Upload 11 files
Browse files- README.md +51 -1
- all_results.json +8 -0
- config.json +32 -0
- generation_config.json +7 -0
- pytorch_model.bin +3 -0
- special_tokens_map.json +4 -0
- tokenizer.json +0 -0
- tokenizer_config.json +6 -0
- train_results.json +8 -0
- trainer_state.json +2863 -0
- training_args.bin +3 -0
README.md
CHANGED
|
@@ -1,3 +1,53 @@
|
|
| 1 |
---
|
| 2 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 3 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
+
tags:
|
| 3 |
+
- generated_from_trainer
|
| 4 |
+
datasets:
|
| 5 |
+
- customized
|
| 6 |
+
model-index:
|
| 7 |
+
- name: my_first_model
|
| 8 |
+
results: []
|
| 9 |
---
|
| 10 |
+
|
| 11 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 12 |
+
should probably proofread and complete it, then remove this comment. -->
|
| 13 |
+
|
| 14 |
+
# my_first_model
|
| 15 |
+
|
| 16 |
+
This model is a fine-tuned version of [/home/sxx/LMFlow/my_models/facebookgalactica-1.3b/](https://huggingface.co//home/sxx/LMFlow/my_models/facebookgalactica-1.3b/) on the customized dataset.
|
| 17 |
+
|
| 18 |
+
## Model description
|
| 19 |
+
|
| 20 |
+
More information needed
|
| 21 |
+
|
| 22 |
+
## Intended uses & limitations
|
| 23 |
+
|
| 24 |
+
More information needed
|
| 25 |
+
|
| 26 |
+
## Training and evaluation data
|
| 27 |
+
|
| 28 |
+
More information needed
|
| 29 |
+
|
| 30 |
+
## Training procedure
|
| 31 |
+
|
| 32 |
+
### Training hyperparameters
|
| 33 |
+
|
| 34 |
+
The following hyperparameters were used during training:
|
| 35 |
+
- learning_rate: 2e-05
|
| 36 |
+
- train_batch_size: 1
|
| 37 |
+
- eval_batch_size: 1
|
| 38 |
+
- seed: 42
|
| 39 |
+
- distributed_type: multi-GPU
|
| 40 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 41 |
+
- lr_scheduler_type: linear
|
| 42 |
+
- num_epochs: 1.0
|
| 43 |
+
|
| 44 |
+
### Training results
|
| 45 |
+
|
| 46 |
+
|
| 47 |
+
|
| 48 |
+
### Framework versions
|
| 49 |
+
|
| 50 |
+
- Transformers 4.28.0.dev0
|
| 51 |
+
- Pytorch 2.0.0+cu117
|
| 52 |
+
- Datasets 2.10.1
|
| 53 |
+
- Tokenizers 0.13.3
|
all_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"train_loss": 0.5445125763019975,
|
| 4 |
+
"train_runtime": 21957.5944,
|
| 5 |
+
"train_samples": 9475,
|
| 6 |
+
"train_samples_per_second": 0.432,
|
| 7 |
+
"train_steps_per_second": 0.432
|
| 8 |
+
}
|
config.json
ADDED
|
@@ -0,0 +1,32 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "/home/sxx/LMFlow/my_models/facebookgalactica-1.3b/",
|
| 3 |
+
"_remove_final_layer_norm": false,
|
| 4 |
+
"activation_dropout": 0.0,
|
| 5 |
+
"activation_function": "gelu",
|
| 6 |
+
"architectures": [
|
| 7 |
+
"OPTForCausalLM"
|
| 8 |
+
],
|
| 9 |
+
"attention_dropout": 0.1,
|
| 10 |
+
"bos_token_id": 0,
|
| 11 |
+
"do_layer_norm_before": true,
|
| 12 |
+
"dropout": 0.1,
|
| 13 |
+
"enable_bias": true,
|
| 14 |
+
"eos_token_id": 2,
|
| 15 |
+
"ffn_dim": 8192,
|
| 16 |
+
"hidden_size": 2048,
|
| 17 |
+
"init_std": 0.02,
|
| 18 |
+
"layer_norm_elementwise_affine": true,
|
| 19 |
+
"layerdrop": 0.0,
|
| 20 |
+
"learned_embeddings": true,
|
| 21 |
+
"max_position_embeddings": 2048,
|
| 22 |
+
"model_type": "opt",
|
| 23 |
+
"num_attention_heads": 32,
|
| 24 |
+
"num_hidden_layers": 24,
|
| 25 |
+
"pad_token_id": 1,
|
| 26 |
+
"scale_embeddings": false,
|
| 27 |
+
"torch_dtype": "bfloat16",
|
| 28 |
+
"transformers_version": "4.28.0.dev0",
|
| 29 |
+
"use_cache": true,
|
| 30 |
+
"vocab_size": 50000,
|
| 31 |
+
"word_embed_proj_dim": 2048
|
| 32 |
+
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_from_model_config": true,
|
| 3 |
+
"bos_token_id": 0,
|
| 4 |
+
"eos_token_id": 2,
|
| 5 |
+
"pad_token_id": 1,
|
| 6 |
+
"transformers_version": "4.28.0.dev0"
|
| 7 |
+
}
|
pytorch_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7e80bab0f60072920c66608c8f2bf8ba3bc0f4a670a22d68b421c7a538ec6a75
|
| 3 |
+
size 2630528157
|
special_tokens_map.json
ADDED
|
@@ -0,0 +1,4 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"eos_token": "</s>",
|
| 3 |
+
"pad_token": "</s>"
|
| 4 |
+
}
|
tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"clean_up_tokenization_spaces": true,
