,
|
| 62 |
+
"step": 4500
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.02,
|
| 66 |
+
"learning_rate": 6.880392646600952e-06,
|
| 67 |
+
"loss": 0.3181,
|
| 68 |
+
"step": 5000
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.02,
|
| 72 |
+
"learning_rate": 6.86844149134542e-06,
|
| 73 |
+
"loss": 0.2855,
|
| 74 |
+
"step": 5500
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.02,
|
| 78 |
+
"learning_rate": 6.856466385878955e-06,
|
| 79 |
+
"loss": 0.32,
|
| 80 |
+
"step": 6000
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.02,
|
| 84 |
+
"learning_rate": 6.844491280412491e-06,
|
| 85 |
+
"loss": 0.3108,
|
| 86 |
+
"step": 6500
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.02,
|
| 90 |
+
"learning_rate": 6.832516174946026e-06,
|
| 91 |
+
"loss": 0.3116,
|
| 92 |
+
"step": 7000
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 0.03,
|
| 96 |
+
"learning_rate": 6.820541069479562e-06,
|
| 97 |
+
"loss": 0.2832,
|
| 98 |
+
"step": 7500
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.03,
|
| 102 |
+
"learning_rate": 6.808565964013098e-06,
|
| 103 |
+
"loss": 0.333,
|
| 104 |
+
"step": 8000
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 0.03,
|
| 108 |
+
"learning_rate": 6.796590858546633e-06,
|
| 109 |
+
"loss": 0.3092,
|
| 110 |
+
"step": 8500
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.03,
|
| 114 |
+
"learning_rate": 6.784615753080168e-06,
|
| 115 |
+
"loss": 0.2621,
|
| 116 |
+
"step": 9000
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 0.03,
|
| 120 |
+
"learning_rate": 6.772664597824636e-06,
|
| 121 |
+
"loss": 0.3031,
|
| 122 |
+
"step": 9500
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.03,
|
| 126 |
+
"learning_rate": 6.760689492358172e-06,
|
| 127 |
+
"loss": 0.2695,
|
| 128 |
+
"step": 10000
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.04,
|
| 132 |
+
"learning_rate": 6.748714386891708e-06,
|
| 133 |
+
"loss": 0.2696,
|
| 134 |
+
"step": 10500
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 0.04,
|
| 138 |
+
"learning_rate": 6.736739281425243e-06,
|
| 139 |
+
"loss": 0.3249,
|
| 140 |
+
"step": 11000
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 0.04,
|
| 144 |
+
"learning_rate": 6.724764175958778e-06,
|
| 145 |
+
"loss": 0.3205,
|
| 146 |
+
"step": 11500
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.04,
|
| 150 |
+
"learning_rate": 6.712789070492313e-06,
|
| 151 |
+
"loss": 0.28,
|
| 152 |
+
"step": 12000
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 0.04,
|
| 156 |
+
"learning_rate": 6.700861865447715e-06,
|
| 157 |
+
"loss": 0.2783,
|
| 158 |
+
"step": 12500
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 0.04,
|
| 162 |
+
"learning_rate": 6.688886759981251e-06,
|
| 163 |
+
"loss": 0.3118,
|
| 164 |
+
"step": 13000
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.05,
|
| 168 |
+
"learning_rate": 6.676911654514786e-06,
|
| 169 |
+
"loss": 0.295,
|
| 170 |
+
"step": 13500
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.05,
|
| 174 |
+
"learning_rate": 6.6649365490483205e-06,
|
| 175 |
+
"loss": 0.2906,
|
| 176 |
+
"step": 14000
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 0.05,
|
| 180 |
+
"learning_rate": 6.652961443581856e-06,
|
| 181 |
+
"loss": 0.2481,
|
| 182 |
+
"step": 14500
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 0.05,
|
| 186 |
+
"learning_rate": 6.640986338115392e-06,
|
| 187 |
+
"loss": 0.3028,
|
| 188 |
+
"step": 15000
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 0.05,
|
| 192 |
+
"learning_rate": 6.629011232648928e-06,
|
| 193 |
+
"loss": 0.2575,
|
| 194 |
+
"step": 15500
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 0.05,
|
| 198 |
+
"learning_rate": 6.617036127182463e-06,
|
| 199 |
+
"loss": 0.2703,
|
| 200 |
+
"step": 16000
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 0.06,
|
| 204 |
+
"learning_rate": 6.605084971926931e-06,
|
| 205 |
+
"loss": 0.2849,
|
| 206 |
+
"step": 16500
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.06,
|
| 210 |
+
"learning_rate": 6.593109866460466e-06,
|
| 211 |
+
"loss": 0.3124,
|
| 212 |
+
"step": 17000
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.06,
|
| 216 |
+
"learning_rate": 6.581134760994002e-06,
|
| 217 |
+
"loss": 0.307,
|
| 218 |
+
"step": 17500
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 0.06,
|
| 222 |
+
"learning_rate": 6.569159655527538e-06,
|
| 223 |
+
"loss": 0.2734,
|
| 224 |
+
"step": 18000
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 0.06,
|
| 228 |
+
"learning_rate": 6.557184550061073e-06,
|
| 229 |
+
"loss": 0.2896,
|
| 230 |
+
"step": 18500
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 0.07,
|
| 234 |
+
"learning_rate": 6.545233394805541e-06,
|
| 235 |
+
"loss": 0.2735,
|
| 236 |
+
"step": 19000
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 0.07,
|
| 240 |
+
"learning_rate": 6.533258289339076e-06,
|
| 241 |
+
"loss": 0.31,
|
| 242 |
+
"step": 19500
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 0.07,
|
| 246 |
+
"learning_rate": 6.521283183872612e-06,
|
| 247 |
+
"loss": 0.2829,
|
| 248 |
+
"step": 20000
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.07,
|
| 252 |
+
"learning_rate": 6.509308078406148e-06,
|
| 253 |
+
"loss": 0.2994,
|
| 254 |
+
"step": 20500
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.07,
|
| 258 |
+
"learning_rate": 6.497356923150616e-06,
|
| 259 |
+
"loss": 0.3024,
|
| 260 |
+
"step": 21000
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 0.07,
|
| 264 |
+
"learning_rate": 6.485381817684151e-06,
|
| 265 |
+
"loss": 0.2991,
|
| 266 |
+
"step": 21500
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 0.08,
|
| 270 |
+
"learning_rate": 6.473406712217687e-06,
|
| 271 |
+
"loss": 0.295,
|
| 272 |
+
"step": 22000
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 0.08,
|
| 276 |
+
"learning_rate": 6.461431606751222e-06,
|
| 277 |
+
"loss": 0.2697,
|
| 278 |
+
"step": 22500
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 0.08,
|
| 282 |
+
"learning_rate": 6.4494565012847576e-06,
|
| 283 |
+
"loss": 0.297,
|
| 284 |
+
"step": 23000
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 0.08,
|
| 288 |
+
"learning_rate": 6.437505346029226e-06,
|
| 289 |
+
"loss": 0.2465,
|
| 290 |
+
"step": 23500
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.08,
|
| 294 |
+
"learning_rate": 6.425530240562761e-06,
|
| 295 |
+
"loss": 0.2666,
|
| 296 |
+
"step": 24000
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.08,
|
| 300 |
+
"learning_rate": 6.413555135096297e-06,
|
| 301 |
+
"loss": 0.2885,
|
| 302 |
+
"step": 24500
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 0.09,
|
| 306 |
+
"learning_rate": 6.401580029629832e-06,
|
| 307 |
+
"loss": 0.285,
|
| 308 |
+
"step": 25000
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 0.09,
|
| 312 |
+
"learning_rate": 6.3896049241633675e-06,
|
| 313 |
+
"loss": 0.2798,
|
| 314 |
+
"step": 25500
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 0.09,
|
| 318 |
+
"learning_rate": 6.377629818696903e-06,
|
| 319 |
+
"loss": 0.3039,
|
| 320 |
+
"step": 26000
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 0.09,
|
| 324 |
+
"learning_rate": 6.365654713230439e-06,
|
| 325 |
+
"loss": 0.2416,
|
| 326 |
+
"step": 26500
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 0.09,
|
| 330 |
+
"learning_rate": 6.353679607763975e-06,
|
| 331 |
+
"loss": 0.3344,
|
| 332 |
+
"step": 27000
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.09,
|
| 336 |
+
"learning_rate": 6.341728452508442e-06,
|
| 337 |
+
"loss": 0.257,
|
| 338 |
+
"step": 27500
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.1,
|
| 342 |
+
"learning_rate": 6.3297533470419775e-06,
|
| 343 |
+
"loss": 0.2741,
|
| 344 |
+
"step": 28000
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 0.1,
|
| 348 |
+
"learning_rate": 6.317778241575513e-06,
|
| 349 |
+
"loss": 0.2977,
|
| 350 |
+
"step": 28500
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 0.1,
|
| 354 |
+
"learning_rate": 6.305827086319982e-06,
|
| 355 |
+
"loss": 0.3057,
|
| 356 |
+
"step": 29000
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 0.1,
|
| 360 |
+
"learning_rate": 6.293851980853517e-06,
|
| 361 |
+
"loss": 0.2691,
|
| 362 |
+
"step": 29500
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 0.1,
|
| 366 |
+
"learning_rate": 6.2818768753870526e-06,
|
| 367 |
+
"loss": 0.2541,
|
| 368 |
+
"step": 30000
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 0.1,
|
| 372 |
+
"learning_rate": 6.2699017699205874e-06,
|
| 373 |
+
"loss": 0.2876,
|
| 374 |
+
"step": 30500
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 0.11,
|
| 378 |
+
"learning_rate": 6.257926664454123e-06,
|
| 379 |
+
"loss": 0.2978,
|
| 380 |
+
"step": 31000
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 0.11,
|
| 384 |
+
"learning_rate": 6.245975509198592e-06,
|
| 385 |
+
"loss": 0.2845,
|
| 386 |
+
"step": 31500
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 0.11,
|
| 390 |
+
"learning_rate": 6.234000403732127e-06,
|
| 391 |
+
"loss": 0.2955,
|
| 392 |
+
"step": 32000
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 0.11,
|
| 396 |
+
"learning_rate": 6.222049248476596e-06,
|
| 397 |
+
"loss": 0.2824,
|
| 398 |
+
"step": 32500
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 0.11,
|
| 402 |
+
"learning_rate": 6.210074143010131e-06,
|
| 403 |
+
"loss": 0.234,
|
| 404 |
+
"step": 33000
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 0.11,
|
| 408 |
+
"learning_rate": 6.198099037543666e-06,
|
| 409 |
+
"loss": 0.2723,
|
| 410 |
+
"step": 33500
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 0.12,
|
| 414 |
+
"learning_rate": 6.186123932077202e-06,
|
| 415 |
+
"loss": 0.3273,
|
| 416 |
+
"step": 34000
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 0.12,
|
| 420 |
+
"learning_rate": 6.174148826610737e-06,
|
| 421 |
+
"loss": 0.2879,
|
| 422 |
+
"step": 34500
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 0.12,
|
| 426 |
+
"learning_rate": 6.1621737211442725e-06,
|
| 427 |
+
"loss": 0.2364,
|
| 428 |
+
"step": 35000
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 0.12,
|
| 432 |
+
"learning_rate": 6.150198615677808e-06,
|
| 433 |
+
"loss": 0.2609,
|
| 434 |
+
"step": 35500
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 0.12,
|
| 438 |
+
"learning_rate": 6.138223510211343e-06,
|
| 439 |
+
"loss": 0.2811,
|
| 440 |
+
"step": 36000
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 0.12,
|
| 444 |
+
"learning_rate": 6.126272354955812e-06,
|
| 445 |
+
"loss": 0.2921,
|
| 446 |
+
"step": 36500
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 0.13,
|
| 450 |
+
"learning_rate": 6.11432119970028e-06,
|
| 451 |
+
"loss": 0.3001,
|
| 452 |
+
"step": 37000
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 0.13,
|
| 456 |
+
"learning_rate": 6.1023460942338155e-06,
|
| 457 |
+
"loss": 0.2618,
|
| 458 |
+
"step": 37500
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 0.13,
|
| 462 |
+
"learning_rate": 6.090370988767351e-06,
|
| 463 |
+
"loss": 0.2811,
|
| 464 |
+
"step": 38000
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 0.13,
|
| 468 |
+
"learning_rate": 6.078395883300887e-06,
|
| 469 |
+
"loss": 0.2683,
|
| 470 |
+
"step": 38500
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 0.13,
|
| 474 |
+
"learning_rate": 6.066420777834422e-06,
|
| 475 |
+
"loss": 0.2466,
|
| 476 |
+
"step": 39000
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 0.14,
|
| 480 |
+
"learning_rate": 6.0544456723679575e-06,
|
| 481 |
+
"loss": 0.2514,
|
| 482 |
+
"step": 39500
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 0.14,
|
| 486 |
+
"learning_rate": 6.042470566901492e-06,
|
| 487 |
+
"loss": 0.2994,
|
| 488 |
+
"step": 40000
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 0.14,
|
| 492 |
+
"learning_rate": 6.030495461435028e-06,
|
| 493 |
+
"loss": 0.2532,
|
| 494 |
+
"step": 40500
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 0.14,
|
| 498 |
+
"learning_rate": 6.018520355968563e-06,
|
| 499 |
+
"loss": 0.2966,
|
| 500 |
+
"step": 41000
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 0.14,
|
| 504 |
+
"learning_rate": 6.006569200713032e-06,