|
| 3 |
+
"model_max_length": 1000000000000000019884624838656,
|
| 4 |
+
"special_tokens_map_file": "/content/tokenizer/special_tokens_map.json",
|
| 5 |
+
"tokenizer_class": "PreTrainedTokenizerFast"
|
| 6 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"train_loss": 0.5445125763019975,
|
| 4 |
+
"train_runtime": 21957.5944,
|
| 5 |
+
"train_samples": 9475,
|
| 6 |
+
"train_samples_per_second": 0.432,
|
| 7 |
+
"train_steps_per_second": 0.432
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2863 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1.0,
|
| 5 |
+
"global_step": 9475,
|
| 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": 1.995778364116095e-05,
|
| 13 |
+
"loss": 0.7213,
|
| 14 |
+
"step": 20
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.0,
|
| 18 |
+
"learning_rate": 1.99155672823219e-05,
|
| 19 |
+
"loss": 0.6524,
|
| 20 |
+
"step": 40
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.01,
|
| 24 |
+
"learning_rate": 1.987335092348285e-05,
|
| 25 |
+
"loss": 0.645,
|
| 26 |
+
"step": 60
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.01,
|
| 30 |
+
"learning_rate": 1.98311345646438e-05,
|
| 31 |
+
"loss": 0.6331,
|
| 32 |
+
"step": 80
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.01,
|
| 36 |
+
"learning_rate": 1.978891820580475e-05,
|
| 37 |
+
"loss": 0.8327,
|
| 38 |
+
"step": 100
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.01,
|
| 42 |
+
"learning_rate": 1.97467018469657e-05,
|
| 43 |
+
"loss": 0.6765,
|
| 44 |
+
"step": 120
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.01,
|
| 48 |
+
"learning_rate": 1.970448548812665e-05,
|
| 49 |
+
"loss": 0.8367,
|
| 50 |
+
"step": 140
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.02,
|
| 54 |
+
"learning_rate": 1.96622691292876e-05,
|
| 55 |
+
"loss": 0.6278,
|
| 56 |
+
"step": 160
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.02,
|
| 60 |
+
"learning_rate": 1.962005277044855e-05,
|
| 61 |
+
"loss": 0.6081,
|
| 62 |
+
"step": 180
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.02,
|
| 66 |
+
"learning_rate": 1.95778364116095e-05,
|
| 67 |
+
"loss": 0.6156,
|
| 68 |
+
"step": 200
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.02,
|
| 72 |
+
"learning_rate": 1.953562005277045e-05,
|
| 73 |
+
"loss": 0.5935,
|
| 74 |
+
"step": 220
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.03,
|
| 78 |
+
"learning_rate": 1.94934036939314e-05,
|
| 79 |
+
"loss": 0.4673,
|
| 80 |
+
"step": 240
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.03,
|
| 84 |
+
"learning_rate": 1.945118733509235e-05,
|
| 85 |
+
"loss": 0.7047,
|
| 86 |
+
"step": 260
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.03,
|
| 90 |
+
"learning_rate": 1.94089709762533e-05,
|
| 91 |
+
"loss": 0.371,
|
| 92 |
+
"step": 280
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 0.03,
|
| 96 |
+
"learning_rate": 1.9366754617414248e-05,
|
| 97 |
+
"loss": 0.4462,
|
| 98 |
+
"step": 300
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.03,
|
| 102 |
+
"learning_rate": 1.93245382585752e-05,
|
| 103 |
+
"loss": 0.4483,
|
| 104 |
+
"step": 320
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 0.04,
|
| 108 |
+
"learning_rate": 1.9282321899736148e-05,
|
| 109 |
+
"loss": 0.6366,
|
| 110 |
+
"step": 340
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.04,
|
| 114 |
+
"learning_rate": 1.92401055408971e-05,
|
| 115 |
+
"loss": 0.5771,
|
| 116 |
+
"step": 360
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 0.04,
|
| 120 |
+
"learning_rate": 1.919788918205805e-05,
|
| 121 |
+
"loss": 0.6417,
|
| 122 |
+
"step": 380
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.04,
|
| 126 |
+
"learning_rate": 1.9155672823219e-05,
|
| 127 |
+
"loss": 0.5874,
|
| 128 |
+
"step": 400
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.04,
|
| 132 |
+
"learning_rate": 1.911345646437995e-05,
|
| 133 |
+
"loss": 0.7817,
|
| 134 |
+
"step": 420
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 0.05,
|
| 138 |
+
"learning_rate": 1.90712401055409e-05,
|
| 139 |
+
"loss": 0.5381,
|
| 140 |
+
"step": 440