|
| 505 |
+
"loss": 0.26,
|
| 506 |
+
"step": 41500
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 0.14,
|
| 510 |
+
"learning_rate": 5.9945940952465675e-06,
|
| 511 |
+
"loss": 0.2524,
|
| 512 |
+
"step": 42000
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 0.15,
|
| 516 |
+
"learning_rate": 5.982618989780102e-06,
|
| 517 |
+
"loss": 0.3518,
|
| 518 |
+
"step": 42500
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 0.15,
|
| 522 |
+
"learning_rate": 5.970667834524571e-06,
|
| 523 |
+
"loss": 0.4247,
|
| 524 |
+
"step": 43000
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 0.15,
|
| 528 |
+
"learning_rate": 5.95871667926904e-06,
|
| 529 |
+
"loss": 0.4605,
|
| 530 |
+
"step": 43500
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 0.15,
|
| 534 |
+
"learning_rate": 5.946765524013508e-06,
|
| 535 |
+
"loss": 0.4111,
|
| 536 |
+
"step": 44000
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 0.15,
|
| 540 |
+
"learning_rate": 5.934790418547044e-06,
|
| 541 |
+
"loss": 0.3416,
|
| 542 |
+
"step": 44500
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 0.15,
|
| 546 |
+
"learning_rate": 5.9228153130805785e-06,
|
| 547 |
+
"loss": 0.2562,
|
| 548 |
+
"step": 45000
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 0.16,
|
| 552 |
+
"learning_rate": 5.910840207614114e-06,
|
| 553 |
+
"loss": 0.2773,
|
| 554 |
+
"step": 45500
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 0.16,
|
| 558 |
+
"learning_rate": 5.89886510214765e-06,
|
| 559 |
+
"loss": 0.2311,
|
| 560 |
+
"step": 46000
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 0.16,
|
| 564 |
+
"learning_rate": 5.886889996681185e-06,
|
| 565 |
+
"loss": 0.3196,
|
| 566 |
+
"step": 46500
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 0.16,
|
| 570 |
+
"learning_rate": 5.87491489121472e-06,
|
| 571 |
+
"loss": 0.2729,
|
| 572 |
+
"step": 47000
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 0.16,
|
| 576 |
+
"learning_rate": 5.862939785748255e-06,
|
| 577 |
+
"loss": 0.3177,
|
| 578 |
+
"step": 47500
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 0.16,
|
| 582 |
+
"learning_rate": 5.850964680281791e-06,
|
| 583 |
+
"loss": 0.2905,
|
| 584 |
+
"step": 48000
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 0.17,
|
| 588 |
+
"learning_rate": 5.838989574815327e-06,
|
| 589 |
+
"loss": 0.322,
|
| 590 |
+
"step": 48500
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 0.17,
|
| 594 |
+
"learning_rate": 5.8270144693488625e-06,
|
| 595 |
+
"loss": 0.2838,
|
| 596 |
+
"step": 49000
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 0.17,
|
| 600 |
+
"learning_rate": 5.815039363882397e-06,
|
| 601 |
+
"loss": 0.2668,
|
| 602 |
+
"step": 49500
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 0.17,
|
| 606 |
+
"learning_rate": 5.803064258415933e-06,
|
| 607 |
+
"loss": 0.2295,
|
| 608 |
+
"step": 50000
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 0.17,
|
| 612 |
+
"learning_rate": 5.791137053371334e-06,
|
| 613 |
+
"loss": 0.2489,
|
| 614 |
+
"step": 50500
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 0.17,
|
| 618 |
+
"learning_rate": 5.77916194790487e-06,
|
| 619 |
+
"loss": 0.2887,
|
| 620 |
+
"step": 51000
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 0.18,
|
| 624 |
+
"learning_rate": 5.7671868424384056e-06,
|
| 625 |
+
"loss": 0.2643,
|
| 626 |
+
"step": 51500
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 0.18,
|
| 630 |
+
"learning_rate": 5.7552117369719404e-06,
|
| 631 |
+
"loss": 0.2563,
|
| 632 |
+
"step": 52000
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 0.18,
|
| 636 |
+
"learning_rate": 5.743236631505475e-06,
|
| 637 |
+
"loss": 0.2604,
|
| 638 |
+
"step": 52500
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 0.18,
|
| 642 |
+
"learning_rate": 5.731285476249944e-06,
|
| 643 |
+
"loss": 0.2382,
|
| 644 |
+
"step": 53000
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 0.18,
|
| 648 |
+
"learning_rate": 5.71931037078348e-06,
|
| 649 |
+
"loss": 0.3063,
|
| 650 |
+
"step": 53500
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 0.18,
|
| 654 |
+
"learning_rate": 5.7073352653170155e-06,
|
| 655 |
+
"loss": 0.285,
|
| 656 |
+
"step": 54000
|
| 657 |
+
},
|
| 658 |
+
{
|
| 659 |
+
"epoch": 0.19,
|
| 660 |
+
"learning_rate": 5.69536015985055e-06,
|
| 661 |
+
"loss": 0.3499,
|
| 662 |
+
"step": 54500
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 0.19,
|
| 666 |
+
"learning_rate": 5.683385054384086e-06,
|
| 667 |
+
"loss": 0.3539,
|
| 668 |
+
"step": 55000
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 0.19,
|
| 672 |
+
"learning_rate": 5.671409948917621e-06,
|
| 673 |
+
"loss": 0.2731,
|
| 674 |
+
"step": 55500
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 0.19,
|
| 678 |
+
"learning_rate": 5.659434843451157e-06,
|
| 679 |
+
"loss": 0.3028,
|
| 680 |
+
"step": 56000
|
| 681 |
+
},
|
| 682 |
+
{
|
| 683 |
+
"epoch": 0.19,
|
| 684 |
+
"learning_rate": 5.6474836881956255e-06,
|
| 685 |
+
"loss": 0.3549,
|
| 686 |
+
"step": 56500
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 0.2,
|
| 690 |
+
"learning_rate": 5.635508582729161e-06,
|
| 691 |
+
"loss": 0.333,
|
| 692 |
+
"step": 57000
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 0.2,
|
| 696 |
+
"learning_rate": 5.623533477262696e-06,
|
| 697 |
+
"loss": 0.2349,
|
| 698 |
+
"step": 57500
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 0.2,
|
| 702 |
+
"learning_rate": 5.611558371796231e-06,
|
| 703 |
+
"loss": 0.329,
|
| 704 |
+
"step": 58000
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 0.2,
|
| 708 |
+
"learning_rate": 5.599583266329767e-06,
|
| 709 |
+
"loss": 0.2739,
|
| 710 |
+
"step": 58500
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 0.2,
|
| 714 |
+
"learning_rate": 5.5876321110742354e-06,
|
| 715 |
+
"loss": 0.3069,
|
| 716 |
+
"step": 59000
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 0.2,
|
| 720 |
+
"learning_rate": 5.575657005607771e-06,
|
| 721 |
+
"loss": 0.2472,
|
| 722 |
+
"step": 59500
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 0.21,
|
| 726 |
+
"learning_rate": 5.563681900141306e-06,
|
| 727 |
+
"loss": 0.2929,
|
| 728 |
+
"step": 60000
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 0.21,
|
| 732 |
+
"learning_rate": 5.551706794674841e-06,
|
| 733 |
+
"loss": 0.2743,
|
| 734 |
+
"step": 60500
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 0.21,
|
| 738 |
+
"learning_rate": 5.539731689208377e-06,
|
| 739 |
+
"loss": 0.304,
|
| 740 |
+
"step": 61000
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 0.21,
|
| 744 |
+
"learning_rate": 5.527756583741912e-06,
|
| 745 |
+
"loss": 0.2828,
|
| 746 |
+
"step": 61500
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 0.21,
|
| 750 |
+
"learning_rate": 5.515805428486381e-06,
|
| 751 |
+
"loss": 0.2407,
|
| 752 |
+
"step": 62000
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 0.21,
|
| 756 |
+
"learning_rate": 5.503830323019916e-06,
|
| 757 |
+
"loss": 0.228,
|
| 758 |
+
"step": 62500
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 0.22,
|
| 762 |
+
"learning_rate": 5.491855217553452e-06,
|
| 763 |
+
"loss": 0.2213,
|
| 764 |
+
"step": 63000
|
| 765 |
+
},
|
| 766 |
+
{
|
| 767 |
+
"epoch": 0.22,
|
| 768 |
+
"learning_rate": 5.479880112086987e-06,
|
| 769 |
+
"loss": 0.2989,
|
| 770 |
+
"step": 63500
|
| 771 |
+
},
|
| 772 |
+
{
|
| 773 |
+
"epoch": 0.22,
|
| 774 |
+
"learning_rate": 5.467905006620522e-06,
|
| 775 |
+
"loss": 0.2999,
|
| 776 |
+
"step": 64000
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 0.22,
|
| 780 |
+
"learning_rate": 5.455929901154058e-06,
|
| 781 |
+
"loss": 0.2966,
|
| 782 |
+
"step": 64500
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 0.22,
|
| 786 |
+
"learning_rate": 5.443954795687594e-06,
|
| 787 |
+
"loss": 0.2806,
|
| 788 |
+
"step": 65000
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 0.22,
|
| 792 |
+
"learning_rate": 5.4319796902211294e-06,
|
| 793 |
+
"loss": 0.3088,
|
| 794 |
+
"step": 65500
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 0.23,
|
| 798 |
+
"learning_rate": 5.420004584754664e-06,
|
| 799 |
+
"loss": 0.3368,
|
| 800 |
+
"step": 66000
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 0.23,
|
| 804 |
+
"learning_rate": 5.408029479288199e-06,
|
| 805 |
+
"loss": 0.2686,
|
| 806 |
+
"step": 66500
|
| 807 |
+
},
|
| 808 |
+
{
|
| 809 |
+
"epoch": 0.23,
|
| 810 |
+
"learning_rate": 5.396054373821735e-06,
|
| 811 |
+
"loss": 0.2887,
|
| 812 |
+
"step": 67000
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 0.23,
|
| 816 |
+
"learning_rate": 5.384079268355271e-06,
|
| 817 |
+
"loss": 0.2259,
|
| 818 |
+
"step": 67500
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 0.23,
|
| 822 |
+
"learning_rate": 5.3721041628888055e-06,
|
| 823 |
+
"loss": 0.2527,
|
| 824 |
+
"step": 68000
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 0.23,
|
| 828 |
+
"learning_rate": 5.360129057422341e-06,
|
| 829 |
+
"loss": 0.2642,
|
| 830 |
+
"step": 68500
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 0.24,
|
| 834 |
+
"learning_rate": 5.348153951955877e-06,
|
| 835 |
+
"loss": 0.2365,
|
| 836 |
+
"step": 69000
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 0.24,
|
| 840 |
+
"learning_rate": 5.336178846489413e-06,
|
| 841 |
+
"loss": 0.2573,
|
| 842 |
+
"step": 69500
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 0.24,
|
| 846 |
+
"learning_rate": 5.324227691233881e-06,
|
| 847 |
+
"loss": 0.2804,
|
| 848 |
+
"step": 70000
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 0.24,
|
| 852 |
+
"learning_rate": 5.312252585767416e-06,
|
| 853 |
+
"loss": 0.2953,
|
| 854 |
+
"step": 70500
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 0.24,
|
| 858 |
+
"learning_rate": 5.300277480300951e-06,
|
| 859 |
+
"loss": 0.2623,
|
| 860 |
+
"step": 71000
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 0.24,
|
| 864 |
+
"learning_rate": 5.288302374834487e-06,
|
| 865 |
+
"loss": 0.2324,
|
| 866 |
+
"step": 71500
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 0.25,
|
| 870 |
+
"learning_rate": 5.276327269368023e-06,
|
| 871 |
+
"loss": 0.305,
|
| 872 |
+
"step": 72000
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 0.25,
|
| 876 |
+
"learning_rate": 5.264352163901558e-06,
|
| 877 |
+
"loss": 0.277,
|
| 878 |
+
"step": 72500
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 0.25,
|
| 882 |
+
"learning_rate": 5.252401008646026e-06,
|
| 883 |
+
"loss": 0.2758,
|
| 884 |
+
"step": 73000
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 0.25,
|
| 888 |
+
"learning_rate": 5.240425903179561e-06,
|
| 889 |
+
"loss": 0.2505,
|
| 890 |
+
"step": 73500
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 0.25,
|
| 894 |
+
"learning_rate": 5.228450797713097e-06,
|
| 895 |
+
"loss": 0.2518,
|
| 896 |
+
"step": 74000
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 0.25,
|
| 900 |
+
"learning_rate": 5.2164756922466326e-06,
|
| 901 |
+
"loss": 0.2642,
|
| 902 |
+
"step": 74500
|
| 903 |
+
},
|
| 904 |
+
{
|
| 905 |
+
"epoch": 0.26,
|
| 906 |
+
"learning_rate": 5.204500586780168e-06,
|
| 907 |
+
"loss": 0.33,
|
| 908 |
+
"step": 75000
|
| 909 |
+
},
|
| 910 |
+
{
|
| 911 |
+
"epoch": 0.26,
|
| 912 |
+
"learning_rate": 5.192525481313703e-06,
|
| 913 |
+
"loss": 0.2136,
|
| 914 |
+
"step": 75500
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"epoch": 0.26,
|
| 918 |
+
"learning_rate": 5.180550375847238e-06,
|
| 919 |
+
"loss": 0.2679,
|
| 920 |
+
"step": 76000
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 0.26,
|
| 924 |
+
"learning_rate": 5.168575270380774e-06,
|
| 925 |
+
"loss": 0.2467,
|
| 926 |
+
"step": 76500
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 0.26,
|
| 930 |
+
"learning_rate": 5.1566241151252425e-06,
|
| 931 |
+
"loss": 0.2634,
|
| 932 |
+
"step": 77000
|
| 933 |
+
},
|
| 934 |
+
{
|
| 935 |
+
"epoch": 0.27,
|
| 936 |
+
"learning_rate": 5.144649009658778e-06,
|
| 937 |
+
"loss": 0.2534,
|
| 938 |
+
"step": 77500
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 0.27,
|
| 942 |
+
"learning_rate": 5.132673904192314e-06,
|
| 943 |
+
"loss": 0.2633,
|
| 944 |
+
"step": 78000
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 0.27,
|
| 948 |
+
"learning_rate": 5.120698798725848e-06,
|
| 949 |
+
"loss": 0.2758,
|
| 950 |
+
"step": 78500
|
| 951 |
+
},
|
| 952 |
+
{
|
| 953 |
+
"epoch": 0.27,
|
| 954 |
+
"learning_rate": 5.108723693259384e-06,
|
| 955 |
+
"loss": 0.2389,
|
| 956 |
+
"step": 79000
|
| 957 |
+
},
|
| 958 |
+
{
|
| 959 |
+
"epoch": 0.27,
|
| 960 |
+
"learning_rate": 5.0967725380038525e-06,
|
| 961 |
+
"loss": 0.2774,
|
| 962 |
+
"step": 79500
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 0.27,
|
| 966 |
+
"learning_rate": 5.084797432537388e-06,
|
| 967 |
+
"loss": 0.2451,
|
| 968 |
+
"step": 80000
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 0.28,
|
| 972 |
+
"learning_rate": 5.072822327070924e-06,
|
| 973 |
+
"loss": 0.2227,
|
| 974 |
+
"step": 80500
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 0.28,
|
| 978 |
+
"learning_rate": 5.060847221604459e-06,
|
| 979 |
+
"loss": 0.2652,
|
| 980 |
+
"step": 81000
|
| 981 |
+
},
|
| 982 |
+
{
|
| 983 |
+
"epoch": 0.28,
|
| 984 |
+
"learning_rate": 5.048872116137994e-06,
|
| 985 |
+
"loss": 0.2727,
|
| 986 |
+
"step": 81500
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 0.28,
|
| 990 |
+
"learning_rate": 5.036897010671529e-06,
|
| 991 |
+
"loss": 0.257,
|
| 992 |
+
"step": 82000
|
| 993 |
+
},
|
| 994 |
+
{
|
| 995 |
+
"epoch": 0.28,
|
| 996 |
+
"learning_rate": 5.024921905205065e-06,
|
| 997 |
+
"loss": 0.209,
|
| 998 |
+
"step": 82500
|
| 999 |
+
},
|
| 1000 |
+
{
|
| 1001 |
+
"epoch": 0.28,
|
| 1002 |
+
"learning_rate": 5.012946799738601e-06,
|
| 1003 |
+
"loss": 0.2342,
|
| 1004 |
+
"step": 83000
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 0.29,
|
| 1008 |
+
"learning_rate": 5.000971694272136e-06,
|
| 1009 |
+
"loss": 0.2429,
|
| 1010 |
+
"step": 83500
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 0.29,
|
| 1014 |
+
"learning_rate": 4.989020539016604e-06,
|
| 1015 |
+
"loss": 0.2346,
|
| 1016 |
+
"step": 84000
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"epoch": 0.29,
|
| 1020 |
+
"learning_rate": 4.977045433550139e-06,
|
| 1021 |
+
"loss": 0.2521,
|
| 1022 |
+
"step": 84500
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 0.29,
|
| 1026 |
+
"learning_rate": 4.965094278294608e-06,
|
| 1027 |
+
"loss": 0.2536,
|
| 1028 |
+
"step": 85000
|
| 1029 |
+
},
|
| 1030 |
+
{
|
| 1031 |
+
"epoch": 0.29,
|
| 1032 |
+
"learning_rate": 4.953119172828144e-06,
|
| 1033 |
+
"loss": 0.2448,
|
| 1034 |
+
"step": 85500
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 0.29,
|
| 1038 |
+
"learning_rate": 4.9411440673616796e-06,
|
| 1039 |
+
"loss": 0.2265,
|
| 1040 |
+
"step": 86000
|
| 1041 |
+
},
|
| 1042 |
+
{
|
| 1043 |
+
"epoch": 0.3,
|
| 1044 |
+
"learning_rate": 4.929168961895214e-06,
|
| 1045 |
+
"loss": 0.2427,
|
| 1046 |
+
"step": 86500
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 0.3,
|
| 1050 |
+
"learning_rate": 4.917193856428749e-06,
|
| 1051 |
+
"loss": 0.2626,
|
| 1052 |
+
"step": 87000
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 0.3,
|
| 1056 |
+
"learning_rate": 4.905218750962285e-06,
|
| 1057 |
+
"loss": 0.2531,
|
| 1058 |
+
"step": 87500
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 0.3,
|
| 1062 |
+
"learning_rate": 4.893267595706754e-06,
|
| 1063 |
+
"loss": 0.2344,
|
| 1064 |
+
"step": 88000
|
| 1065 |
+
},
|
| 1066 |
+
{
|
| 1067 |
+
"epoch": 0.3,
|
| 1068 |
+
"learning_rate": 4.8812924902402895e-06,
|
| 1069 |
+
"loss": 0.2213,
|
| 1070 |
+
"step": 88500
|
| 1071 |
+
},
|
| 1072 |
+
{
|
| 1073 |
+
"epoch": 0.3,
|
| 1074 |
+
"learning_rate": 4.869317384773824e-06,
|
| 1075 |
+
"loss": 0.2384,
|
| 1076 |
+
"step": 89000
|
| 1077 |
+
},
|
| 1078 |
+
{
|
| 1079 |
+
"epoch": 0.31,
|
| 1080 |
+
"learning_rate": 4.857342279307359e-06,
|
| 1081 |
+
"loss": 0.2676,
|
| 1082 |
+
"step": 89500
|
| 1083 |
+
},
|
| 1084 |
+
{
|
| 1085 |
+
"epoch": 0.31,
|
| 1086 |
+
"learning_rate": 4.845367173840895e-06,
|
| 1087 |
+
"loss": 0.271,
|
| 1088 |
+
"step": 90000
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 0.31,
|
| 1092 |
+
"learning_rate": 4.833416018585364e-06,
|
| 1093 |
+
"loss": 0.2691,
|
| 1094 |
+
"step": 90500
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 0.31,
|
| 1098 |
+
"learning_rate": 4.8214409131188995e-06,
|
| 1099 |
+
"loss": 0.2609,
|
| 1100 |
+
"step": 91000
|
| 1101 |
+
},
|
| 1102 |
+
{
|
| 1103 |
+
"epoch": 0.31,
|
| 1104 |
+
"learning_rate": 4.809465807652434e-06,
|
| 1105 |
+
"loss": 0.2414,
|
| 1106 |
+
"step": 91500
|
| 1107 |
+
},
|
| 1108 |
+
{
|
| 1109 |
+
"epoch": 0.31,
|
| 1110 |
+
"learning_rate": 4.797490702185969e-06,
|
| 1111 |
+
"loss": 0.2542,
|
| 1112 |
+
"step": 92000
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 0.32,
|
| 1116 |
+
"learning_rate": 4.785539546930438e-06,
|
| 1117 |
+
"loss": 0.2575,
|
| 1118 |
+
"step": 92500
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 0.32,
|
| 1122 |
+
"learning_rate": 4.773564441463974e-06,
|
| 1123 |
+
"loss": 0.2182,
|
| 1124 |
+
"step": 93000
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 0.32,
|
| 1128 |
+
"learning_rate": 4.7615893359975094e-06,
|
| 1129 |
+
"loss": 0.2975,
|
| 1130 |
+
"step": 93500
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 0.32,
|
| 1134 |
+
"learning_rate": 4.749614230531045e-06,
|
| 1135 |
+
"loss": 0.2542,
|
| 1136 |
+
"step": 94000
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 0.32,
|
| 1140 |
+
"learning_rate": 4.73763912506458e-06,
|
| 1141 |
+
"loss": 0.2297,
|
| 1142 |
+
"step": 94500
|
| 1143 |
+
},
|
| 1144 |
+
{
|
| 1145 |
+
"epoch": 0.33,
|
| 1146 |
+
"learning_rate": 4.725664019598115e-06,
|
| 1147 |
+
"loss": 0.2704,
|
| 1148 |
+
"step": 95000
|
| 1149 |
+
},
|
| 1150 |
+
{
|
| 1151 |
+
"epoch": 0.33,
|
| 1152 |
+
"learning_rate": 4.713712864342584e-06,
|
| 1153 |
+
"loss": 0.2101,
|
| 1154 |
+
"step": 95500
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 0.33,
|
| 1158 |
+
"learning_rate": 4.701737758876119e-06,
|
| 1159 |
+
"loss": 0.2597,
|
| 1160 |
+
"step": 96000
|
| 1161 |
+
},
|
| 1162 |
+
{
|
| 1163 |
+
"epoch": 0.33,
|
| 1164 |
+
"learning_rate": 4.689762653409655e-06,
|
| 1165 |
+
"loss": 0.2194,
|
| 1166 |
+
"step": 96500
|
| 1167 |
+
},
|
| 1168 |
+
{
|
| 1169 |
+
"epoch": 0.33,
|
| 1170 |
+
"learning_rate": 4.67778754794319e-06,
|
| 1171 |
+
"loss": 0.2235,
|
| 1172 |
+
"step": 97000
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 0.33,
|
| 1176 |
+
"learning_rate": 4.665860342898592e-06,
|
| 1177 |
+
"loss": 0.2419,
|
| 1178 |
+
"step": 97500
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 0.34,
|
| 1182 |
+
"learning_rate": 4.653885237432127e-06,
|
| 1183 |
+
"loss": 0.2434,
|
| 1184 |
+
"step": 98000
|
| 1185 |
+
},
|
| 1186 |
+
{
|
| 1187 |
+
"epoch": 0.34,
|
| 1188 |
+
"learning_rate": 4.641910131965662e-06,
|
| 1189 |
+
"loss": 0.2861,
|
| 1190 |
+
"step": 98500
|
| 1191 |
+
},
|
| 1192 |
+
{
|
| 1193 |
+
"epoch": 0.34,
|
| 1194 |
+
"learning_rate": 4.629935026499197e-06,
|
| 1195 |
+
"loss": 0.2311,
|
| 1196 |
+
"step": 99000
|
| 1197 |
+
},
|
| 1198 |
+
{
|
| 1199 |
+
"epoch": 0.34,
|
| 1200 |
+
"learning_rate": 4.617959921032733e-06,
|
| 1201 |
+
"loss": 0.2198,
|
| 1202 |
+
"step": 99500
|
| 1203 |
+
},
|
| 1204 |
+
{
|
| 1205 |
+
"epoch": 0.34,
|
| 1206 |
+
"learning_rate": 4.605984815566269e-06,
|
| 1207 |
+
"loss": 0.243,
|
| 1208 |
+
"step": 100000
|
| 1209 |
+
},
|
| 1210 |
+
{
|
| 1211 |
+
"epoch": 0.34,
|
| 1212 |
+
"learning_rate": 4.5940336603107375e-06,
|
| 1213 |
+
"loss": 0.2348,
|
| 1214 |
+
"step": 100500
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 0.35,
|
| 1218 |
+
"learning_rate": 4.582058554844272e-06,
|
| 1219 |
+
"loss": 0.25,
|
| 1220 |
+
"step": 101000
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 0.35,
|
| 1224 |
+
"learning_rate": 4.570083449377807e-06,
|
| 1225 |
+
"loss": 0.2267,
|
| 1226 |
+
"step": 101500
|
| 1227 |
+
},
|
| 1228 |
+
{
|
| 1229 |
+
"epoch": 0.35,
|
| 1230 |
+
"learning_rate": 4.558108343911343e-06,
|
| 1231 |
+
"loss": 0.2492,
|
| 1232 |
+
"step": 102000
|
| 1233 |
+
},
|
| 1234 |
+
{
|
| 1235 |
+
"epoch": 0.35,
|
| 1236 |
+
"learning_rate": 4.546133238444879e-06,
|
| 1237 |
+
"loss": 0.2534,
|
| 1238 |
+
"step": 102500
|
| 1239 |
+
},
|
| 1240 |
+
{
|
| 1241 |
+
"epoch": 0.35,
|
| 1242 |
+
"learning_rate": 4.5341581329784136e-06,
|
| 1243 |
+
"loss": 0.2082,
|
| 1244 |
+
"step": 103000
|
| 1245 |
+
},
|
| 1246 |
+
{
|
| 1247 |
+
"epoch": 0.35,
|
| 1248 |
+
"learning_rate": 4.522206977722882e-06,
|
| 1249 |
+
"loss": 0.263,
|
| 1250 |
+
"step": 103500
|
| 1251 |
+
},
|
| 1252 |
+
{
|
| 1253 |
+
"epoch": 0.36,
|
| 1254 |
+
"learning_rate": 4.510231872256417e-06,
|
| 1255 |
+
"loss": 0.2707,
|
| 1256 |
+
"step": 104000
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 0.36,
|
| 1260 |
+
"learning_rate": 4.498256766789953e-06,
|
| 1261 |
+
"loss": 0.2593,
|
| 1262 |
+
"step": 104500
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 0.36,
|
| 1266 |
+
"learning_rate": 4.486281661323489e-06,
|
| 1267 |
+
"loss": 0.2735,
|
| 1268 |
+
"step": 105000
|
| 1269 |
+
},
|
| 1270 |
+
{
|
| 1271 |
+
"epoch": 0.36,
|
| 1272 |
+
"learning_rate": 4.474306555857024e-06,
|
| 1273 |
+
"loss": 0.2743,
|
| 1274 |
+
"step": 105500
|
| 1275 |
+
},
|
| 1276 |
+
{
|
| 1277 |
+
"epoch": 0.36,
|
| 1278 |
+
"learning_rate": 4.462331450390559e-06,
|
| 1279 |
+
"loss": 0.2314,
|
| 1280 |
+
"step": 106000
|
| 1281 |
+
},
|
| 1282 |
+
{
|
| 1283 |
+
"epoch": 0.36,
|
| 1284 |
+
"learning_rate": 4.450356344924095e-06,
|
| 1285 |
+
"loss": 0.2241,
|
| 1286 |
+
"step": 106500
|
| 1287 |
+
},
|
| 1288 |
+
{
|
| 1289 |
+
"epoch": 0.37,
|
| 1290 |
+
"learning_rate": 4.438381239457631e-06,
|
| 1291 |
+
"loss": 0.2471,
|
| 1292 |
+
"step": 107000
|
| 1293 |
+
},
|
| 1294 |
+
{
|
| 1295 |
+
"epoch": 0.37,
|
| 1296 |
+
"learning_rate": 4.426430084202099e-06,
|
| 1297 |
+
"loss": 0.2347,
|
| 1298 |
+
"step": 107500
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 0.37,
|
| 1302 |
+
"learning_rate": 4.414454978735634e-06,
|
| 1303 |
+
"loss": 0.2391,
|
| 1304 |
+
"step": 108000
|
| 1305 |
+
},
|
| 1306 |
+
{
|
| 1307 |
+
"epoch": 0.37,
|
| 1308 |
+
"learning_rate": 4.402479873269169e-06,
|
| 1309 |
+
"loss": 0.2819,
|
| 1310 |
+
"step": 108500
|
| 1311 |
+
},
|
| 1312 |
+
{
|
| 1313 |
+
"epoch": 0.37,
|
| 1314 |
+
"learning_rate": 4.390504767802705e-06,
|
| 1315 |
+
"loss": 0.2278,
|
| 1316 |
+
"step": 109000
|
| 1317 |
+
},
|
| 1318 |
+
{
|
| 1319 |
+
"epoch": 0.37,
|
| 1320 |
+
"learning_rate": 4.378529662336241e-06,
|
| 1321 |
+
"loss": 0.2467,
|
| 1322 |
+
"step": 109500
|
| 1323 |
+
},
|
| 1324 |
+
{
|
| 1325 |
+
"epoch": 0.38,
|
| 1326 |
+
"learning_rate": 4.366554556869776e-06,
|
| 1327 |
+
"loss": 0.2494,
|
| 1328 |
+
"step": 110000
|
| 1329 |
+
},
|
| 1330 |
+
{
|
| 1331 |
+
"epoch": 0.38,
|
| 1332 |
+
"learning_rate": 4.354579451403311e-06,
|
| 1333 |
+
"loss": 0.2059,
|
| 1334 |
+
"step": 110500
|
| 1335 |
+
},
|
| 1336 |
+
{
|
| 1337 |
+
"epoch": 0.38,
|
| 1338 |
+
"learning_rate": 4.342604345936846e-06,
|
| 1339 |
+
"loss": 0.2388,
|
| 1340 |
+
"step": 111000
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 0.38,
|
| 1344 |
+
"learning_rate": 4.330653190681315e-06,
|
| 1345 |
+
"loss": 0.2286,
|
| 1346 |
+
"step": 111500
|
| 1347 |
+
},
|
| 1348 |
+
{
|
| 1349 |
+
"epoch": 0.38,
|
| 1350 |
+
"learning_rate": 4.318678085214851e-06,
|
| 1351 |
+
"loss": 0.2519,
|
| 1352 |
+
"step": 112000
|
| 1353 |
+
},
|
| 1354 |
+
{
|
| 1355 |
+
"epoch": 0.38,
|
| 1356 |
+
"learning_rate": 4.3067269299593185e-06,
|
| 1357 |
+
"loss": 0.2444,
|
| 1358 |
+
"step": 112500
|
| 1359 |
+
},
|
| 1360 |
+
{
|
| 1361 |
+
"epoch": 0.39,
|
| 1362 |
+
"learning_rate": 4.294751824492854e-06,
|
| 1363 |
+
"loss": 0.2278,
|
| 1364 |
+
"step": 113000
|
| 1365 |
+
},
|
| 1366 |
+
{
|
| 1367 |
+
"epoch": 0.39,
|
| 1368 |
+
"learning_rate": 4.28277671902639e-06,
|
| 1369 |
+
"loss": 0.2176,
|
| 1370 |
+
"step": 113500
|
| 1371 |
+
},
|
| 1372 |
+
{
|
| 1373 |
+
"epoch": 0.39,
|
| 1374 |
+
"learning_rate": 4.270801613559925e-06,
|
| 1375 |
+
"loss": 0.2382,
|
| 1376 |
+
"step": 114000
|
| 1377 |
+
},
|
| 1378 |
+
{
|
| 1379 |
+
"epoch": 0.39,
|
| 1380 |
+