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 0.05,
|
| 144 |
+
"learning_rate": 1.902902374670185e-05,
|
| 145 |
+
"loss": 0.6428,
|
| 146 |
+
"step": 460
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.05,
|
| 150 |
+
"learning_rate": 1.89868073878628e-05,
|
| 151 |
+
"loss": 0.6244,
|
| 152 |
+
"step": 480
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 0.05,
|
| 156 |
+
"learning_rate": 1.894459102902375e-05,
|
| 157 |
+
"loss": 0.5345,
|
| 158 |
+
"step": 500
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 0.05,
|
| 162 |
+
"learning_rate": 1.89023746701847e-05,
|
| 163 |
+
"loss": 0.6411,
|
| 164 |
+
"step": 520
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.06,
|
| 168 |
+
"learning_rate": 1.886015831134565e-05,
|
| 169 |
+
"loss": 0.6699,
|
| 170 |
+
"step": 540
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.06,
|
| 174 |
+
"learning_rate": 1.8817941952506598e-05,
|
| 175 |
+
"loss": 0.5649,
|
| 176 |
+
"step": 560
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 0.06,
|
| 180 |
+
"learning_rate": 1.877572559366755e-05,
|
| 181 |
+
"loss": 0.4107,
|
| 182 |
+
"step": 580
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 0.06,
|
| 186 |
+
"learning_rate": 1.8733509234828498e-05,
|
| 187 |
+
"loss": 0.3438,
|
| 188 |
+
"step": 600
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 0.07,
|
| 192 |
+
"learning_rate": 1.869129287598945e-05,
|
| 193 |
+
"loss": 0.5882,
|
| 194 |
+
"step": 620
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 0.07,
|
| 198 |
+
"learning_rate": 1.8649076517150398e-05,
|
| 199 |
+
"loss": 0.5149,
|
| 200 |
+
"step": 640
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 0.07,
|
| 204 |
+
"learning_rate": 1.8606860158311346e-05,
|
| 205 |
+
"loss": 0.4901,
|
| 206 |
+
"step": 660
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.07,
|
| 210 |
+
"learning_rate": 1.8564643799472297e-05,
|
| 211 |
+
"loss": 0.6303,
|
| 212 |
+
"step": 680
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.07,
|
| 216 |
+
"learning_rate": 1.8522427440633246e-05,
|
| 217 |
+
"loss": 0.5993,
|
| 218 |
+
"step": 700
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 0.08,
|
| 222 |
+
"learning_rate": 1.8480211081794197e-05,
|
| 223 |
+
"loss": 0.5937,
|
| 224 |
+
"step": 720
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 0.08,
|
| 228 |
+
"learning_rate": 1.8437994722955145e-05,
|
| 229 |
+
"loss": 0.6804,
|
| 230 |
+
"step": 740
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 0.08,
|
| 234 |
+
"learning_rate": 1.8395778364116097e-05,
|
| 235 |
+
"loss": 0.6918,
|
| 236 |
+
"step": 760
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 0.08,
|
| 240 |
+
"learning_rate": 1.8353562005277045e-05,
|
| 241 |
+
"loss": 0.648,
|
| 242 |
+
"step": 780
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 0.08,
|
| 246 |
+
"learning_rate": 1.8311345646437997e-05,
|
| 247 |
+
"loss": 0.4871,
|
| 248 |
+
"step": 800
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.09,
|
| 252 |
+
"learning_rate": 1.8269129287598945e-05,
|
| 253 |
+
"loss": 0.6124,
|
| 254 |
+
"step": 820
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.09,
|
| 258 |
+
"learning_rate": 1.8226912928759896e-05,
|
| 259 |
+
"loss": 0.6574,
|
| 260 |
+
"step": 840
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 0.09,
|
| 264 |
+
"learning_rate": 1.8184696569920845e-05,
|
| 265 |
+
"loss": 0.4374,
|
| 266 |
+
"step": 860
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 0.09,
|
| 270 |
+
"learning_rate": 1.8142480211081796e-05,
|
| 271 |
+
"loss": 0.6513,
|
| 272 |
+
"step": 880
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 0.09,
|
| 276 |
+
"learning_rate": 1.8100263852242744e-05,
|
| 277 |
+
"loss": 0.7598,
|
| 278 |
+
"step": 900
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 0.1,
|
| 282 |
+
"learning_rate": 1.8058047493403696e-05,
|
| 283 |
+
"loss": 0.6105,
|
| 284 |
+
"step": 920
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 0.1,
|
| 288 |
+
"learning_rate": 1.8015831134564644e-05,
|
| 289 |
+
"loss": 0.6589,
|
| 290 |
+
"step": 940
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.1,
|
| 294 |
+
"learning_rate": 1.7973614775725596e-05,
|
| 295 |
+
"loss": 0.4167,
|
| 296 |
+
"step": 960
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.1,
|
| 300 |
+
"learning_rate": 1.7931398416886544e-05,
|
| 301 |
+
"loss": 0.5617,
|
| 302 |
+
"step": 980
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 0.11,
|
| 306 |
+
"learning_rate": 1.7889182058047495e-05,
|
| 307 |
+
"loss": 0.4674,
|
| 308 |
+
"step": 1000
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 0.11,
|
| 312 |
+
"learning_rate": 1.7846965699208444e-05,
|
| 313 |
+
"loss": 0.7633,
|
| 314 |
+
"step": 1020
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 0.11,
|
| 318 |
+
"learning_rate": 1.7804749340369395e-05,
|
| 319 |
+
"loss": 0.5497,
|
| 320 |
+
"step": 1040
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 0.11,
|
| 324 |
+
"learning_rate": 1.7762532981530343e-05,
|
| 325 |
+
"loss": 0.3297,
|
| 326 |
+
"step": 1060
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 0.11,
|
| 330 |
+
"learning_rate": 1.7720316622691295e-05,
|
| 331 |
+
"loss": 0.5251,
|
| 332 |
+
"step": 1080
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.12,
|
| 336 |
+
"learning_rate": 1.7678100263852246e-05,
|
| 337 |
+
"loss": 0.6771,
|
| 338 |
+
"step": 1100
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.12,
|
| 342 |
+
"learning_rate": 1.7635883905013195e-05,
|
| 343 |
+
"loss": 0.6371,
|
| 344 |
+
"step": 1120
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 0.12,
|
| 348 |
+
"learning_rate": 1.7593667546174146e-05,
|
| 349 |
+
"loss": 0.5774,
|
| 350 |
+
"step": 1140
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 0.12,
|
| 354 |
+
"learning_rate": 1.7551451187335094e-05,
|
| 355 |
+
"loss": 0.5764,
|
| 356 |
+
"step": 1160
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 0.12,
|
| 360 |
+
"learning_rate": 1.7509234828496046e-05,
|
| 361 |
+
"loss": 0.5529,
|
| 362 |
+
"step": 1180
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 0.13,
|
| 366 |
+
"learning_rate": 1.7467018469656994e-05,
|
| 367 |
+
"loss": 0.6113,
|
| 368 |
+
"step": 1200
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 0.13,
|
| 372 |
+
"learning_rate": 1.7424802110817946e-05,
|
| 373 |
+
"loss": 0.5255,
|
| 374 |
+
"step": 1220
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 0.13,
|
| 378 |
+
"learning_rate": 1.7382585751978894e-05,
|
| 379 |
+
"loss": 0.4578,
|
| 380 |
+
"step": 1240
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 0.13,
|
| 384 |
+
"learning_rate": 1.7340369393139842e-05,
|
| 385 |
+
"loss": 0.4262,
|
| 386 |
+
"step": 1260
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 0.14,
|
| 390 |
+
"learning_rate": 1.7298153034300794e-05,
|
| 391 |
+
"loss": 0.5446,
|
| 392 |
+
"step": 1280
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 0.14,
|
| 396 |
+
"learning_rate": 1.7255936675461742e-05,
|
| 397 |
+
"loss": 0.5746,
|
| 398 |
+
"step": 1300
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 0.14,
|
| 402 |
+
"learning_rate": 1.7213720316622693e-05,
|
| 403 |
+
"loss": 0.6115,
|
| 404 |
+
"step": 1320
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 0.14,
|
| 408 |
+
"learning_rate": 1.717150395778364e-05,
|
| 409 |
+
"loss": 0.5768,
|
| 410 |
+
"step": 1340
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 0.14,
|
| 414 |
+
"learning_rate": 1.7129287598944593e-05,
|
| 415 |
+
"loss": 0.5883,
|
| 416 |
+
"step": 1360
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 0.15,
|
| 420 |
+
"learning_rate": 1.708707124010554e-05,
|
| 421 |
+
"loss": 0.4796,
|
| 422 |
+
"step": 1380
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 0.15,
|
| 426 |
+
"learning_rate": 1.7044854881266493e-05,
|
| 427 |
+
"loss": 0.3795,
|
| 428 |
+
"step": 1400
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 0.15,
|
| 432 |
+
"learning_rate": 1.700263852242744e-05,
|
| 433 |
+
"loss": 0.4755,
|
| 434 |
+
"step": 1420
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 0.15,
|
| 438 |
+
"learning_rate": 1.6960422163588393e-05,
|
| 439 |
+
"loss": 0.3378,
|
| 440 |
+
"step": 1440