"learning_rate": 4.258850458304393e-06,
|
| 1381 |
+
"loss": 0.2402,
|
| 1382 |
+
"step": 114500
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 0.39,
|
| 1386 |
+
"learning_rate": 4.2468753528379285e-06,
|
| 1387 |
+
"loss": 0.2632,
|
| 1388 |
+
"step": 115000
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 0.4,
|
| 1392 |
+
"learning_rate": 4.234924197582397e-06,
|
| 1393 |
+
"loss": 0.2324,
|
| 1394 |
+
"step": 115500
|
| 1395 |
+
},
|
| 1396 |
+
{
|
| 1397 |
+
"epoch": 0.4,
|
| 1398 |
+
"learning_rate": 4.222949092115933e-06,
|
| 1399 |
+
"loss": 0.2159,
|
| 1400 |
+
"step": 116000
|
| 1401 |
+
},
|
| 1402 |
+
{
|
| 1403 |
+
"epoch": 0.4,
|
| 1404 |
+
"learning_rate": 4.210973986649469e-06,
|
| 1405 |
+
"loss": 0.2862,
|
| 1406 |
+
"step": 116500
|
| 1407 |
+
},
|
| 1408 |
+
{
|
| 1409 |
+
"epoch": 0.4,
|
| 1410 |
+
"learning_rate": 4.198998881183003e-06,
|
| 1411 |
+
"loss": 0.2385,
|
| 1412 |
+
"step": 117000
|
| 1413 |
+
},
|
| 1414 |
+
{
|
| 1415 |
+
"epoch": 0.4,
|
| 1416 |
+
"learning_rate": 4.1870477259274715e-06,
|
| 1417 |
+
"loss": 0.2146,
|
| 1418 |
+
"step": 117500
|
| 1419 |
+
},
|
| 1420 |
+
{
|
| 1421 |
+
"epoch": 0.4,
|
| 1422 |
+
"learning_rate": 4.175072620461007e-06,
|
| 1423 |
+
"loss": 0.2749,
|
| 1424 |
+
"step": 118000
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 0.41,
|
| 1428 |
+
"learning_rate": 4.163097514994543e-06,
|
| 1429 |
+
"loss": 0.2164,
|
| 1430 |
+
"step": 118500
|
| 1431 |
+
},
|
| 1432 |
+
{
|
| 1433 |
+
"epoch": 0.41,
|
| 1434 |
+
"learning_rate": 4.151122409528079e-06,
|
| 1435 |
+
"loss": 0.2521,
|
| 1436 |
+
"step": 119000
|
| 1437 |
+
},
|
| 1438 |
+
{
|
| 1439 |
+
"epoch": 0.41,
|
| 1440 |
+
"learning_rate": 4.1391473040616136e-06,
|
| 1441 |
+
"loss": 0.1924,
|
| 1442 |
+
"step": 119500
|
| 1443 |
+
},
|
| 1444 |
+
{
|
| 1445 |
+
"epoch": 0.41,
|
| 1446 |
+
"learning_rate": 4.1271721985951484e-06,
|
| 1447 |
+
"loss": 0.2564,
|
| 1448 |
+
"step": 120000
|
| 1449 |
+
},
|
| 1450 |
+
{
|
| 1451 |
+
"epoch": 0.41,
|
| 1452 |
+
"learning_rate": 4.115197093128684e-06,
|
| 1453 |
+
"loss": 0.1907,
|
| 1454 |
+
"step": 120500
|
| 1455 |
+
},
|
| 1456 |
+
{
|
| 1457 |
+
"epoch": 0.41,
|
| 1458 |
+
"learning_rate": 4.103245937873153e-06,
|
| 1459 |
+
"loss": 0.2629,
|
| 1460 |
+
"step": 121000
|
| 1461 |
+
},
|
| 1462 |
+
{
|
| 1463 |
+
"epoch": 0.42,
|
| 1464 |
+
"learning_rate": 4.091270832406689e-06,
|
| 1465 |
+
"loss": 0.1845,
|
| 1466 |
+
"step": 121500
|
| 1467 |
+
},
|
| 1468 |
+
{
|
| 1469 |
+
"epoch": 0.42,
|
| 1470 |
+
"learning_rate": 4.079295726940224e-06,
|
| 1471 |
+
"loss": 0.1999,
|
| 1472 |
+
"step": 122000
|
| 1473 |
+
},
|
| 1474 |
+
{
|
| 1475 |
+
"epoch": 0.42,
|
| 1476 |
+
"learning_rate": 4.067320621473758e-06,
|
| 1477 |
+
"loss": 0.2527,
|
| 1478 |
+
"step": 122500
|
| 1479 |
+
},
|
| 1480 |
+
{
|
| 1481 |
+
"epoch": 0.42,
|
| 1482 |
+
"learning_rate": 4.055345516007294e-06,
|
| 1483 |
+
"loss": 0.2153,
|
| 1484 |
+
"step": 123000
|
| 1485 |
+
},
|
| 1486 |
+
{
|
| 1487 |
+
"epoch": 0.42,
|
| 1488 |
+
"learning_rate": 4.04337041054083e-06,
|
| 1489 |
+
"loss": 0.2089,
|
| 1490 |
+
"step": 123500
|
| 1491 |
+
},
|
| 1492 |
+
{
|
| 1493 |
+
"epoch": 0.42,
|
| 1494 |
+
"learning_rate": 4.0313953050743655e-06,
|
| 1495 |
+
"loss": 0.2335,
|
| 1496 |
+
"step": 124000
|
| 1497 |
+
},
|
| 1498 |
+
{
|
| 1499 |
+
"epoch": 0.43,
|
| 1500 |
+
"learning_rate": 4.019420199607901e-06,
|
| 1501 |
+
"loss": 0.2252,
|
| 1502 |
+
"step": 124500
|
| 1503 |
+
},
|
| 1504 |
+
{
|
| 1505 |
+
"epoch": 0.43,
|
| 1506 |
+
"learning_rate": 4.007445094141436e-06,
|
| 1507 |
+
"loss": 0.2381,
|
| 1508 |
+
"step": 125000
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"epoch": 0.43,
|
| 1512 |
+
"learning_rate": 3.995469988674972e-06,
|
| 1513 |
+
"loss": 0.2327,
|
| 1514 |
+
"step": 125500
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 0.43,
|
| 1518 |
+
"learning_rate": 3.983494883208507e-06,
|
| 1519 |
+
"loss": 0.247,
|
| 1520 |
+
"step": 126000
|
| 1521 |
+
},
|
| 1522 |
+
{
|
| 1523 |
+
"epoch": 0.43,
|
| 1524 |
+
"learning_rate": 3.9715197777420424e-06,
|
| 1525 |
+
"loss": 0.2659,
|
| 1526 |
+
"step": 126500
|
| 1527 |
+
},
|
| 1528 |
+
{
|
| 1529 |
+
"epoch": 0.43,
|
| 1530 |
+
"learning_rate": 3.959544672275577e-06,
|
| 1531 |
+
"loss": 0.2212,
|
| 1532 |
+
"step": 127000
|
| 1533 |
+
},
|
| 1534 |
+
{
|
| 1535 |
+
"epoch": 0.44,
|
| 1536 |
+
"learning_rate": 3.947569566809113e-06,
|
| 1537 |
+
"loss": 0.2265,
|
| 1538 |
+
"step": 127500
|
| 1539 |
+
},
|
| 1540 |
+
{
|
| 1541 |
+
"epoch": 0.44,
|
| 1542 |
+
"learning_rate": 3.935618411553582e-06,
|
| 1543 |
+
"loss": 0.2389,
|
| 1544 |
+
"step": 128000
|
| 1545 |
+
},
|
| 1546 |
+
{
|
| 1547 |
+
"epoch": 0.44,
|
| 1548 |
+
"learning_rate": 3.9236433060871175e-06,
|
| 1549 |
+
"loss": 0.2342,
|
| 1550 |
+
"step": 128500
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 0.44,
|
| 1554 |
+
"learning_rate": 3.911668200620652e-06,
|
| 1555 |
+
"loss": 0.2181,
|
| 1556 |
+
"step": 129000
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 0.44,
|
| 1560 |
+
"learning_rate": 3.899693095154188e-06,
|
| 1561 |
+
"loss": 0.2378,
|
| 1562 |
+
"step": 129500
|
| 1563 |
+
},
|
| 1564 |
+
{
|
| 1565 |
+
"epoch": 0.44,
|
| 1566 |
+
"learning_rate": 3.887717989687723e-06,
|
| 1567 |
+
"loss": 0.2397,
|
| 1568 |
+
"step": 130000
|
| 1569 |
+
},
|
| 1570 |
+
{
|
| 1571 |
+
"epoch": 0.45,
|
| 1572 |
+
"learning_rate": 3.875742884221259e-06,
|
| 1573 |
+
"loss": 0.2606,
|
| 1574 |
+
"step": 130500
|
| 1575 |
+
},
|
| 1576 |
+
{
|
| 1577 |
+
"epoch": 0.45,
|
| 1578 |
+
"learning_rate": 3.863767778754794e-06,
|
| 1579 |
+
"loss": 0.2198,
|
| 1580 |
+
"step": 131000
|
| 1581 |
+
},
|
| 1582 |
+
{
|
| 1583 |
+
"epoch": 0.45,
|
| 1584 |
+
"learning_rate": 3.851816623499262e-06,
|
| 1585 |
+
"loss": 0.2213,
|
| 1586 |
+
"step": 131500
|
| 1587 |
+
},
|
| 1588 |
+
{
|
| 1589 |
+
"epoch": 0.45,
|
| 1590 |
+
"learning_rate": 3.839841518032798e-06,
|
| 1591 |
+
"loss": 0.2663,
|
| 1592 |
+
"step": 132000
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 0.45,
|
| 1596 |
+
"learning_rate": 3.827866412566333e-06,
|
| 1597 |
+
"loss": 0.2195,
|
| 1598 |
+
"step": 132500
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 0.46,
|
| 1602 |
+
"learning_rate": 3.815891307099869e-06,
|
| 1603 |
+
"loss": 0.1806,
|
| 1604 |
+
"step": 133000
|
| 1605 |
+
},
|
| 1606 |
+
{
|
| 1607 |
+
"epoch": 0.46,
|
| 1608 |
+
"learning_rate": 3.803940151844337e-06,
|
| 1609 |
+
"loss": 0.2167,
|
| 1610 |
+
"step": 133500
|
| 1611 |
+
},
|
| 1612 |
+
{
|
| 1613 |
+
"epoch": 0.46,
|
| 1614 |
+
"learning_rate": 3.7919650463778727e-06,
|
| 1615 |
+
"loss": 0.2563,
|
| 1616 |
+
"step": 134000
|
| 1617 |
+
},
|
| 1618 |
+
{
|
| 1619 |
+
"epoch": 0.46,
|
| 1620 |
+
"learning_rate": 3.7799899409114085e-06,
|
| 1621 |
+
"loss": 0.2136,
|
| 1622 |
+
"step": 134500
|
| 1623 |
+
},
|
| 1624 |
+
{
|
| 1625 |
+
"epoch": 0.46,
|
| 1626 |
+
"learning_rate": 3.7680148354449438e-06,
|
| 1627 |
+
"loss": 0.2307,
|
| 1628 |
+
"step": 135000
|
| 1629 |
+
},
|
| 1630 |
+
{
|
| 1631 |
+
"epoch": 0.46,
|
| 1632 |
+
"learning_rate": 3.7560397299784786e-06,
|
| 1633 |
+
"loss": 0.2135,
|
| 1634 |
+
"step": 135500
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 0.47,
|
| 1638 |
+
"learning_rate": 3.744088574722947e-06,
|
| 1639 |
+
"loss": 0.253,
|
| 1640 |
+
"step": 136000
|
| 1641 |
+
},
|
| 1642 |
+
{
|
| 1643 |
+
"epoch": 0.47,
|
| 1644 |
+
"learning_rate": 3.7321134692564827e-06,
|
| 1645 |
+
"loss": 0.2402,
|
| 1646 |
+
"step": 136500
|
| 1647 |
+
},
|
| 1648 |
+
{
|
| 1649 |
+
"epoch": 0.47,
|
| 1650 |
+
"learning_rate": 3.7201383637900184e-06,
|
| 1651 |
+
"loss": 0.2199,
|
| 1652 |
+
"step": 137000
|
| 1653 |
+
},
|
| 1654 |
+
{
|
| 1655 |
+
"epoch": 0.47,
|
| 1656 |
+
"learning_rate": 3.7081632583235537e-06,
|
| 1657 |
+
"loss": 0.2072,
|
| 1658 |
+
"step": 137500
|
| 1659 |
+
},
|
| 1660 |
+
{
|
| 1661 |
+
"epoch": 0.47,
|
| 1662 |
+
"learning_rate": 3.6961881528570886e-06,
|
| 1663 |
+
"loss": 0.2046,
|
| 1664 |
+
"step": 138000
|
| 1665 |
+
},
|
| 1666 |
+
{
|
| 1667 |
+
"epoch": 0.47,
|
| 1668 |
+
"learning_rate": 3.684236997601557e-06,
|
| 1669 |
+
"loss": 0.1959,
|
| 1670 |
+
"step": 138500
|
| 1671 |
+
},
|
| 1672 |
+
{
|
| 1673 |
+
"epoch": 0.48,
|
| 1674 |
+
"learning_rate": 3.6722618921350927e-06,
|
| 1675 |
+
"loss": 0.2218,
|
| 1676 |
+
"step": 139000
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"epoch": 0.48,
|
| 1680 |
+
"learning_rate": 3.6602867866686284e-06,
|
| 1681 |
+
"loss": 0.2931,
|
| 1682 |
+
"step": 139500
|
| 1683 |
+
},
|
| 1684 |
+
{
|
| 1685 |
+
"epoch": 0.48,
|
| 1686 |
+
"learning_rate": 3.648311681202164e-06,
|
| 1687 |
+
"loss": 0.2156,
|
| 1688 |
+
"step": 140000
|
| 1689 |
+
},
|
| 1690 |
+
{
|
| 1691 |
+
"epoch": 0.48,
|
| 1692 |
+
"learning_rate": 3.6363605259466325e-06,
|
| 1693 |
+
"loss": 0.2149,
|
| 1694 |
+
"step": 140500
|
| 1695 |
+
},
|
| 1696 |
+
{
|
| 1697 |
+
"epoch": 0.48,
|
| 1698 |
+
"learning_rate": 3.6243854204801673e-06,
|
| 1699 |
+
"loss": 0.234,
|
| 1700 |
+
"step": 141000
|
| 1701 |
+
},
|
| 1702 |
+
{
|
| 1703 |
+
"epoch": 0.48,
|
| 1704 |
+
"learning_rate": 3.6124103150137026e-06,
|
| 1705 |
+
"loss": 0.2632,
|
| 1706 |
+
"step": 141500
|
| 1707 |
+
},
|
| 1708 |
+
{
|
| 1709 |
+
"epoch": 0.49,
|
| 1710 |
+
"learning_rate": 3.6004352095472383e-06,
|
| 1711 |
+
"loss": 0.2561,
|
| 1712 |
+
"step": 142000
|
| 1713 |
+
},
|
| 1714 |
+
{
|
| 1715 |
+
"epoch": 0.49,
|
| 1716 |
+
"learning_rate": 3.5884840542917067e-06,
|
| 1717 |
+
"loss": 0.2184,
|
| 1718 |
+
"step": 142500
|
| 1719 |
+
},
|
| 1720 |
+
{
|
| 1721 |
+
"epoch": 0.49,
|
| 1722 |
+
"learning_rate": 3.5765089488252424e-06,
|
| 1723 |
+
"loss": 0.2336,
|
| 1724 |
+
"step": 143000
|
| 1725 |
+
},
|
| 1726 |
+
{
|
| 1727 |
+
"epoch": 0.49,
|
| 1728 |
+
"learning_rate": 3.5645338433587773e-06,
|
| 1729 |
+
"loss": 0.2321,
|
| 1730 |
+
"step": 143500
|
| 1731 |
+
},
|
| 1732 |
+
{
|
| 1733 |
+
"epoch": 0.49,
|
| 1734 |
+
"learning_rate": 3.5525587378923126e-06,
|
| 1735 |
+
"loss": 0.2539,
|
| 1736 |
+
"step": 144000
|
| 1737 |
+
},
|
| 1738 |
+
{
|
| 1739 |
+
"epoch": 0.49,
|
| 1740 |
+
"learning_rate": 3.540607582636781e-06,
|
| 1741 |
+
"loss": 0.2314,
|
| 1742 |
+
"step": 144500
|
| 1743 |
+
},
|
| 1744 |
+
{
|
| 1745 |
+
"epoch": 0.5,
|
| 1746 |
+
"learning_rate": 3.5286324771703167e-06,
|
| 1747 |
+
"loss": 0.2352,
|
| 1748 |
+
"step": 145000
|
| 1749 |
+
},
|
| 1750 |
+
{
|
| 1751 |
+
"epoch": 0.5,
|
| 1752 |
+
"learning_rate": 3.5166573717038524e-06,
|
| 1753 |
+
"loss": 0.2136,
|
| 1754 |
+
"step": 145500
|
| 1755 |
+
},
|
| 1756 |
+
{
|
| 1757 |
+
"epoch": 0.5,
|
| 1758 |
+
"learning_rate": 3.504682266237388e-06,
|
| 1759 |
+
"loss": 0.2662,
|
| 1760 |
+
"step": 146000
|
| 1761 |
+
},
|
| 1762 |
+
{
|
| 1763 |
+
"epoch": 0.5,
|
| 1764 |
+
"learning_rate": 3.492731110981856e-06,
|
| 1765 |
+
"loss": 0.2461,
|
| 1766 |
+
"step": 146500
|
| 1767 |
+
},
|
| 1768 |
+
{
|
| 1769 |
+
"epoch": 0.5,
|
| 1770 |
+
"learning_rate": 3.4807560055153918e-06,
|
| 1771 |
+
"loss": 0.1742,
|
| 1772 |
+
"step": 147000
|
| 1773 |
+
},
|
| 1774 |
+
{
|
| 1775 |
+
"epoch": 0.5,
|
| 1776 |
+
"learning_rate": 3.4687809000489266e-06,
|
| 1777 |
+
"loss": 0.1936,
|
| 1778 |
+
"step": 147500
|
| 1779 |
+
},
|
| 1780 |
+
{
|
| 1781 |
+
"epoch": 0.51,
|
| 1782 |
+
"learning_rate": 3.4568057945824623e-06,
|
| 1783 |
+
"loss": 0.1879,
|
| 1784 |
+
"step": 148000
|
| 1785 |
+
},
|
| 1786 |
+
{
|
| 1787 |
+
"epoch": 0.51,
|
| 1788 |
+
"learning_rate": 3.4448306891159976e-06,
|
| 1789 |
+
"loss": 0.24,
|
| 1790 |
+
"step": 148500
|
| 1791 |
+
},
|
| 1792 |
+
{
|
| 1793 |
+
"epoch": 0.51,
|
| 1794 |
+
"learning_rate": 3.432855583649533e-06,
|
| 1795 |
+
"loss": 0.2371,
|
| 1796 |
+
"step": 149000
|
| 1797 |
+
},
|
| 1798 |
+
{
|
| 1799 |
+
"epoch": 0.51,
|
| 1800 |
+
"learning_rate": 3.4208804781830682e-06,
|
| 1801 |
+
"loss": 0.253,
|
| 1802 |
+
"step": 149500
|
| 1803 |
+
},
|
| 1804 |
+
{
|
| 1805 |
+
"epoch": 0.51,
|
| 1806 |
+
"learning_rate": 3.408905372716604e-06,
|
| 1807 |
+
"loss": 0.1954,
|
| 1808 |
+
"step": 150000
|
| 1809 |
+
},
|
| 1810 |
+
{
|
| 1811 |