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 0.15,
|
| 444 |
+
"learning_rate": 1.691820580474934e-05,
|
| 445 |
+
"loss": 0.55,
|
| 446 |
+
"step": 1460
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 0.16,
|
| 450 |
+
"learning_rate": 1.6875989445910292e-05,
|
| 451 |
+
"loss": 0.6011,
|
| 452 |
+
"step": 1480
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 0.16,
|
| 456 |
+
"learning_rate": 1.683377308707124e-05,
|
| 457 |
+
"loss": 0.8145,
|
| 458 |
+
"step": 1500
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 0.16,
|
| 462 |
+
"learning_rate": 1.6791556728232192e-05,
|
| 463 |
+
"loss": 0.5853,
|
| 464 |
+
"step": 1520
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 0.16,
|
| 468 |
+
"learning_rate": 1.674934036939314e-05,
|
| 469 |
+
"loss": 0.4712,
|
| 470 |
+
"step": 1540
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 0.16,
|
| 474 |
+
"learning_rate": 1.6707124010554092e-05,
|
| 475 |
+
"loss": 0.6111,
|
| 476 |
+
"step": 1560
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 0.17,
|
| 480 |
+
"learning_rate": 1.666490765171504e-05,
|
| 481 |
+
"loss": 0.7256,
|
| 482 |
+
"step": 1580
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 0.17,
|
| 486 |
+
"learning_rate": 1.6622691292875988e-05,
|
| 487 |
+
"loss": 0.6924,
|
| 488 |
+
"step": 1600
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 0.17,
|
| 492 |
+
"learning_rate": 1.658047493403694e-05,
|
| 493 |
+
"loss": 0.3561,
|
| 494 |
+
"step": 1620
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 0.17,
|
| 498 |
+
"learning_rate": 1.653825857519789e-05,
|
| 499 |
+
"loss": 0.9116,
|
| 500 |
+
"step": 1640
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 0.18,
|
| 504 |
+
"learning_rate": 1.649604221635884e-05,
|
| 505 |
+
"loss": 0.5273,
|
| 506 |
+
"step": 1660
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 0.18,
|
| 510 |
+
"learning_rate": 1.645382585751979e-05,
|
| 511 |
+
"loss": 0.6573,
|
| 512 |
+
"step": 1680
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 0.18,
|
| 516 |
+
"learning_rate": 1.641160949868074e-05,
|
| 517 |
+
"loss": 0.537,
|
| 518 |
+
"step": 1700
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 0.18,
|
| 522 |
+
"learning_rate": 1.636939313984169e-05,
|
| 523 |
+
"loss": 0.4107,
|
| 524 |
+
"step": 1720
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 0.18,
|
| 528 |
+
"learning_rate": 1.632717678100264e-05,
|
| 529 |
+
"loss": 0.5239,
|
| 530 |
+
"step": 1740
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 0.19,
|
| 534 |
+
"learning_rate": 1.628496042216359e-05,
|
| 535 |
+
"loss": 0.4254,
|
| 536 |
+
"step": 1760
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 0.19,
|
| 540 |
+
"learning_rate": 1.624274406332454e-05,
|
| 541 |
+
"loss": 0.7679,
|
| 542 |
+
"step": 1780
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 0.19,
|
| 546 |
+
"learning_rate": 1.620052770448549e-05,
|
| 547 |
+
"loss": 0.5916,
|
| 548 |
+
"step": 1800
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 0.19,
|
| 552 |
+
"learning_rate": 1.6158311345646442e-05,
|
| 553 |
+
"loss": 0.4871,
|
| 554 |
+
"step": 1820
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 0.19,
|
| 558 |
+
"learning_rate": 1.611609498680739e-05,
|
| 559 |
+
"loss": 0.354,
|
| 560 |
+
"step": 1840
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 0.2,
|
| 564 |
+
"learning_rate": 1.607387862796834e-05,
|
| 565 |
+
"loss": 0.7787,
|
| 566 |
+
"step": 1860
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 0.2,
|
| 570 |
+
"learning_rate": 1.603166226912929e-05,
|
| 571 |
+
"loss": 0.5968,
|
| 572 |
+
"step": 1880
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 0.2,
|
| 576 |
+
"learning_rate": 1.5989445910290238e-05,
|
| 577 |
+
"loss": 0.6694,
|
| 578 |
+
"step": 1900
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 0.2,
|
| 582 |
+
"learning_rate": 1.594722955145119e-05,
|
| 583 |
+
"loss": 0.6532,
|
| 584 |
+
"step": 1920
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 0.2,
|
| 588 |
+
"learning_rate": 1.5905013192612138e-05,
|
| 589 |
+
"loss": 0.6289,
|
| 590 |
+