+
"epoch": 0.51,
|
| 1812 |
+
"learning_rate": 3.3969302672501397e-06,
|
| 1813 |
+
"loss": 0.2494,
|
| 1814 |
+
"step": 150500
|
| 1815 |
+
},
|
| 1816 |
+
{
|
| 1817 |
+
"epoch": 0.52,
|
| 1818 |
+
"learning_rate": 3.3849791119946076e-06,
|
| 1819 |
+
"loss": 0.1934,
|
| 1820 |
+
"step": 151000
|
| 1821 |
+
},
|
| 1822 |
+
{
|
| 1823 |
+
"epoch": 0.52,
|
| 1824 |
+
"learning_rate": 3.3730040065281433e-06,
|
| 1825 |
+
"loss": 0.2544,
|
| 1826 |
+
"step": 151500
|
| 1827 |
+
},
|
| 1828 |
+
{
|
| 1829 |
+
"epoch": 0.52,
|
| 1830 |
+
"learning_rate": 3.3610289010616786e-06,
|
| 1831 |
+
"loss": 0.2255,
|
| 1832 |
+
"step": 152000
|
| 1833 |
+
},
|
| 1834 |
+
{
|
| 1835 |
+
"epoch": 0.52,
|
| 1836 |
+
"learning_rate": 3.349053795595214e-06,
|
| 1837 |
+
"loss": 0.2103,
|
| 1838 |
+
"step": 152500
|
| 1839 |
+
},
|
| 1840 |
+
{
|
| 1841 |
+
"epoch": 0.52,
|
| 1842 |
+
"learning_rate": 3.3370786901287496e-06,
|
| 1843 |
+
"loss": 0.2033,
|
| 1844 |
+
"step": 153000
|
| 1845 |
+
},
|
| 1846 |
+
{
|
| 1847 |
+
"epoch": 0.53,
|
| 1848 |
+
"learning_rate": 3.325103584662285e-06,
|
| 1849 |
+
"loss": 0.1838,
|
| 1850 |
+
"step": 153500
|
| 1851 |
+
},
|
| 1852 |
+
{
|
| 1853 |
+
"epoch": 0.53,
|
| 1854 |
+
"learning_rate": 3.31312847919582e-06,
|
| 1855 |
+
"loss": 0.3,
|
| 1856 |
+
"step": 154000
|
| 1857 |
+
},
|
| 1858 |
+
{
|
| 1859 |
+
"epoch": 0.53,
|
| 1860 |
+
"learning_rate": 3.301153373729356e-06,
|
| 1861 |
+
"loss": 0.2067,
|
| 1862 |
+
"step": 154500
|
| 1863 |
+
},
|
| 1864 |
+
{
|
| 1865 |
+
"epoch": 0.53,
|
| 1866 |
+
"learning_rate": 3.289202218473824e-06,
|
| 1867 |
+
"loss": 0.2176,
|
| 1868 |
+
"step": 155000
|
| 1869 |
+
},
|
| 1870 |
+
{
|
| 1871 |
+
"epoch": 0.53,
|
| 1872 |
+
"learning_rate": 3.2772510632182922e-06,
|
| 1873 |
+
"loss": 0.2316,
|
| 1874 |
+
"step": 155500
|
| 1875 |
+
},
|
| 1876 |
+
{
|
| 1877 |
+
"epoch": 0.53,
|
| 1878 |
+
"learning_rate": 3.265275957751828e-06,
|
| 1879 |
+
"loss": 0.235,
|
| 1880 |
+
"step": 156000
|
| 1881 |
+
},
|
| 1882 |
+
{
|
| 1883 |
+
"epoch": 0.54,
|
| 1884 |
+
"learning_rate": 3.2533008522853632e-06,
|
| 1885 |
+
"loss": 0.2208,
|
| 1886 |
+
"step": 156500
|
| 1887 |
+
},
|
| 1888 |
+
{
|
| 1889 |
+
"epoch": 0.54,
|
| 1890 |
+
"learning_rate": 3.2413257468188985e-06,
|
| 1891 |
+
"loss": 0.2194,
|
| 1892 |
+
"step": 157000
|
| 1893 |
+
},
|
| 1894 |
+
{
|
| 1895 |
+
"epoch": 0.54,
|
| 1896 |
+
"learning_rate": 3.2293745915633673e-06,
|
| 1897 |
+
"loss": 0.2314,
|
| 1898 |
+
"step": 157500
|
| 1899 |
+
},
|
| 1900 |
+
{
|
| 1901 |
+
"epoch": 0.54,
|
| 1902 |
+
"learning_rate": 3.217399486096902e-06,
|
| 1903 |
+
"loss": 0.2367,
|
| 1904 |
+
"step": 158000
|
| 1905 |
+
},
|
| 1906 |
+
{
|
| 1907 |
+
"epoch": 0.54,
|
| 1908 |
+
"learning_rate": 3.205424380630438e-06,
|
| 1909 |
+
"loss": 0.184,
|
| 1910 |
+
"step": 158500
|
| 1911 |
+
},
|
| 1912 |
+
{
|
| 1913 |
+
"epoch": 0.54,
|
| 1914 |
+
"learning_rate": 3.1934492751639736e-06,
|
| 1915 |
+
"loss": 0.2172,
|
| 1916 |
+
"step": 159000
|
| 1917 |
+
},
|
| 1918 |
+
{
|
| 1919 |
+
"epoch": 0.55,
|
| 1920 |
+
"learning_rate": 3.1814981199084416e-06,
|
| 1921 |
+
"loss": 0.2203,
|
| 1922 |
+
"step": 159500
|
| 1923 |
+
},
|
| 1924 |
+
{
|
| 1925 |
+
"epoch": 0.55,
|
| 1926 |
+
"learning_rate": 3.1695230144419773e-06,
|
| 1927 |
+
"loss": 0.2066,
|
| 1928 |
+
"step": 160000
|
| 1929 |
+
},
|
| 1930 |
+
{
|
| 1931 |
+
"epoch": 0.55,
|
| 1932 |
+
"learning_rate": 3.1575479089755126e-06,
|
| 1933 |
+
"loss": 0.226,
|
| 1934 |
+
"step": 160500
|
| 1935 |
+
},
|
| 1936 |
+
{
|
| 1937 |
+
"epoch": 0.55,
|
| 1938 |
+
"learning_rate": 3.145572803509048e-06,
|
| 1939 |
+
"loss": 0.1993,
|
| 1940 |
+
"step": 161000
|
| 1941 |
+
},
|
| 1942 |
+
{
|
| 1943 |
+
"epoch": 0.55,
|
| 1944 |
+
"learning_rate": 3.1335976980425836e-06,
|
| 1945 |
+
"loss": 0.2209,
|
| 1946 |
+
"step": 161500
|
| 1947 |
+
},
|
| 1948 |
+
{
|
| 1949 |
+
"epoch": 0.55,
|
| 1950 |
+
"learning_rate": 3.121622592576119e-06,
|
| 1951 |
+
"loss": 0.2078,
|
| 1952 |
+
"step": 162000
|
| 1953 |
+
},
|
| 1954 |
+
{
|
| 1955 |
+
"epoch": 0.56,
|
| 1956 |
+
"learning_rate": 3.109647487109654e-06,
|
| 1957 |
+
"loss": 0.2263,
|
| 1958 |
+
"step": 162500
|
| 1959 |
+
},
|
| 1960 |
+
{
|
| 1961 |
+
"epoch": 0.56,
|
| 1962 |
+
"learning_rate": 3.0976723816431895e-06,
|
| 1963 |
+
"loss": 0.2194,
|
| 1964 |
+
"step": 163000
|
| 1965 |
+
},
|
| 1966 |
+
{
|
| 1967 |
+
"epoch": 0.56,
|
| 1968 |
+
"learning_rate": 3.085697276176725e-06,
|
| 1969 |
+
"loss": 0.2015,
|
| 1970 |
+
"step": 163500
|
| 1971 |
+
},
|
| 1972 |
+
{
|
| 1973 |
+
"epoch": 0.56,
|
| 1974 |
+
"learning_rate": 3.0737461209211936e-06,
|
| 1975 |
+
"loss": 0.2078,
|
| 1976 |
+
"step": 164000
|
| 1977 |
+
},
|
| 1978 |
+
{
|
| 1979 |
+
"epoch": 0.56,
|
| 1980 |
+
"learning_rate": 3.061771015454729e-06,
|
| 1981 |
+
"loss": 0.2465,
|
| 1982 |
+
"step": 164500
|
| 1983 |
+
},
|
| 1984 |
+
{
|
| 1985 |
+
"epoch": 0.56,
|
| 1986 |
+
"learning_rate": 3.049795909988264e-06,
|
| 1987 |
+
"loss": 0.2301,
|
| 1988 |
+
"step": 165000
|
| 1989 |
+
},
|
| 1990 |
+
{
|
| 1991 |
+
"epoch": 0.57,
|
| 1992 |
+
"learning_rate": 3.0378208045218e-06,
|
| 1993 |
+
"loss": 0.2014,
|
| 1994 |
+
"step": 165500
|
| 1995 |
+
},
|
| 1996 |
+
{
|
| 1997 |
+
"epoch": 0.57,
|
| 1998 |
+
"learning_rate": 3.025845699055335e-06,
|
| 1999 |
+
"loss": 0.1913,
|
| 2000 |
+
"step": 166000
|
| 2001 |
+
},
|
| 2002 |
+
{
|
| 2003 |
+
"epoch": 0.57,
|
| 2004 |
+
"learning_rate": 3.013870593588871e-06,
|
| 2005 |
+
"loss": 0.2305,
|
| 2006 |
+
"step": 166500
|
| 2007 |
+
},
|
| 2008 |
+
{
|
| 2009 |
+
"epoch": 0.57,
|
| 2010 |
+
"learning_rate": 3.0018954881224057e-06,
|
| 2011 |
+
"loss": 0.2185,
|
| 2012 |
+
"step": 167000
|
| 2013 |
+
},
|
| 2014 |
+
{
|
| 2015 |
+
"epoch": 0.57,
|
| 2016 |
+
"learning_rate": 2.9899443328668745e-06,
|
| 2017 |
+
"loss": 0.222,
|
| 2018 |
+
"step": 167500
|
| 2019 |
+
},
|
| 2020 |
+
{
|
| 2021 |
+
"epoch": 0.57,
|
| 2022 |
+
"learning_rate": 2.97796922740041e-06,
|
| 2023 |
+
"loss": 0.2073,
|
| 2024 |
+
"step": 168000
|
| 2025 |
+
},
|
| 2026 |
+
{
|
| 2027 |
+
"epoch": 0.58,
|
| 2028 |
+
"learning_rate": 2.965994121933945e-06,
|
| 2029 |
+
"loss": 0.189,
|
| 2030 |
+
"step": 168500
|
| 2031 |
+
},
|
| 2032 |
+
{
|
| 2033 |
+
"epoch": 0.58,
|
| 2034 |
+
"learning_rate": 2.954019016467481e-06,
|
| 2035 |
+
"loss": 0.205,
|
| 2036 |
+
"step": 169000
|
| 2037 |
+
},
|
| 2038 |
+
{
|
| 2039 |
+
"epoch": 0.58,
|
| 2040 |
+
"learning_rate": 2.942043911001016e-06,
|
| 2041 |
+
"loss": 0.211,
|
| 2042 |
+
"step": 169500
|
| 2043 |
+
},
|
| 2044 |
+
{
|
| 2045 |
+
"epoch": 0.58,
|
| 2046 |
+
"learning_rate": 2.9300688055345514e-06,
|
| 2047 |
+
"loss": 0.2355,
|
| 2048 |
+
"step": 170000
|
| 2049 |
+
},
|
| 2050 |
+
{
|
| 2051 |
+
"epoch": 0.58,
|
| 2052 |
+
"learning_rate": 2.918093700068087e-06,
|
| 2053 |
+
"loss": 0.2351,
|
| 2054 |
+
"step": 170500
|
| 2055 |
+
},
|
| 2056 |
+
{
|
| 2057 |
+
"epoch": 0.59,
|
| 2058 |
+
"learning_rate": 2.9061185946016224e-06,
|
| 2059 |
+
"loss": 0.2413,
|
| 2060 |
+
"step": 171000
|
| 2061 |
+
},
|
| 2062 |
+
{
|
| 2063 |
+
"epoch": 0.59,
|
| 2064 |
+
"learning_rate": 2.894167439346091e-06,
|
| 2065 |
+
"loss": 0.2304,
|
| 2066 |
+
"step": 171500
|
| 2067 |
+
},
|
| 2068 |
+
{
|
| 2069 |
+
"epoch": 0.59,
|
| 2070 |
+
"learning_rate": 2.8821923338796265e-06,
|
| 2071 |
+
"loss": 0.2138,
|
| 2072 |
+
"step": 172000
|
| 2073 |
+
},
|
| 2074 |
+
{
|
| 2075 |
+
"epoch": 0.59,
|
| 2076 |
+
"learning_rate": 2.8702172284131614e-06,
|
| 2077 |
+
"loss": 0.2223,
|
| 2078 |
+
"step": 172500
|
| 2079 |
+
},
|
| 2080 |
+
{
|
| 2081 |
+
"epoch": 0.59,
|
| 2082 |
+
"learning_rate": 2.858242122946697e-06,
|
| 2083 |
+
"loss": 0.2214,
|
| 2084 |
+
"step": 173000
|
| 2085 |
+
},
|
| 2086 |
+
{
|
| 2087 |
+
"epoch": 0.59,
|
| 2088 |
+
"learning_rate": 2.8462909676911655e-06,
|
| 2089 |
+
"loss": 0.2426,
|
| 2090 |
+
"step": 173500
|
| 2091 |
+
},
|
| 2092 |
+
{
|
| 2093 |
+
"epoch": 0.6,
|
| 2094 |
+
"learning_rate": 2.8343158622247008e-06,
|
| 2095 |
+
"loss": 0.2301,
|
| 2096 |
+
"step": 174000
|
| 2097 |
+
},
|
| 2098 |
+
{
|
| 2099 |
+
"epoch": 0.6,
|
| 2100 |
+
"learning_rate": 2.8223407567582365e-06,
|
| 2101 |
+
"loss": 0.2265,
|
| 2102 |
+
"step": 174500
|
| 2103 |
+
},
|
| 2104 |
+
{
|
| 2105 |
+
"epoch": 0.6,
|
| 2106 |
+
"learning_rate": 2.8103656512917713e-06,
|
| 2107 |
+
"loss": 0.2501,
|
| 2108 |
+
"step": 175000
|
| 2109 |
+
},
|
| 2110 |
+
{
|
| 2111 |
+
"epoch": 0.6,
|
| 2112 |
+
"learning_rate": 2.798390545825307e-06,
|
| 2113 |
+
"loss": 0.2074,
|
| 2114 |
+
"step": 175500
|
| 2115 |
+
},
|
| 2116 |
+
{
|
| 2117 |
+
"epoch": 0.6,
|
| 2118 |
+
"learning_rate": 2.7864154403588428e-06,
|
| 2119 |
+
"loss": 0.2511,
|
| 2120 |
+
"step": 176000
|
| 2121 |
+
},
|
| 2122 |
+
{
|
| 2123 |
+
"epoch": 0.6,
|
| 2124 |
+
"learning_rate": 2.774440334892378e-06,
|
| 2125 |
+
"loss": 0.2096,
|
| 2126 |
+
"step": 176500
|
| 2127 |
+
},
|
| 2128 |
+
{
|
| 2129 |
+
"epoch": 0.61,
|
| 2130 |
+
"learning_rate": 2.7624652294259134e-06,
|
| 2131 |
+
"loss": 0.222,
|
| 2132 |
+
"step": 177000
|
| 2133 |
+
},
|
| 2134 |
+
{
|
| 2135 |
+
"epoch": 0.61,
|
| 2136 |
+
"learning_rate": 2.7505140741703817e-06,
|
| 2137 |
+
"loss": 0.1944,
|
| 2138 |
+
"step": 177500
|
| 2139 |
+
},
|
| 2140 |
+
{
|
| 2141 |
+
"epoch": 0.61,
|
| 2142 |
+
"learning_rate": 2.738538968703917e-06,
|
| 2143 |
+
"loss": 0.2324,
|
| 2144 |
+
"step": 178000
|
| 2145 |
+
},
|
| 2146 |
+
{
|
| 2147 |
+
"epoch": 0.61,
|
| 2148 |
+
"learning_rate": 2.7265638632374527e-06,
|
| 2149 |
+
"loss": 0.229,
|
| 2150 |
+
"step": 178500
|
| 2151 |
+
},
|
| 2152 |
+
{
|
| 2153 |
+
"epoch": 0.61,
|
| 2154 |
+
"learning_rate": 2.714612707981921e-06,
|
| 2155 |
+
"loss": 0.249,
|
| 2156 |
+
"step": 179000
|
| 2157 |
+
},
|
| 2158 |
+
{
|
| 2159 |
+
"epoch": 0.61,
|
| 2160 |
+
"learning_rate": 2.7026376025154564e-06,
|
| 2161 |
+
"loss": 0.189,
|
| 2162 |
+
"step": 179500
|
| 2163 |
+
},
|
| 2164 |
+
{
|
| 2165 |
+
"epoch": 0.62,
|
| 2166 |
+
"learning_rate": 2.690662497048992e-06,
|
| 2167 |
+
"loss": 0.2328,
|
| 2168 |
+
"step": 180000
|
| 2169 |
+
},
|
| 2170 |
+
{
|
| 2171 |
+
"epoch": 0.62,
|
| 2172 |
+
"learning_rate": 2.678687391582527e-06,
|
| 2173 |
+
"loss": 0.2474,
|
| 2174 |
+
"step": 180500
|
| 2175 |
+
},
|
| 2176 |
+
{
|
| 2177 |
+
"epoch": 0.62,
|
| 2178 |
+
"learning_rate": 2.6667122861160627e-06,
|
| 2179 |
+
"loss": 0.1939,
|
| 2180 |
+
"step": 181000
|
| 2181 |
+
},
|
| 2182 |
+
{
|
| 2183 |
+
"epoch": 0.62,
|
| 2184 |
+
"learning_rate": 2.654737180649598e-06,
|
| 2185 |
+
"loss": 0.19,
|
| 2186 |
+
"step": 181500
|
| 2187 |
+
},
|
| 2188 |
+
{
|
| 2189 |
+
"epoch": 0.62,
|
| 2190 |
+
"learning_rate": 2.6427620751831337e-06,
|
| 2191 |
+
"loss": 0.2433,
|
| 2192 |
+
"step": 182000
|
| 2193 |
+
},
|
| 2194 |
+
{
|
| 2195 |
+
"epoch": 0.62,
|
| 2196 |
+
"learning_rate": 2.630786969716669e-06,
|
| 2197 |
+
"loss": 0.2171,
|
| 2198 |
+
"step": 182500
|
| 2199 |
+
},
|
| 2200 |
+
{
|
| 2201 |
+
"epoch": 0.63,
|
| 2202 |
+
"learning_rate": 2.618835814461137e-06,
|
| 2203 |
+
"loss": 0.1911,
|
| 2204 |
+
"step": 183000
|
| 2205 |
+
},
|
| 2206 |
+
{
|
| 2207 |
+
"epoch": 0.63,
|
| 2208 |
+
"learning_rate": 2.6068607089946727e-06,
|
| 2209 |
+
"loss": 0.2519,
|
| 2210 |
+
"step": 183500
|
| 2211 |
+
},
|
| 2212 |
+
{
|
| 2213 |
+
"epoch": 0.63,
|
| 2214 |
+
"learning_rate": 2.594909553739141e-06,
|
| 2215 |
+
"loss": 0.2094,
|
| 2216 |
+
"step": 184000
|
| 2217 |
+
},
|
| 2218 |
+
{
|
| 2219 |
+
"epoch": 0.63,
|
| 2220 |
+
"learning_rate": 2.5829344482726767e-06,
|
| 2221 |
+
"loss": 0.1911,
|
| 2222 |
+
"step": 184500
|
| 2223 |
+
},
|
| 2224 |
+
{
|
| 2225 |
+
"epoch": 0.63,
|
| 2226 |
+
"learning_rate": 2.570959342806212e-06,
|
| 2227 |
+
"loss": 0.218,
|
| 2228 |
+
"step": 185000
|
| 2229 |
+
},
|
| 2230 |
+
{
|
| 2231 |
+
"epoch": 0.63,
|
| 2232 |
+
"learning_rate": 2.5589842373397473e-06,
|
| 2233 |
+
"loss": 0.2081,
|
| 2234 |
+
"step": 185500
|
| 2235 |
+
},
|
| 2236 |
+
{
|
| 2237 |
+
"epoch": 0.64,
|
| 2238 |
+
"learning_rate": 2.547033082084216e-06,
|
| 2239 |
+
"loss": 0.2282,
|
| 2240 |
+
"step": 186000
|
| 2241 |
+
},
|
| 2242 |
+
{
|
| 2243 |
+
"epoch": 0.64,
|
| 2244 |
+
"learning_rate": 2.535057976617751e-06,
|
| 2245 |
+
"loss": 0.1888,
|
| 2246 |
+
"step": 186500
|
| 2247 |
+
},
|
| 2248 |
+
{
|
| 2249 |
+
"epoch": 0.64,
|
| 2250 |
+
"learning_rate": 2.5230828711512867e-06,
|
| 2251 |
+
"loss": 0.1858,
|
| 2252 |
+
"step": 187000
|
| 2253 |
+
},
|
| 2254 |
+
{
|
| 2255 |
+
"epoch": 0.64,
|
| 2256 |
+
"learning_rate": 2.511107765684822e-06,
|
| 2257 |
+
"loss": 0.218,
|
| 2258 |
+
"step": 187500
|
| 2259 |
+
},
|
| 2260 |
+
{
|
| 2261 |
+
"epoch": 0.64,
|
| 2262 |
+
"learning_rate": 2.4991326602183577e-06,
|
| 2263 |
+
"loss": 0.2614,
|
| 2264 |
+
"step": 188000
|
| 2265 |
+
},
|
| 2266 |
+
{
|
| 2267 |
+
"epoch": 0.64,
|
| 2268 |
+
"learning_rate": 2.4871575547518926e-06,
|
| 2269 |
+
"loss": 0.2407,
|
| 2270 |
+
"step": 188500
|
| 2271 |
+
},
|
| 2272 |
+
{
|
| 2273 |
+
"epoch": 0.65,
|
| 2274 |
+
"learning_rate": 2.475206399496361e-06,
|
| 2275 |
+
"loss": 0.2338,
|
| 2276 |
+
"step": 189000
|
| 2277 |
+
},
|
| 2278 |
+
{
|
| 2279 |
+
"epoch": 0.65,
|
| 2280 |
+
"learning_rate": 2.4632312940298967e-06,
|
| 2281 |
+
"loss": 0.1825,
|
| 2282 |
+
"step": 189500
|
| 2283 |
+
},
|
| 2284 |
+
{
|
| 2285 |
+
"epoch": 0.65,
|
| 2286 |
+
"learning_rate": 2.451256188563432e-06,
|
| 2287 |
+
"loss": 0.2095,
|
| 2288 |
+
"step": 190000
|
| 2289 |
+
},
|
| 2290 |
+
{
|
| 2291 |
+
"epoch": 0.65,
|
| 2292 |
+
"learning_rate": 2.4392810830969677e-06,
|
| 2293 |
+
"loss": 0.2295,
|
| 2294 |
+
"step": 190500
|
| 2295 |
+
},
|
| 2296 |
+
{
|
| 2297 |
+
"epoch": 0.65,
|
| 2298 |
+
"learning_rate": 2.427305977630503e-06,
|
| 2299 |
+
"loss": 0.2118,
|
| 2300 |
+
"step": 191000
|
| 2301 |
+
},
|
| 2302 |
+
{
|
| 2303 |
+
"epoch": 0.66,
|
| 2304 |
+
"learning_rate": 2.4153308721640383e-06,
|
| 2305 |
+
"loss": 0.2096,
|
| 2306 |
+
"step": 191500
|
| 2307 |
+
},
|
| 2308 |
+
{
|
| 2309 |
+
"epoch": 0.66,
|
| 2310 |
+
"learning_rate": 2.403355766697574e-06,
|
| 2311 |
+
"loss": 0.1916,
|
| 2312 |
+
"step": 192000
|
| 2313 |
+
},
|
| 2314 |
+
{
|
| 2315 |
+
"epoch": 0.66,
|
| 2316 |
+
"learning_rate": 2.3913806612311093e-06,
|
| 2317 |
+
"loss": 0.2059,
|
| 2318 |
+
"step": 192500
|
| 2319 |
+
},
|
| 2320 |
+
{
|
| 2321 |
+
"epoch": 0.66,
|
| 2322 |
+
"learning_rate": 2.3794055557646446e-06,
|
| 2323 |
+
"loss": 0.1891,
|
| 2324 |
+
"step": 193000
|
| 2325 |
+
},
|
| 2326 |
+
{
|
| 2327 |
+
"epoch": 0.66,
|
| 2328 |
+
"learning_rate": 2.367454400509113e-06,
|
| 2329 |
+
"loss": 0.2334,
|
| 2330 |
+
"step": 193500
|
| 2331 |
+
},
|
| 2332 |
+
{
|
| 2333 |
+
"epoch": 0.66,
|
| 2334 |
+
"learning_rate": 2.3554792950426482e-06,
|
| 2335 |
+
"loss": 0.2112,
|
| 2336 |
+
"step": 194000
|
| 2337 |
+
},
|
| 2338 |
+
{
|
| 2339 |
+
"epoch": 0.67,
|
| 2340 |
+
"learning_rate": 2.343504189576184e-06,
|
| 2341 |
+
"loss": 0.2117,
|
| 2342 |
+
"step": 194500
|
| 2343 |
+
},
|
| 2344 |
+
{
|
| 2345 |
+
"epoch": 0.67,
|
| 2346 |
+
"learning_rate": 2.3315290841097192e-06,
|
| 2347 |
+
"loss": 0.2576,
|
| 2348 |
+
"step": 195000
|
| 2349 |
+
},
|
| 2350 |
+
{
|
| 2351 |
+
"epoch": 0.67,
|
| 2352 |
+
"learning_rate": 2.3195539786432545e-06,
|
| 2353 |
+
"loss": 0.2264,
|
| 2354 |
+
"step": 195500
|
| 2355 |
+
},
|
| 2356 |
+
{
|
| 2357 |
+
"epoch": 0.67,
|
| 2358 |
+
"learning_rate": 2.3076028233877233e-06,
|
| 2359 |
+
"loss": 0.2078,
|
| 2360 |
+
"step": 196000
|
| 2361 |
+
},
|
| 2362 |
+
{
|
| 2363 |
+
"epoch": 0.67,
|
| 2364 |
+
"learning_rate": 2.295627717921258e-06,
|
| 2365 |
+
"loss": 0.2008,
|
| 2366 |
+
"step": 196500
|
| 2367 |
+
},
|
| 2368 |
+
{
|
| 2369 |
+
"epoch": 0.67,
|
| 2370 |
+
"learning_rate": 2.283652612454794e-06,
|
| 2371 |
+
"loss": 0.2493,
|
| 2372 |
+
"step": 197000
|
| 2373 |
+
},
|
| 2374 |
+
{
|
| 2375 |
+
"epoch": 0.68,
|
| 2376 |
+
"learning_rate": 2.2716775069883296e-06,
|
| 2377 |
+
"loss": 0.1625,
|
| 2378 |
+
"step": 197500
|
| 2379 |
+
},
|
| 2380 |
+
{
|
| 2381 |
+
"epoch": 0.68,
|
| 2382 |
+
"learning_rate": 2.259702401521865e-06,
|
| 2383 |
+
"loss": 0.2086,
|
| 2384 |
+
"step": 198000
|
| 2385 |
+
},
|
| 2386 |
+
{
|
| 2387 |
+
"epoch": 0.68,
|
| 2388 |
+
"learning_rate": 2.2477272960554e-06,
|
| 2389 |
+
"loss": 0.2326,
|
| 2390 |
+
"step": 198500
|
| 2391 |
+
},
|
| 2392 |
+
{
|
| 2393 |
+
"epoch": 0.68,
|
| 2394 |
+
"learning_rate": 2.2357761407998686e-06,
|
| 2395 |
+
"loss": 0.1936,
|
| 2396 |
+
"step": 199000
|
| 2397 |
+
},
|
| 2398 |
+
{
|
| 2399 |
+
"epoch": 0.68,
|
| 2400 |
+
"learning_rate": 2.223824985544337e-06,
|
| 2401 |
+
"loss": 0.2182,
|
| 2402 |
+
"step": 199500
|
| 2403 |
+
},
|
| 2404 |
+
{
|
| 2405 |
+
"epoch": 0.68,
|
| 2406 |
+
"learning_rate": 2.2118498800778722e-06,
|
| 2407 |
+
"loss": 0.2449,
|
| 2408 |
+
"step": 200000
|
| 2409 |
+
},
|
| 2410 |
+
{
|
| 2411 |
+
"epoch": 0.69,
|
| 2412 |
+
"learning_rate": 2.199874774611408e-06,
|
| 2413 |
+
"loss": 0.1762,
|
| 2414 |
+
"step": 200500
|
| 2415 |
+
},
|
| 2416 |
+
{
|
| 2417 |
+
"epoch": 0.69,
|
| 2418 |
+
"learning_rate": 2.1878996691449432e-06,
|
| 2419 |
+
"loss": 0.2033,
|
| 2420 |
+
"step": 201000
|
| 2421 |
+
},
|
| 2422 |
+
{
|
| 2423 |
+
"epoch": 0.69,
|
| 2424 |
+
"learning_rate": 2.1759245636784785e-06,
|
| 2425 |
+
"loss": 0.2165,
|
| 2426 |
+
"step": 201500
|
| 2427 |
+
},
|
| 2428 |
+
{
|
| 2429 |
+
"epoch": 0.69,
|
| 2430 |
+
"learning_rate": 2.163949458212014e-06,
|
| 2431 |
+
"loss": 0.1966,
|
| 2432 |
+
"step": 202000
|
| 2433 |
+
},
|
| 2434 |
+
{
|
| 2435 |
+
"epoch": 0.69,
|
| 2436 |
+
"learning_rate": 2.1519743527455495e-06,
|
| 2437 |
+
"loss": 0.207,
|
| 2438 |
+
"step": 202500
|
| 2439 |
+
},
|
| 2440 |
+
{
|
| 2441 |
+
"epoch": 0.69,
|
| 2442 |
+
"learning_rate": 2.1399992472790853e-06,
|
| 2443 |
+
"loss": 0.2306,
|
| 2444 |
+
"step": 203000
|
| 2445 |
+
},
|
| 2446 |
+
{
|
| 2447 |
+
"epoch": 0.7,
|
| 2448 |
+
"learning_rate": 2.12802414181262e-06,
|
| 2449 |
+
"loss": 0.2208,
|
| 2450 |
+
"step": 203500
|
| 2451 |
+
},
|
| 2452 |
+
{
|
| 2453 |
+
"epoch": 0.7,
|
| 2454 |
+
"learning_rate": 2.116072986557089e-06,
|
| 2455 |
+
"loss": 0.2058,
|
| 2456 |
+
"step": 204000
|
| 2457 |
+
},
|
| 2458 |
+
{
|
| 2459 |
+
"epoch": 0.7,
|
| 2460 |
+
"learning_rate": 2.104097881090624e-06,
|
| 2461 |
+
"loss": 0.1907,
|
| 2462 |
+
"step": 204500
|
| 2463 |
+
},
|
| 2464 |
+
{
|
| 2465 |
+
"epoch": 0.7,
|
| 2466 |
+
"learning_rate": 2.092146725835092e-06,
|
| 2467 |
+
"loss": 0.2182,
|
| 2468 |
+
"step": 205000
|
| 2469 |
+
},
|
| 2470 |
+
{
|
| 2471 |
+
"epoch": 0.7,
|
| 2472 |
+
"learning_rate": 2.080171620368628e-06,
|
| 2473 |
+
"loss": 0.2206,
|
| 2474 |
+
"step": 205500
|
| 2475 |
+
},
|
| 2476 |
+
{
|
| 2477 |
+
"epoch": 0.7,
|
| 2478 |
+
"learning_rate": 2.0681965149021636e-06,
|
| 2479 |
+
"loss": 0.1957,
|
| 2480 |
+
"step": 206000
|
| 2481 |
+
},
|
| 2482 |
+
{
|
| 2483 |
+
"epoch": 0.71,
|
| 2484 |
+
"learning_rate": 2.056221409435699e-06,
|
| 2485 |
+
"loss": 0.1911,
|
| 2486 |
+
"step": 206500
|
| 2487 |
+
},
|
| 2488 |
+
{
|
| 2489 |
+
"epoch": 0.71,
|
| 2490 |
+
"learning_rate": 2.044246303969234e-06,
|
| 2491 |
+
"loss": 0.2188,
|
| 2492 |
+
"step": 207000
|
| 2493 |
+
},
|
| 2494 |
+
{
|
| 2495 |
+
"epoch": 0.71,
|
| 2496 |
+
"learning_rate": 2.0322711985027695e-06,
|
| 2497 |
+
"loss": 0.2,
|
| 2498 |
+
"step": 207500
|
| 2499 |
+
},
|
| 2500 |
+
{
|
| 2501 |
+
"epoch": 0.71,
|
| 2502 |
+
"learning_rate": 2.020296093036305e-06,
|
| 2503 |
+
"loss": 0.1907,
|
| 2504 |
+
"step": 208000
|
| 2505 |
+
},
|
| 2506 |
+
{
|
| 2507 |
+
"epoch": 0.71,
|
| 2508 |
+
"learning_rate": 2.0083209875698405e-06,
|
| 2509 |
+
"loss": 0.2258,
|
| 2510 |
+
"step": 208500
|
| 2511 |
+
},
|
| 2512 |
+
{
|
| 2513 |
+
"epoch": 0.72,
|
| 2514 |
+
"learning_rate": 1.996369832314309e-06,
|
| 2515 |
+
"loss": 0.2048,
|
| 2516 |
+
"step": 209000
|
| 2517 |
+
},
|
| 2518 |
+
{
|
| 2519 |
+
"epoch": 0.72,
|
| 2520 |
+
"learning_rate": 1.984418677058777e-06,
|
| 2521 |
+
"loss": 0.213,
|
| 2522 |
+
"step": 209500
|
| 2523 |
+
},
|
| 2524 |
+
{
|
| 2525 |
+
"epoch": 0.72,
|
| 2526 |
+
"learning_rate": 1.972443571592313e-06,
|
| 2527 |
+
"loss": 0.2089,
|
| 2528 |
+
"step": 210000
|
| 2529 |
+
},
|
| 2530 |
+
{
|
| 2531 |
+
"epoch": 0.72,
|
| 2532 |
+
"learning_rate": 1.9604924163367813e-06,
|
| 2533 |
+
"loss": 0.2252,
|
| 2534 |
+
"step": 210500
|
| 2535 |
+
},
|
| 2536 |
+
{
|
| 2537 |
+
"epoch": 0.72,
|
| 2538 |
+
"learning_rate": 1.948517310870316e-06,
|
| 2539 |
+
"loss": 0.188,
|
| 2540 |
+
"step": 211000
|
| 2541 |
+
},
|
| 2542 |
+
{
|
| 2543 |
+
"epoch": 0.72,
|
| 2544 |
+
"learning_rate": 1.936542205403852e-06,
|
| 2545 |
+
"loss": 0.1943,
|
| 2546 |
+
"step": 211500
|
| 2547 |
+
},
|
| 2548 |
+
{
|
| 2549 |
+
"epoch": 0.73,
|
| 2550 |
+
"learning_rate": 1.924567099937387e-06,
|
| 2551 |
+
"loss": 0.1994,
|
| 2552 |
+
"step": 212000
|
| 2553 |
+
},
|
| 2554 |
+
{
|
| 2555 |
+
"epoch": 0.73,
|
| 2556 |
+
"learning_rate": 1.912591994470923e-06,
|
| 2557 |
+
"loss": 0.1892,
|
| 2558 |
+
"step": 212500
|
| 2559 |
+
},
|
| 2560 |
+
{
|
| 2561 |
+
"epoch": 0.73,
|
| 2562 |
+
"learning_rate": 1.9006168890044584e-06,
|
| 2563 |
+
"loss": 0.2104,
|
| 2564 |
+
"step": 213000
|
| 2565 |
+
},
|
| 2566 |
+
{
|
| 2567 |
+
"epoch": 0.73,
|
| 2568 |
+
"learning_rate": 1.8886417835379935e-06,
|
| 2569 |
+
"loss": 0.189,
|
| 2570 |
+
"step": 213500
|
| 2571 |
+
},
|
| 2572 |
+
{
|
| 2573 |
+
"epoch": 0.73,
|
| 2574 |
+
"learning_rate": 1.8766666780715292e-06,
|
| 2575 |
+
"loss": 0.203,
|
| 2576 |
+
"step": 214000
|
| 2577 |
+
},
|
| 2578 |
+
{
|
| 2579 |
+
"epoch": 0.73,
|
| 2580 |
+
"learning_rate": 1.8646915726050643e-06,
|
| 2581 |
+
"loss": 0.1794,
|
| 2582 |
+
"step": 214500
|
| 2583 |
+
},
|
| 2584 |
+
{
|
| 2585 |
+
"epoch": 0.74,
|
| 2586 |
+
"learning_rate": 1.8527164671386e-06,
|
| 2587 |
+
"loss": 0.2119,
|
| 2588 |
+
"step": 215000
|
| 2589 |
+
},
|
| 2590 |
+
{
|
| 2591 |
+
"epoch": 0.74,
|
| 2592 |
+
"learning_rate": 1.840741361672135e-06,
|
| 2593 |
+
"loss": 0.183,
|
| 2594 |
+
"step": 215500
|
| 2595 |
+
},
|
| 2596 |
+
{
|
| 2597 |
+
"epoch": 0.74,
|
| 2598 |
+
"learning_rate": 1.8287662562056708e-06,
|
| 2599 |
+
"loss": 0.1655,
|
| 2600 |
+
"step": 216000
|
| 2601 |
+
},
|
| 2602 |
+
{
|
| 2603 |
+
"epoch": 0.74,
|
| 2604 |
+
"learning_rate": 1.8167911507392063e-06,
|
| 2605 |
+
"loss": 0.1866,
|
| 2606 |
+
"step": 216500
|
| 2607 |
+
},
|
| 2608 |
+
{
|
| 2609 |
+
"epoch": 0.74,
|
| 2610 |
+
"learning_rate": 1.8048399954836742e-06,
|
| 2611 |
+
"loss": 0.202,
|
| 2612 |
+
"step": 217000
|
| 2613 |
+
},
|
| 2614 |
+
{
|
| 2615 |
+
"epoch": 0.74,
|
| 2616 |
+
"learning_rate": 1.79286489001721e-06,
|
| 2617 |
+
"loss": 0.2151,
|
| 2618 |
+
"step": 217500
|
| 2619 |
+
},
|
| 2620 |
+
{
|
| 2621 |
+
"epoch": 0.75,
|
| 2622 |
+
"learning_rate": 1.7808897845507454e-06,
|
| 2623 |
+
"loss": 0.1977,
|
| 2624 |
+
"step": 218000
|
| 2625 |
+
},
|
| 2626 |
+
{
|
| 2627 |
+
"epoch": 0.75,
|
| 2628 |
+
"learning_rate": 1.7689386292952136e-06,
|
| 2629 |
+
"loss": 0.2263,
|
| 2630 |
+
"step": 218500
|
| 2631 |
+
},
|
| 2632 |
+
{
|
| 2633 |
+
"epoch": 0.75,
|
| 2634 |
+
"learning_rate": 1.756963523828749e-06,
|
| 2635 |
+
"loss": 0.1919,
|
| 2636 |
+
"step": 219000
|
| 2637 |
+
},
|
| 2638 |
+
{
|
| 2639 |
+
"epoch": 0.75,
|
| 2640 |
+
"learning_rate": 1.7449884183622846e-06,
|
| 2641 |
+
"loss": 0.1726,
|
| 2642 |
+
"step": 219500
|
| 2643 |
+
},
|
| 2644 |
+
{
|
| 2645 |
+
"epoch": 0.75,
|
| 2646 |
+
"learning_rate": 1.73301331289582e-06,
|
| 2647 |
+
"loss": 0.198,
|
| 2648 |
+
"step": 220000
|
| 2649 |
+
},
|
| 2650 |
+
{
|
| 2651 |
+
"epoch": 0.75,
|
| 2652 |
+
"learning_rate": 1.7210382074293554e-06,
|
| 2653 |
+
"loss": 0.2128,
|
| 2654 |
+
"step": 220500
|
| 2655 |
+
},
|
| 2656 |
+
{
|
| 2657 |
+
"epoch": 0.76,
|
| 2658 |
+
"learning_rate": 1.709063101962891e-06,
|
| 2659 |
+
"loss": 0.2186,
|
| 2660 |
+
"step": 221000
|
| 2661 |
+
},
|
| 2662 |
+
{
|
| 2663 |
+
"epoch": 0.76,
|
| 2664 |
+
"learning_rate": 1.6970879964964262e-06,
|
| 2665 |
+
"loss": 0.2037,
|
| 2666 |
+
"step": 221500
|
| 2667 |
+
},
|
| 2668 |
+
{
|
| 2669 |
+
"epoch": 0.76,
|
| 2670 |
+
"learning_rate": 1.6851128910299617e-06,
|
| 2671 |
+
"loss": 0.1783,
|
| 2672 |
+
"step": 222000
|
| 2673 |
+
},
|
| 2674 |