"step": 1940
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 0.21,
|
| 594 |
+
"learning_rate": 1.586279683377309e-05,
|
| 595 |
+
"loss": 0.3906,
|
| 596 |
+
"step": 1960
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 0.21,
|
| 600 |
+
"learning_rate": 1.5820580474934038e-05,
|
| 601 |
+
"loss": 0.5025,
|
| 602 |
+
"step": 1980
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 0.21,
|
| 606 |
+
"learning_rate": 1.577836411609499e-05,
|
| 607 |
+
"loss": 0.7357,
|
| 608 |
+
"step": 2000
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 0.21,
|
| 612 |
+
"learning_rate": 1.5736147757255937e-05,
|
| 613 |
+
"loss": 0.5729,
|
| 614 |
+
"step": 2020
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 0.22,
|
| 618 |
+
"learning_rate": 1.569393139841689e-05,
|
| 619 |
+
"loss": 0.4374,
|
| 620 |
+
"step": 2040
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 0.22,
|
| 624 |
+
"learning_rate": 1.5651715039577837e-05,
|
| 625 |
+
"loss": 0.5097,
|
| 626 |
+
"step": 2060
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 0.22,
|
| 630 |
+
"learning_rate": 1.560949868073879e-05,
|
| 631 |
+
"loss": 0.5712,
|
| 632 |
+
"step": 2080
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 0.22,
|
| 636 |
+
"learning_rate": 1.5567282321899737e-05,
|
| 637 |
+
"loss": 0.6074,
|
| 638 |
+
"step": 2100
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 0.22,
|
| 642 |
+
"learning_rate": 1.552506596306069e-05,
|
| 643 |
+
"loss": 0.3627,
|
| 644 |
+
"step": 2120
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 0.23,
|
| 648 |
+
"learning_rate": 1.5482849604221637e-05,
|
| 649 |
+
"loss": 0.448,
|
| 650 |
+
"step": 2140
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 0.23,
|
| 654 |
+
"learning_rate": 1.5440633245382588e-05,
|
| 655 |
+
"loss": 0.5785,
|
| 656 |
+
"step": 2160
|
| 657 |
+
},
|
| 658 |
+
{
|
| 659 |
+
"epoch": 0.23,
|
| 660 |
+
"learning_rate": 1.5398416886543536e-05,
|
| 661 |
+
"loss": 0.4907,
|
| 662 |
+
"step": 2180
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 0.23,
|
| 666 |
+
"learning_rate": 1.5356200527704484e-05,
|
| 667 |
+
"loss": 0.6272,
|
| 668 |
+
"step": 2200
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 0.23,
|
| 672 |
+
"learning_rate": 1.5313984168865436e-05,
|
| 673 |
+
"loss": 0.527,
|
| 674 |
+
"step": 2220
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 0.24,
|
| 678 |
+
"learning_rate": 1.5271767810026384e-05,
|
| 679 |
+
"loss": 0.6151,
|
| 680 |
+
"step": 2240
|
| 681 |
+
},
|
| 682 |
+
{
|
| 683 |
+
"epoch": 0.24,
|
| 684 |
+
"learning_rate": 1.5229551451187336e-05,
|
| 685 |
+
"loss": 0.37,
|
| 686 |
+
"step": 2260
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 0.24,
|
| 690 |
+
"learning_rate": 1.5187335092348286e-05,
|
| 691 |
+
"loss": 0.4636,
|
| 692 |
+
"step": 2280
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 0.24,
|
| 696 |
+
"learning_rate": 1.5145118733509237e-05,
|
| 697 |
+
"loss": 0.5352,
|
| 698 |
+
"step": 2300
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 0.24,
|
| 702 |
+
"learning_rate": 1.5102902374670185e-05,
|
| 703 |
+
"loss": 0.7749,
|
| 704 |
+
"step": 2320
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 0.25,
|
| 708 |
+
"learning_rate": 1.5060686015831137e-05,
|
| 709 |
+
"loss": 0.6404,
|
| 710 |
+
"step": 2340
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 0.25,
|
| 714 |
+
"learning_rate": 1.5018469656992085e-05,
|
| 715 |
+
"loss": 0.5735,
|
| 716 |
+
"step": 2360
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 0.25,
|
| 720 |
+
"learning_rate": 1.4976253298153037e-05,
|
| 721 |
+
"loss": 0.5506,
|
| 722 |
+
"step": 2380
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 0.25,
|
| 726 |
+
"learning_rate": 1.4934036939313985e-05,
|
| 727 |
+
"loss": 0.5916,
|
| 728 |
+
"step": 2400
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 0.26,
|
| 732 |
+
"learning_rate": 1.4891820580474936e-05,
|
| 733 |
+
"loss": 0.4593,
|
| 734 |
+
"step": 2420
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 0.26,
|
| 738 |
+
"learning_rate": 1.4849604221635885e-05,
|
| 739 |
+