+
{
|
| 2675 |
+
"epoch": 0.76,
|
| 2676 |
+
"learning_rate": 1.67316173577443e-06,
|
| 2677 |
+
"loss": 0.1906,
|
| 2678 |
+
"step": 222500
|
| 2679 |
+
},
|
| 2680 |
+
{
|
| 2681 |
+
"epoch": 0.76,
|
| 2682 |
+
"learning_rate": 1.6611866303079656e-06,
|
| 2683 |
+
"loss": 0.23,
|
| 2684 |
+
"step": 223000
|
| 2685 |
+
},
|
| 2686 |
+
{
|
| 2687 |
+
"epoch": 0.76,
|
| 2688 |
+
"learning_rate": 1.6492115248415009e-06,
|
| 2689 |
+
"loss": 0.2613,
|
| 2690 |
+
"step": 223500
|
| 2691 |
+
},
|
| 2692 |
+
{
|
| 2693 |
+
"epoch": 0.77,
|
| 2694 |
+
"learning_rate": 1.6372364193750364e-06,
|
| 2695 |
+
"loss": 0.2053,
|
| 2696 |
+
"step": 224000
|
| 2697 |
+
},
|
| 2698 |
+
{
|
| 2699 |
+
"epoch": 0.77,
|
| 2700 |
+
"learning_rate": 1.6252613139085717e-06,
|
| 2701 |
+
"loss": 0.2254,
|
| 2702 |
+
"step": 224500
|
| 2703 |
+
},
|
| 2704 |
+
{
|
| 2705 |
+
"epoch": 0.77,
|
| 2706 |
+
"learning_rate": 1.6133341088639731e-06,
|
| 2707 |
+
"loss": 0.185,
|
| 2708 |
+
"step": 225000
|
| 2709 |
+
},
|
| 2710 |
+
{
|
| 2711 |
+
"epoch": 0.77,
|
| 2712 |
+
"learning_rate": 1.6013590033975086e-06,
|
| 2713 |
+
"loss": 0.205,
|
| 2714 |
+
"step": 225500
|
| 2715 |
+
},
|
| 2716 |
+
{
|
| 2717 |
+
"epoch": 0.77,
|
| 2718 |
+
"learning_rate": 1.589383897931044e-06,
|
| 2719 |
+
"loss": 0.1897,
|
| 2720 |
+
"step": 226000
|
| 2721 |
+
},
|
| 2722 |
+
{
|
| 2723 |
+
"epoch": 0.77,
|
| 2724 |
+
"learning_rate": 1.5774327426755123e-06,
|
| 2725 |
+
"loss": 0.1755,
|
| 2726 |
+
"step": 226500
|
| 2727 |
+
},
|
| 2728 |
+
{
|
| 2729 |
+
"epoch": 0.78,
|
| 2730 |
+
"learning_rate": 1.5654576372090478e-06,
|
| 2731 |
+
"loss": 0.2278,
|
| 2732 |
+
"step": 227000
|
| 2733 |
+
},
|
| 2734 |
+
{
|
| 2735 |
+
"epoch": 0.78,
|
| 2736 |
+
"learning_rate": 1.553482531742583e-06,
|
| 2737 |
+
"loss": 0.1753,
|
| 2738 |
+
"step": 227500
|
| 2739 |
+
},
|
| 2740 |
+
{
|
| 2741 |
+
"epoch": 0.78,
|
| 2742 |
+
"learning_rate": 1.5415074262761186e-06,
|
| 2743 |
+
"loss": 0.207,
|
| 2744 |
+
"step": 228000
|
| 2745 |
+
},
|
| 2746 |
+
{
|
| 2747 |
+
"epoch": 0.78,
|
| 2748 |
+
"learning_rate": 1.5295323208096539e-06,
|
| 2749 |
+
"loss": 0.1999,
|
| 2750 |
+
"step": 228500
|
| 2751 |
+
},
|
| 2752 |
+
{
|
| 2753 |
+
"epoch": 0.78,
|
| 2754 |
+
"learning_rate": 1.5175572153431894e-06,
|
| 2755 |
+
"loss": 0.2763,
|
| 2756 |
+
"step": 229000
|
| 2757 |
+
},
|
| 2758 |
+
{
|
| 2759 |
+
"epoch": 0.79,
|
| 2760 |
+
"learning_rate": 1.5055821098767247e-06,
|
| 2761 |
+
"loss": 0.2363,
|
| 2762 |
+
"step": 229500
|
| 2763 |
+
},
|
| 2764 |
+
{
|
| 2765 |
+
"epoch": 0.79,
|
| 2766 |
+
"learning_rate": 1.4936070044102604e-06,
|
| 2767 |
+
"loss": 0.2022,
|
| 2768 |
+
"step": 230000
|
| 2769 |
+
},
|
| 2770 |
+
{
|
| 2771 |
+
"epoch": 0.79,
|
| 2772 |
+
"learning_rate": 1.4816318989437957e-06,
|
| 2773 |
+
"loss": 0.2565,
|
| 2774 |
+
"step": 230500
|
| 2775 |
+
},
|
| 2776 |
+
{
|
| 2777 |
+
"epoch": 0.79,
|
| 2778 |
+
"learning_rate": 1.4696567934773312e-06,
|
| 2779 |
+
"loss": 0.2224,
|
| 2780 |
+
"step": 231000
|
| 2781 |
+
},
|
| 2782 |
+
{
|
| 2783 |
+
"epoch": 0.79,
|
| 2784 |
+
"learning_rate": 1.4576816880108665e-06,
|
| 2785 |
+
"loss": 0.2018,
|
| 2786 |
+
"step": 231500
|
| 2787 |
+
},
|
| 2788 |
+
{
|
| 2789 |
+
"epoch": 0.79,
|
| 2790 |
+
"learning_rate": 1.445706582544402e-06,
|
| 2791 |
+
"loss": 0.1966,
|
| 2792 |
+
"step": 232000
|
| 2793 |
+
},
|
| 2794 |
+
{
|
| 2795 |
+
"epoch": 0.8,
|
| 2796 |
+
"learning_rate": 1.4337554272888703e-06,
|
| 2797 |
+
"loss": 0.2034,
|
| 2798 |
+
"step": 232500
|
| 2799 |
+
},
|
| 2800 |
+
{
|
| 2801 |
+
"epoch": 0.8,
|
| 2802 |
+
"learning_rate": 1.4217803218224056e-06,
|
| 2803 |
+
"loss": 0.2089,
|
| 2804 |
+
"step": 233000
|
| 2805 |
+
},
|
| 2806 |
+
{
|
| 2807 |
+
"epoch": 0.8,
|
| 2808 |
+
"learning_rate": 1.4098052163559411e-06,
|
| 2809 |
+
"loss": 0.1942,
|
| 2810 |
+
"step": 233500
|
| 2811 |
+
},
|
| 2812 |
+
{
|
| 2813 |
+
"epoch": 0.8,
|
| 2814 |
+
"learning_rate": 1.3978301108894764e-06,
|
| 2815 |
+
"loss": 0.1786,
|
| 2816 |
+
"step": 234000
|
| 2817 |
+
},
|
| 2818 |
+
{
|
| 2819 |
+
"epoch": 0.8,
|
| 2820 |
+
"learning_rate": 1.385878955633945e-06,
|
| 2821 |
+
"loss": 0.2404,
|
| 2822 |
+
"step": 234500
|
| 2823 |
+
},
|
| 2824 |
+
{
|
| 2825 |
+
"epoch": 0.8,
|
| 2826 |
+
"learning_rate": 1.3739038501674803e-06,
|
| 2827 |
+
"loss": 0.1703,
|
| 2828 |
+
"step": 235000
|
| 2829 |
+
},
|
| 2830 |
+
{
|
| 2831 |
+
"epoch": 0.81,
|
| 2832 |
+
"learning_rate": 1.3619526949119487e-06,
|
| 2833 |
+
"loss": 0.2025,
|
| 2834 |
+
"step": 235500
|
| 2835 |
+
},
|
| 2836 |
+
{
|
| 2837 |
+
"epoch": 0.81,
|
| 2838 |
+
"learning_rate": 1.3499775894454842e-06,
|
| 2839 |
+
"loss": 0.2186,
|
| 2840 |
+
"step": 236000
|
| 2841 |
+
},
|
| 2842 |
+
{
|
| 2843 |
+
"epoch": 0.81,
|
| 2844 |
+
"learning_rate": 1.3380024839790195e-06,
|
| 2845 |
+
"loss": 0.2175,
|
| 2846 |
+
"step": 236500
|
| 2847 |
+
},
|
| 2848 |
+
{
|
| 2849 |
+
"epoch": 0.81,
|
| 2850 |
+
"learning_rate": 1.326027378512555e-06,
|
| 2851 |
+
"loss": 0.2511,
|
| 2852 |
+
"step": 237000
|
| 2853 |
+
},
|
| 2854 |
+
{
|
| 2855 |
+
"epoch": 0.81,
|
| 2856 |
+
"learning_rate": 1.3140522730460905e-06,
|
| 2857 |
+
"loss": 0.2199,
|
| 2858 |
+
"step": 237500
|
| 2859 |
+
},
|
| 2860 |
+
{
|
| 2861 |
+
"epoch": 0.81,
|
| 2862 |
+
"learning_rate": 1.302077167579626e-06,
|
| 2863 |
+
"loss": 0.2199,
|
| 2864 |
+
"step": 238000
|
| 2865 |
+
},
|
| 2866 |
+
{
|
| 2867 |
+
"epoch": 0.82,
|
| 2868 |
+
"learning_rate": 1.2901020621131613e-06,
|
| 2869 |
+
"loss": 0.2085,
|
| 2870 |
+
"step": 238500
|
| 2871 |
+
},
|
| 2872 |
+
{
|
| 2873 |
+
"epoch": 0.82,
|
| 2874 |
+
"learning_rate": 1.2781269566466968e-06,
|
| 2875 |
+
"loss": 0.2038,
|
| 2876 |
+
"step": 239000
|
| 2877 |
+
},
|
| 2878 |
+
{
|
| 2879 |
+
"epoch": 0.82,
|
| 2880 |
+
"learning_rate": 1.2661758013911651e-06,
|
| 2881 |
+
"loss": 0.2175,
|
| 2882 |
+
"step": 239500
|
| 2883 |
+
},
|
| 2884 |
+
{
|
| 2885 |
+
"epoch": 0.82,
|
| 2886 |
+
"learning_rate": 1.2542006959247004e-06,
|
| 2887 |
+
"loss": 0.229,
|
| 2888 |
+
"step": 240000
|
| 2889 |
+
},
|
| 2890 |
+
{
|
| 2891 |
+
"epoch": 0.82,
|
| 2892 |
+
"learning_rate": 1.242225590458236e-06,
|
| 2893 |
+
"loss": 0.1929,
|
| 2894 |
+
"step": 240500
|
| 2895 |
+
},
|
| 2896 |
+
{
|
| 2897 |
+
"epoch": 0.82,
|
| 2898 |
+
"learning_rate": 1.2302504849917712e-06,
|
| 2899 |
+
"loss": 0.1896,
|
| 2900 |
+
"step": 241000
|
| 2901 |
+
},
|
| 2902 |
+
{
|
| 2903 |
+
"epoch": 0.83,
|
| 2904 |
+
"learning_rate": 1.2182753795253067e-06,
|
| 2905 |
+
"loss": 0.2221,
|
| 2906 |
+
"step": 241500
|
| 2907 |
+
},
|
| 2908 |
+
{
|
| 2909 |
+
"epoch": 0.83,
|
| 2910 |
+
"learning_rate": 1.2063002740588423e-06,
|
| 2911 |
+
"loss": 0.1957,
|
| 2912 |
+
"step": 242000
|
| 2913 |
+
},
|
| 2914 |
+
{
|
| 2915 |
+
"epoch": 0.83,
|
| 2916 |
+
"learning_rate": 1.1943251685923778e-06,
|
| 2917 |
+
"loss": 0.2127,
|
| 2918 |
+
"step": 242500
|
| 2919 |
+
},
|
| 2920 |
+
{
|
| 2921 |
+
"epoch": 0.83,
|
| 2922 |
+
"learning_rate": 1.182350063125913e-06,
|
| 2923 |
+
"loss": 0.1977,
|
| 2924 |
+
"step": 243000
|
| 2925 |
+
},
|
| 2926 |
+
{
|
| 2927 |
+
"epoch": 0.83,
|
| 2928 |
+
"learning_rate": 1.1703749576594486e-06,
|
| 2929 |
+
"loss": 0.1765,
|
| 2930 |
+
"step": 243500
|
| 2931 |
+
},
|
| 2932 |
+
{
|
| 2933 |
+
"epoch": 0.83,
|
| 2934 |
+
"learning_rate": 1.158423802403917e-06,
|
| 2935 |
+
"loss": 0.2475,
|
| 2936 |
+
"step": 244000
|
| 2937 |
+
},
|
| 2938 |
+
{
|
| 2939 |
+
"epoch": 0.84,
|
| 2940 |
+
"learning_rate": 1.1464486969374522e-06,
|
| 2941 |
+
"loss": 0.2178,
|
| 2942 |
+
"step": 244500
|
| 2943 |
+
},
|
| 2944 |
+
{
|
| 2945 |
+
"epoch": 0.84,
|
| 2946 |
+
"learning_rate": 1.1344735914709877e-06,
|
| 2947 |
+
"loss": 0.2157,
|
| 2948 |
+
"step": 245000
|
| 2949 |
+
},
|
| 2950 |
+
{
|
| 2951 |
+
"epoch": 0.84,
|
| 2952 |
+
"learning_rate": 1.122498486004523e-06,
|
| 2953 |
+
"loss": 0.2147,
|
| 2954 |
+
"step": 245500
|
| 2955 |
+
},
|
| 2956 |
+
{
|
| 2957 |
+
"epoch": 0.84,
|
| 2958 |
+
"learning_rate": 1.1105712809599244e-06,
|
| 2959 |
+
"loss": 0.1948,
|
| 2960 |
+
"step": 246000
|
| 2961 |
+
},
|
| 2962 |
+
{
|
| 2963 |
+
"epoch": 0.84,
|
| 2964 |
+
"learning_rate": 1.09859617549346e-06,
|
| 2965 |
+
"loss": 0.1968,
|
| 2966 |
+
"step": 246500
|
| 2967 |
+
},
|
| 2968 |
+
{
|
| 2969 |
+
"epoch": 0.85,
|
| 2970 |
+
"learning_rate": 1.0866210700269952e-06,
|
| 2971 |
+
"loss": 0.1725,
|
| 2972 |
+
"step": 247000
|
| 2973 |
+
},
|
| 2974 |
+
{
|
| 2975 |
+
"epoch": 0.85,
|
| 2976 |
+
"learning_rate": 1.0746459645605308e-06,
|
| 2977 |
+
"loss": 0.1864,
|
| 2978 |
+
"step": 247500
|
| 2979 |
+
},
|
| 2980 |
+
{
|
| 2981 |
+
"epoch": 0.85,
|
| 2982 |
+
"learning_rate": 1.062670859094066e-06,
|
| 2983 |
+
"loss": 0.2123,
|
| 2984 |
+
"step": 248000
|
| 2985 |
+
},
|
| 2986 |
+
{
|
| 2987 |
+
"epoch": 0.85,
|
| 2988 |
+
"learning_rate": 1.0506957536276015e-06,
|
| 2989 |
+
"loss": 0.1684,
|
| 2990 |
+
"step": 248500
|
| 2991 |
+
},
|
| 2992 |
+
{
|
| 2993 |
+
"epoch": 0.85,
|
| 2994 |
+
"learning_rate": 1.0387206481611368e-06,
|
| 2995 |
+
"loss": 0.138,
|
| 2996 |
+
"step": 249000
|
| 2997 |
+
},
|
| 2998 |
+
{
|
| 2999 |
+
"epoch": 0.85,
|
| 3000 |
+
"learning_rate": 1.0267455426946726e-06,
|
| 3001 |
+
"loss": 0.2054,
|
| 3002 |
+
"step": 249500
|
| 3003 |
+
},
|
| 3004 |
+
{
|
| 3005 |
+
"epoch": 0.86,
|
| 3006 |
+
"learning_rate": 1.0147704372282079e-06,
|
| 3007 |
+
"loss": 0.2042,
|
| 3008 |
+
"step": 250000
|
| 3009 |
+
},
|
| 3010 |
+
{
|
| 3011 |
+
"epoch": 0.86,
|
| 3012 |
+
"learning_rate": 1.0028192819726762e-06,
|
| 3013 |
+
"loss": 0.2019,
|
| 3014 |
+
"step": 250500
|
| 3015 |
+
},
|
| 3016 |
+
{
|
| 3017 |
+
"epoch": 0.86,
|
| 3018 |
+
"learning_rate": 9.908441765062117e-07,
|
| 3019 |
+
"loss": 0.2051,
|
| 3020 |
+
"step": 251000
|
| 3021 |
+
},
|
| 3022 |
+
{
|
| 3023 |
+
"epoch": 0.86,
|
| 3024 |
+
"learning_rate": 9.788930212506799e-07,
|
| 3025 |
+
"loss": 0.2206,
|
| 3026 |
+
"step": 251500
|
| 3027 |
+
},
|
| 3028 |
+
{
|
| 3029 |
+
"epoch": 0.86,
|
| 3030 |
+
"learning_rate": 9.669179157842154e-07,
|
| 3031 |
+
"loss": 0.1939,
|
| 3032 |
+
"step": 252000
|
| 3033 |
+
},
|
| 3034 |
+
{
|
| 3035 |
+
"epoch": 0.86,
|
| 3036 |
+
"learning_rate": 9.549428103177509e-07,
|
| 3037 |
+
"loss": 0.1783,
|
| 3038 |
+
"step": 252500
|
| 3039 |
+
},
|
| 3040 |
+
{
|
| 3041 |
+
"epoch": 0.87,
|
| 3042 |
+
"learning_rate": 9.429677048512863e-07,
|
| 3043 |
+
"loss": 0.2094,
|
| 3044 |
+
"step": 253000
|
| 3045 |
+
},
|
| 3046 |
+
{
|
| 3047 |
+
"epoch": 0.87,
|
| 3048 |
+
"learning_rate": 9.310165495957548e-07,
|
| 3049 |
+
"loss": 0.2094,
|
| 3050 |
+
"step": 253500
|
| 3051 |
+
},
|
| 3052 |
+
{
|
| 3053 |
+
"epoch": 0.87,
|
| 3054 |
+
"learning_rate": 9.190414441292902e-07,
|
| 3055 |
+
"loss": 0.2089,
|
| 3056 |
+
"step": 254000
|
| 3057 |
+
},
|
| 3058 |
+
{
|
| 3059 |
+
"epoch": 0.87,
|
| 3060 |
+
"learning_rate": 9.070663386628256e-07,
|
| 3061 |
+
"loss": 0.2105,
|
| 3062 |
+
"step": 254500
|
| 3063 |
+
},
|
| 3064 |
+
{
|
| 3065 |
+
"epoch": 0.87,
|
| 3066 |
+
"learning_rate": 8.95091233196361e-07,
|
| 3067 |
+
"loss": 0.2205,
|
| 3068 |
+
"step": 255000
|
| 3069 |
+
},
|
| 3070 |
+
{
|
| 3071 |
+
"epoch": 0.87,
|
| 3072 |
+
"learning_rate": 8.831161277298964e-07,
|
| 3073 |
+
"loss": 0.2192,
|
| 3074 |
+
"step": 255500
|
| 3075 |
+
},
|
| 3076 |
+
{
|
| 3077 |
+
"epoch": 0.88,
|
| 3078 |
+
"learning_rate": 8.711410222634319e-07,
|
| 3079 |
+
"loss": 0.2001,
|
| 3080 |
+
"step": 256000
|
| 3081 |
+
},
|
| 3082 |
+
{
|
| 3083 |
+
"epoch": 0.88,
|
| 3084 |
+
"learning_rate": 8.591659167969673e-07,
|
| 3085 |
+
"loss": 0.18,
|
| 3086 |
+
"step": 256500
|
| 3087 |
+
},
|
| 3088 |
+
{
|
| 3089 |
+
"epoch": 0.88,
|
| 3090 |
+
"learning_rate": 8.472147615414355e-07,
|
| 3091 |
+
"loss": 0.1814,
|
| 3092 |
+
"step": 257000
|
| 3093 |
+
},
|
| 3094 |
+
{
|
| 3095 |
+
"epoch": 0.88,
|
| 3096 |
+
"learning_rate": 8.35239656074971e-07,
|
| 3097 |
+
"loss": 0.2044,
|
| 3098 |
+
"step": 257500
|
| 3099 |
+
},
|
| 3100 |
+
{
|
| 3101 |
+
"epoch": 0.88,
|
| 3102 |
+
"learning_rate": 8.232645506085064e-07,
|
| 3103 |
+
"loss": 0.1876,
|
| 3104 |
+
"step": 258000