"loss": 0.5974,
|
| 740 |
+
"step": 2440
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 0.26,
|
| 744 |
+
"learning_rate": 1.4807387862796836e-05,
|
| 745 |
+
"loss": 0.8557,
|
| 746 |
+
"step": 2460
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 0.26,
|
| 750 |
+
"learning_rate": 1.4765171503957784e-05,
|
| 751 |
+
"loss": 0.6341,
|
| 752 |
+
"step": 2480
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 0.26,
|
| 756 |
+
"learning_rate": 1.4722955145118736e-05,
|
| 757 |
+
"loss": 0.5251,
|
| 758 |
+
"step": 2500
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 0.27,
|
| 762 |
+
"learning_rate": 1.4680738786279684e-05,
|
| 763 |
+
"loss": 0.5603,
|
| 764 |
+
"step": 2520
|
| 765 |
+
},
|
| 766 |
+
{
|
| 767 |
+
"epoch": 0.27,
|
| 768 |
+
"learning_rate": 1.4638522427440634e-05,
|
| 769 |
+
"loss": 0.5832,
|
| 770 |
+
"step": 2540
|
| 771 |
+
},
|
| 772 |
+
{
|
| 773 |
+
"epoch": 0.27,
|
| 774 |
+
"learning_rate": 1.4596306068601584e-05,
|
| 775 |
+
"loss": 0.5408,
|
| 776 |
+
"step": 2560
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 0.27,
|
| 780 |
+
"learning_rate": 1.4554089709762534e-05,
|
| 781 |
+
"loss": 0.5024,
|
| 782 |
+
"step": 2580
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 0.27,
|
| 786 |
+
"learning_rate": 1.4511873350923484e-05,
|
| 787 |
+
"loss": 0.5729,
|
| 788 |
+
"step": 2600
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 0.28,
|
| 792 |
+
"learning_rate": 1.4469656992084433e-05,
|
| 793 |
+
"loss": 0.6356,
|
| 794 |
+
"step": 2620
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 0.28,
|
| 798 |
+
"learning_rate": 1.4427440633245385e-05,
|
| 799 |
+
"loss": 0.3953,
|
| 800 |
+
"step": 2640
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 0.28,
|
| 804 |
+
"learning_rate": 1.4385224274406333e-05,
|
| 805 |
+
"loss": 0.634,
|
| 806 |
+
"step": 2660
|
| 807 |
+
},
|
| 808 |
+
{
|
| 809 |
+
"epoch": 0.28,
|
| 810 |
+
"learning_rate": 1.4343007915567285e-05,
|
| 811 |
+
"loss": 0.4937,
|
| 812 |
+
"step": 2680
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 0.28,
|
| 816 |
+
"learning_rate": 1.4300791556728233e-05,
|
| 817 |
+
"loss": 0.5123,
|
| 818 |
+
"step": 2700
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 0.29,
|
| 822 |
+
"learning_rate": 1.4258575197889185e-05,
|
| 823 |
+
"loss": 0.5106,
|
| 824 |
+
"step": 2720
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 0.29,
|
| 828 |
+
"learning_rate": 1.4216358839050133e-05,
|
| 829 |
+
"loss": 0.634,
|
| 830 |
+
"step": 2740
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 0.29,
|
| 834 |
+
"learning_rate": 1.4174142480211084e-05,
|
| 835 |
+
"loss": 0.7666,
|
| 836 |
+
"step": 2760
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 0.29,
|
| 840 |
+
"learning_rate": 1.4131926121372032e-05,
|
| 841 |
+
"loss": 0.5169,
|
| 842 |
+
"step": 2780
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 0.3,
|
| 846 |
+
"learning_rate": 1.4089709762532984e-05,
|
| 847 |
+
"loss": 0.7099,
|
| 848 |
+
"step": 2800
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 0.3,
|
| 852 |
+
"learning_rate": 1.4047493403693932e-05,
|
| 853 |
+
"loss": 0.6799,
|
| 854 |
+
"step": 2820
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 0.3,
|
| 858 |
+
"learning_rate": 1.4005277044854882e-05,
|
| 859 |
+
"loss": 0.5833,
|
| 860 |
+
"step": 2840
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 0.3,
|
| 864 |
+
"learning_rate": 1.3963060686015832e-05,
|
| 865 |
+
"loss": 0.528,
|
| 866 |
+
"step": 2860
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 0.3,
|
| 870 |
+
"learning_rate": 1.3920844327176782e-05,
|
| 871 |
+
"loss": 0.8601,
|
| 872 |
+
"step": 2880
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 0.31,
|
| 876 |
+
"learning_rate": 1.3878627968337732e-05,
|
| 877 |
+
"loss": 0.6472,
|
| 878 |
+
"step": 2900
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 0.31,
|
| 882 |
+
"learning_rate": 1.3836411609498682e-05,
|
| 883 |
+
"loss": 0.6743,
|
| 884 |
+
"step": 2920
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 0.31,
|
| 888 |
+
"learning_rate": 1.3794195250659631e-05,