|
| 3105 |
+
},
|
| 3106 |
+
{
|
| 3107 |
+
"epoch": 0.88,
|
| 3108 |
+
"learning_rate": 8.112894451420418e-07,
|
| 3109 |
+
"loss": 0.1998,
|
| 3110 |
+
"step": 258500
|
| 3111 |
+
},
|
| 3112 |
+
{
|
| 3113 |
+
"epoch": 0.89,
|
| 3114 |
+
"learning_rate": 7.993143396755773e-07,
|
| 3115 |
+
"loss": 0.1667,
|
| 3116 |
+
"step": 259000
|
| 3117 |
+
},
|
| 3118 |
+
{
|
| 3119 |
+
"epoch": 0.89,
|
| 3120 |
+
"learning_rate": 7.873392342091127e-07,
|
| 3121 |
+
"loss": 0.2043,
|
| 3122 |
+
"step": 259500
|
| 3123 |
+
},
|
| 3124 |
+
{
|
| 3125 |
+
"epoch": 0.89,
|
| 3126 |
+
"learning_rate": 7.753641287426481e-07,
|
| 3127 |
+
"loss": 0.1759,
|
| 3128 |
+
"step": 260000
|
| 3129 |
+
},
|
| 3130 |
+
{
|
| 3131 |
+
"epoch": 0.89,
|
| 3132 |
+
"learning_rate": 7.633890232761835e-07,
|
| 3133 |
+
"loss": 0.1607,
|
| 3134 |
+
"step": 260500
|
| 3135 |
+
},
|
| 3136 |
+
{
|
| 3137 |
+
"epoch": 0.89,
|
| 3138 |
+
"learning_rate": 7.514378680206519e-07,
|
| 3139 |
+
"loss": 0.2327,
|
| 3140 |
+
"step": 261000
|
| 3141 |
+
},
|
| 3142 |
+
{
|
| 3143 |
+
"epoch": 0.89,
|
| 3144 |
+
"learning_rate": 7.394627625541873e-07,
|
| 3145 |
+
"loss": 0.205,
|
| 3146 |
+
"step": 261500
|
| 3147 |
+
},
|
| 3148 |
+
{
|
| 3149 |
+
"epoch": 0.9,
|
| 3150 |
+
"learning_rate": 7.274876570877228e-07,
|
| 3151 |
+
"loss": 0.202,
|
| 3152 |
+
"step": 262000
|
| 3153 |
+
},
|
| 3154 |
+
{
|
| 3155 |
+
"epoch": 0.9,
|
| 3156 |
+
"learning_rate": 7.155365018321912e-07,
|
| 3157 |
+
"loss": 0.1992,
|
| 3158 |
+
"step": 262500
|
| 3159 |
+
},
|
| 3160 |
+
{
|
| 3161 |
+
"epoch": 0.9,
|
| 3162 |
+
"learning_rate": 7.035613963657266e-07,
|
| 3163 |
+
"loss": 0.2099,
|
| 3164 |
+
"step": 263000
|
| 3165 |
+
},
|
| 3166 |
+
{
|
| 3167 |
+
"epoch": 0.9,
|
| 3168 |
+
"learning_rate": 6.915862908992621e-07,
|
| 3169 |
+
"loss": 0.225,
|
| 3170 |
+
"step": 263500
|
| 3171 |
+
},
|
| 3172 |
+
{
|
| 3173 |
+
"epoch": 0.9,
|
| 3174 |
+
"learning_rate": 6.796111854327975e-07,
|
| 3175 |
+
"loss": 0.1722,
|
| 3176 |
+
"step": 264000
|
| 3177 |
+
},
|
| 3178 |
+
{
|
| 3179 |
+
"epoch": 0.9,
|
| 3180 |
+
"learning_rate": 6.676360799663329e-07,
|
| 3181 |
+
"loss": 0.2016,
|
| 3182 |
+
"step": 264500
|
| 3183 |
+
},
|
| 3184 |
+
{
|
| 3185 |
+
"epoch": 0.91,
|
| 3186 |
+
"learning_rate": 6.556609744998684e-07,
|
| 3187 |
+
"loss": 0.193,
|
| 3188 |
+
"step": 265000
|
| 3189 |
+
},
|
| 3190 |
+
{
|
| 3191 |
+
"epoch": 0.91,
|
| 3192 |
+
"learning_rate": 6.436858690334037e-07,
|
| 3193 |
+
"loss": 0.222,
|
| 3194 |
+
"step": 265500
|
| 3195 |
+
},
|
| 3196 |
+
{
|
| 3197 |
+
"epoch": 0.91,
|
| 3198 |
+
"learning_rate": 6.317107635669391e-07,
|
| 3199 |
+
"loss": 0.1912,
|
| 3200 |
+
"step": 266000
|
| 3201 |
+
},
|
| 3202 |
+
{
|
| 3203 |
+
"epoch": 0.91,
|
| 3204 |
+
"learning_rate": 6.197356581004745e-07,
|
| 3205 |
+
"loss": 0.2108,
|
| 3206 |
+
"step": 266500
|
| 3207 |
+
},
|
| 3208 |
+
{
|
| 3209 |
+
"epoch": 0.91,
|
| 3210 |
+
"learning_rate": 6.0776055263401e-07,
|
| 3211 |
+
"loss": 0.1917,
|
| 3212 |
+
"step": 267000
|
| 3213 |
+
},
|
| 3214 |
+
{
|
| 3215 |
+
"epoch": 0.92,
|
| 3216 |
+
"learning_rate": 5.958093973784783e-07,
|
| 3217 |
+
"loss": 0.1795,
|
| 3218 |
+
"step": 267500
|
| 3219 |
+
},
|
| 3220 |
+
{
|
| 3221 |
+
"epoch": 0.92,
|
| 3222 |
+
"learning_rate": 5.838342919120137e-07,
|
| 3223 |
+
"loss": 0.2032,
|
| 3224 |
+
"step": 268000
|
| 3225 |
+
},
|
| 3226 |
+
{
|
| 3227 |
+
"epoch": 0.92,
|
| 3228 |
+
"learning_rate": 5.718591864455492e-07,
|
| 3229 |
+
"loss": 0.1946,
|
| 3230 |
+
"step": 268500
|
| 3231 |
+
},
|
| 3232 |
+
{
|
| 3233 |
+
"epoch": 0.92,
|
| 3234 |
+
"learning_rate": 5.598840809790846e-07,
|
| 3235 |
+
"loss": 0.1944,
|
| 3236 |
+
"step": 269000
|
| 3237 |
+
},
|
| 3238 |
+
{
|
| 3239 |
+
"epoch": 0.92,
|
| 3240 |
+
"learning_rate": 5.47932925723553e-07,
|
| 3241 |
+
"loss": 0.2188,
|
| 3242 |
+
"step": 269500
|
| 3243 |
+
},
|
| 3244 |
+
{
|
| 3245 |
+
"epoch": 0.92,
|
| 3246 |
+
"learning_rate": 5.359578202570884e-07,
|
| 3247 |
+
"loss": 0.2265,
|
| 3248 |
+
"step": 270000
|
| 3249 |
+
},
|
| 3250 |
+
{
|
| 3251 |
+
"epoch": 0.93,
|
| 3252 |
+
"learning_rate": 5.239827147906238e-07,
|
| 3253 |
+
"loss": 0.2165,
|
| 3254 |
+
"step": 270500
|
| 3255 |
+
},
|
| 3256 |
+
{
|
| 3257 |
+
"epoch": 0.93,
|
| 3258 |
+
"learning_rate": 5.120076093241592e-07,
|
| 3259 |
+
"loss": 0.1648,
|
| 3260 |
+
"step": 271000
|
| 3261 |
+
},
|
| 3262 |
+
{
|
| 3263 |
+
"epoch": 0.93,
|
| 3264 |
+
"learning_rate": 5.000325038576947e-07,
|
| 3265 |
+
"loss": 0.2164,
|
| 3266 |
+
"step": 271500
|
| 3267 |
+
},
|
| 3268 |
+
{
|
| 3269 |
+
"epoch": 0.93,
|
| 3270 |
+
"learning_rate": 4.880813486021631e-07,
|
| 3271 |
+
"loss": 0.2216,
|
| 3272 |
+
"step": 272000
|
| 3273 |
+
},
|
| 3274 |
+
{
|
| 3275 |
+
"epoch": 0.93,
|
| 3276 |
+
"learning_rate": 4.761062431356985e-07,
|
| 3277 |
+
"loss": 0.1969,
|
| 3278 |
+
"step": 272500
|
| 3279 |
+
},
|
| 3280 |
+
{
|
| 3281 |
+
"epoch": 0.93,
|
| 3282 |
+
"learning_rate": 4.641550878801668e-07,
|
| 3283 |
+
"loss": 0.1804,
|
| 3284 |
+
"step": 273000
|
| 3285 |
+
},
|
| 3286 |
+
{
|
| 3287 |
+
"epoch": 0.94,
|
| 3288 |
+
"learning_rate": 4.521799824137022e-07,
|
| 3289 |
+
"loss": 0.1979,
|
| 3290 |
+
"step": 273500
|
| 3291 |
+
},
|
| 3292 |
+
{
|
| 3293 |
+
"epoch": 0.94,
|
| 3294 |
+
"learning_rate": 4.4020487694723773e-07,
|
| 3295 |
+
"loss": 0.2244,
|
| 3296 |
+
"step": 274000
|
| 3297 |
+
},
|
| 3298 |
+
{
|
| 3299 |
+
"epoch": 0.94,
|
| 3300 |
+
"learning_rate": 4.2822977148077313e-07,
|
| 3301 |
+
"loss": 0.218,
|
| 3302 |
+
"step": 274500
|
| 3303 |
+
},
|
| 3304 |
+
{
|
| 3305 |
+
"epoch": 0.94,
|
| 3306 |
+
"learning_rate": 4.1625466601430853e-07,
|
| 3307 |
+
"loss": 0.2156,
|
| 3308 |
+
"step": 275000
|
| 3309 |
+
},
|
| 3310 |
+
{
|
| 3311 |
+
"epoch": 0.94,
|
| 3312 |
+
"learning_rate": 4.04279560547844e-07,
|
| 3313 |
+
"loss": 0.1865,
|
| 3314 |
+
"step": 275500
|
| 3315 |
+
},
|
| 3316 |
+
{
|
| 3317 |
+
"epoch": 0.94,
|
| 3318 |
+
"learning_rate": 3.9230445508137943e-07,
|
| 3319 |
+
"loss": 0.1838,
|
| 3320 |
+
"step": 276000
|
| 3321 |
+
},
|
| 3322 |
+
{
|
| 3323 |
+
"epoch": 0.95,
|
| 3324 |
+
"learning_rate": 3.8032934961491483e-07,
|
| 3325 |
+
"loss": 0.164,
|
| 3326 |
+
"step": 276500
|
| 3327 |
+
},
|
| 3328 |
+
{
|
| 3329 |
+
"epoch": 0.95,
|
| 3330 |
+
"learning_rate": 3.6835424414845023e-07,
|
| 3331 |
+
"loss": 0.1726,
|
| 3332 |
+
"step": 277000
|
| 3333 |
+
},
|
| 3334 |
+
{
|
| 3335 |
+
"epoch": 0.95,
|
| 3336 |
+
"learning_rate": 3.564030888929186e-07,
|
| 3337 |
+
"loss": 0.1883,
|
| 3338 |
+
"step": 277500
|
| 3339 |
+
},
|
| 3340 |
+
{
|
| 3341 |
+
"epoch": 0.95,
|
| 3342 |
+
"learning_rate": 3.44427983426454e-07,
|
| 3343 |
+
"loss": 0.1768,
|
| 3344 |
+
"step": 278000
|
| 3345 |
+
},
|
| 3346 |
+
{
|
| 3347 |
+
"epoch": 0.95,
|
| 3348 |
+
"learning_rate": 3.3245287795998945e-07,
|
| 3349 |
+
"loss": 0.2227,
|
| 3350 |
+
"step": 278500
|
| 3351 |
+
},
|
| 3352 |
+
{
|
| 3353 |
+
"epoch": 0.95,
|
| 3354 |
+
"learning_rate": 3.204777724935249e-07,
|
| 3355 |
+
"loss": 0.1388,
|
| 3356 |
+
"step": 279000
|
| 3357 |
+
},
|
| 3358 |
+
{
|
| 3359 |
+
"epoch": 0.96,
|
| 3360 |
+
"learning_rate": 3.085266172379932e-07,
|
| 3361 |
+
"loss": 0.2058,
|
| 3362 |
+
"step": 279500
|
| 3363 |
+
},
|
| 3364 |
+
{
|
| 3365 |
+
"epoch": 0.96,
|
| 3366 |
+
"learning_rate": 2.9655151177152866e-07,
|
| 3367 |
+
"loss": 0.2163,
|
| 3368 |
+
"step": 280000
|
| 3369 |
+
},
|
| 3370 |
+
{
|
| 3371 |
+
"epoch": 0.96,
|
| 3372 |
+
"learning_rate": 2.845764063050641e-07,
|
| 3373 |
+
"loss": 0.1807,
|
| 3374 |
+
"step": 280500
|
| 3375 |
+
},
|
| 3376 |
+
{
|
| 3377 |
+
"epoch": 0.96,
|
| 3378 |
+
"learning_rate": 2.726013008385995e-07,
|
| 3379 |
+
"loss": 0.1744,
|
| 3380 |
+
"step": 281000
|
| 3381 |
+
},
|
| 3382 |
+
{
|
| 3383 |
+
"epoch": 0.96,
|
| 3384 |
+
"learning_rate": 2.6062619537213497e-07,
|
| 3385 |
+
"loss": 0.1752,
|
| 3386 |
+
"step": 281500
|
| 3387 |
+
},
|
| 3388 |
+
{
|
| 3389 |
+
"epoch": 0.96,
|
| 3390 |
+
"learning_rate": 2.486510899056704e-07,
|
| 3391 |
+
"loss": 0.1851,
|
| 3392 |
+
"step": 282000
|
| 3393 |
+
},
|
| 3394 |
+
{
|
| 3395 |
+
"epoch": 0.97,
|
| 3396 |
+
"learning_rate": 2.366759844392058e-07,
|
| 3397 |
+
"loss": 0.194,
|
| 3398 |
+
"step": 282500
|
| 3399 |
+
},
|
| 3400 |
+
{
|
| 3401 |
+
"epoch": 0.97,
|
| 3402 |
+
"learning_rate": 2.2470087897274124e-07,
|
| 3403 |
+
"loss": 0.1932,
|
| 3404 |
+
"step": 283000
|
| 3405 |
+
},
|
| 3406 |
+
{
|
| 3407 |
+
"epoch": 0.97,
|
| 3408 |
+
"learning_rate": 2.1272577350627667e-07,
|
| 3409 |
+
"loss": 0.2132,
|
| 3410 |
+
"step": 283500
|
| 3411 |
+
},
|
| 3412 |
+
{
|
| 3413 |
+
"epoch": 0.97,
|
| 3414 |
+
"learning_rate": 2.0077461825074503e-07,
|
| 3415 |
+
"loss": 0.2121,
|
| 3416 |
+
"step": 284000
|
| 3417 |
+
},
|
| 3418 |
+
{
|
| 3419 |
+
"epoch": 0.97,
|
| 3420 |
+
"learning_rate": 1.8879951278428046e-07,
|
| 3421 |
+
"loss": 0.1818,
|
| 3422 |
+
"step": 284500
|
| 3423 |
+
},
|
| 3424 |
+
{
|
| 3425 |
+
"epoch": 0.98,
|
| 3426 |
+
"learning_rate": 1.7682440731781588e-07,
|
| 3427 |
+
"loss": 0.1572,
|
| 3428 |
+
"step": 285000
|
| 3429 |
+
},
|
| 3430 |
+
{
|
| 3431 |
+
"epoch": 0.98,
|
| 3432 |
+
"learning_rate": 1.648493018513513e-07,
|
| 3433 |
+
"loss": 0.1862,
|
| 3434 |
+
"step": 285500
|
| 3435 |
+
},
|
| 3436 |
+
{
|
| 3437 |
+
"epoch": 0.98,
|
| 3438 |
+
"learning_rate": 1.529220968067526e-07,
|
| 3439 |
+
"loss": 0.1985,
|
| 3440 |
+
"step": 286000
|
| 3441 |
+
},
|
| 3442 |
+
{
|
| 3443 |
+
"epoch": 0.98,
|
| 3444 |
+
"learning_rate": 1.4094699134028801e-07,
|
| 3445 |
+
"loss": 0.2008,
|
| 3446 |
+
"step": 286500
|
| 3447 |
+
},
|
| 3448 |
+
{
|
| 3449 |
+
"epoch": 0.98,
|
| 3450 |
+
"learning_rate": 1.2897188587382344e-07,
|
| 3451 |
+
"loss": 0.2292,
|
| 3452 |
+
"step": 287000
|
| 3453 |
+
},
|
| 3454 |
+
{
|
| 3455 |
+
"epoch": 0.98,
|
| 3456 |
+
"learning_rate": 1.1699678040735886e-07,
|
| 3457 |
+
"loss": 0.2061,
|
| 3458 |
+
"step": 287500
|
| 3459 |
+
},
|
| 3460 |
+
{
|
| 3461 |
+
"epoch": 0.99,
|
| 3462 |
+
"learning_rate": 1.050216749408943e-07,
|
| 3463 |
+
"loss": 0.2021,
|
| 3464 |
+
"step": 288000
|
| 3465 |
+
},
|
| 3466 |
+
{
|
| 3467 |
+
"epoch": 0.99,
|
| 3468 |
+
"learning_rate": 9.304656947442973e-08,
|
| 3469 |
+
"loss": 0.1685,
|
| 3470 |
+
"step": 288500
|
| 3471 |
+
},
|
| 3472 |
+
{
|
| 3473 |
+
"epoch": 0.99,
|
| 3474 |
+
"learning_rate": 8.107146400796515e-08,
|
| 3475 |
+
"loss": 0.1779,
|
| 3476 |
+
"step": 289000
|
| 3477 |
+
},
|
| 3478 |
+
{
|
| 3479 |
+
"epoch": 0.99,
|
| 3480 |
+
"learning_rate": 6.912030875243351e-08,
|
| 3481 |
+
"loss": 0.1905,
|
| 3482 |
+
"step": 289500
|
| 3483 |
+
},
|
| 3484 |
+
{
|
| 3485 |
+
"epoch": 0.99,
|
| 3486 |
+
"learning_rate": 5.7145203285968946e-08,
|
| 3487 |
+
"loss": 0.1863,
|
| 3488 |
+
"step": 290000
|
| 3489 |
+
},
|
| 3490 |
+
{
|
| 3491 |
+
"epoch": 0.99,
|
| 3492 |
+
"learning_rate": 4.5170097819504365e-08,
|
| 3493 |
+
"loss": 0.1859,
|
| 3494 |
+
"step": 290500
|
| 3495 |
+
},
|
| 3496 |
+
{
|
| 3497 |
+
"epoch": 1.0,
|
| 3498 |
+
"learning_rate": 3.3194992353039797e-08,
|
| 3499 |
+
"loss": 0.1752,
|
| 3500 |
+
"step": 291000
|
| 3501 |
+
},
|
| 3502 |
+
{
|
| 3503 |
+
"epoch": 1.0,
|
| 3504 |
+
"learning_rate": 2.1219886886575222e-08,
|
| 3505 |
+
"loss": 0.1982,
|
| 3506 |
+
"step": 291500
|
| 3507 |
+
},
|
| 3508 |
+
{
|
| 3509 |
+
"epoch": 1.0,
|
| 3510 |
+
"learning_rate": 9.244781420110651e-09,
|
| 3511 |
+
"loss": 0.1291,
|
| 3512 |
+
"step": 292000
|
| 3513 |
+
},
|
| 3514 |
+
{
|
| 3515 |
+
"epoch": 1.0,
|
| 3516 |
+
"eval_loss": 1.2412053346633911,
|
| 3517 |
+
"eval_rmse": 1.1140938997268677,
|
| 3518 |
+
"eval_runtime": 11.2425,
|
| 3519 |
+
"eval_samples_per_second": 136.18,
|
| 3520 |
+
"eval_steps_per_second": 17.078,
|
| 3521 |
+
"step": 292273
|
| 3522 |
+
}
|
| 3523 |
+
],
|
| 3524 |
+
"max_steps": 292273,
|
| 3525 |
+
"num_train_epochs": 1,
|
| 3526 |
+
"total_flos": 1.9031292466854525e+17,
|
| 3527 |
+
"trial_name": null,
|
| 3528 |
+
"trial_params": null
|
| 3529 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5481b3865e7aadda5e5f6fda38fb7738e69861f7eac074f1d541806e85ba32aa
|
| 3 |
+
size 3439
|