|
| 889 |
+
"loss": 0.5751,
|
| 890 |
+
"step": 2940
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 0.31,
|
| 894 |
+
"learning_rate": 1.3751978891820581e-05,
|
| 895 |
+
"loss": 0.5818,
|
| 896 |
+
"step": 2960
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 0.31,
|
| 900 |
+
"learning_rate": 1.3709762532981531e-05,
|
| 901 |
+
"loss": 0.6772,
|
| 902 |
+
"step": 2980
|
| 903 |
+
},
|
| 904 |
+
{
|
| 905 |
+
"epoch": 0.32,
|
| 906 |
+
"learning_rate": 1.3667546174142481e-05,
|
| 907 |
+
"loss": 0.6443,
|
| 908 |
+
"step": 3000
|
| 909 |
+
},
|
| 910 |
+
{
|
| 911 |
+
"epoch": 0.32,
|
| 912 |
+
"learning_rate": 1.3625329815303433e-05,
|
| 913 |
+
"loss": 0.7561,
|
| 914 |
+
"step": 3020
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"epoch": 0.32,
|
| 918 |
+
"learning_rate": 1.3583113456464381e-05,
|
| 919 |
+
"loss": 0.5368,
|
| 920 |
+
"step": 3040
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 0.32,
|
| 924 |
+
"learning_rate": 1.3540897097625332e-05,
|
| 925 |
+
"loss": 0.5932,
|
| 926 |
+
"step": 3060
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 0.33,
|
| 930 |
+
"learning_rate": 1.349868073878628e-05,
|
| 931 |
+
"loss": 0.646,
|
| 932 |
+
"step": 3080
|
| 933 |
+
},
|
| 934 |
+
{
|
| 935 |
+
"epoch": 0.33,
|
| 936 |
+
"learning_rate": 1.3456464379947232e-05,
|
| 937 |
+
"loss": 0.5333,
|
| 938 |
+
"step": 3100
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 0.33,
|
| 942 |
+
"learning_rate": 1.341424802110818e-05,
|
| 943 |
+
"loss": 0.5377,
|
| 944 |
+
"step": 3120
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 0.33,
|
| 948 |
+
"learning_rate": 1.3372031662269129e-05,
|
| 949 |
+
"loss": 0.3827,
|
| 950 |
+
"step": 3140
|
| 951 |
+
},
|
| 952 |
+
{
|
| 953 |
+
"epoch": 0.33,
|
| 954 |
+
"learning_rate": 1.332981530343008e-05,
|
| 955 |
+
"loss": 0.4472,
|
| 956 |
+
"step": 3160
|
| 957 |
+
},
|
| 958 |
+
{
|
| 959 |
+
"epoch": 0.34,
|
| 960 |
+
"learning_rate": 1.328759894459103e-05,
|
| 961 |
+
"loss": 0.6726,
|
| 962 |
+
"step": 3180
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 0.34,
|
| 966 |
+
"learning_rate": 1.324538258575198e-05,
|
| 967 |
+
"loss": 0.5824,
|
| 968 |
+
"step": 3200
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 0.34,
|
| 972 |
+
"learning_rate": 1.320316622691293e-05,
|
| 973 |
+
"loss": 0.6857,
|
| 974 |
+
"step": 3220
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 0.34,
|
| 978 |
+
"learning_rate": 1.316094986807388e-05,
|
| 979 |
+
"loss": 0.4948,
|
| 980 |
+
"step": 3240
|
| 981 |
+
},
|
| 982 |
+
{
|
| 983 |
+
"epoch": 0.34,
|
| 984 |
+
"learning_rate": 1.311873350923483e-05,
|
| 985 |
+
"loss": 0.4676,
|
| 986 |
+
"step": 3260
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 0.35,
|
| 990 |
+
"learning_rate": 1.307651715039578e-05,
|
| 991 |
+
"loss": 0.4412,
|
| 992 |
+
"step": 3280
|
| 993 |
+
},
|
| 994 |
+
{
|
| 995 |
+
"epoch": 0.35,
|
| 996 |
+
"learning_rate": 1.303430079155673e-05,
|
| 997 |
+
"loss": 0.5293,
|
| 998 |
+
"step": 3300
|
| 999 |
+
},
|
| 1000 |
+
{
|
| 1001 |
+
"epoch": 0.35,
|
| 1002 |
+
"learning_rate": 1.2992084432717679e-05,
|
| 1003 |
+
"loss": 0.6052,
|
| 1004 |
+
"step": 3320
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 0.35,
|
| 1008 |
+
"learning_rate": 1.2949868073878629e-05,
|
| 1009 |
+
"loss": 0.5418,
|
| 1010 |
+
"step": 3340
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 0.35,
|
| 1014 |
+
"learning_rate": 1.290765171503958e-05,
|
| 1015 |
+
"loss": 0.6428,
|
| 1016 |
+
"step": 3360
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"epoch": 0.36,
|
| 1020 |
+
"learning_rate": 1.2865435356200529e-05,
|
| 1021 |
+
"loss": 0.6619,
|
| 1022 |
+
"step": 3380
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 0.36,
|
| 1026 |
+
"learning_rate": 1.282321899736148e-05,
|
| 1027 |
+
"loss": 0.5869,
|
| 1028 |
+
"step": 3400
|
| 1029 |
+
},
|
| 1030 |
+
{
|
| 1031 |
+
"epoch": 0.36,
|
| 1032 |
+
"learning_rate": 1.2781002638522428e-05,
|
| 1033 |
+
"loss": 0.6139,
|
| 1034 |
+
"step": 3420
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 0.36,
|
| 1038 |
+
"learning_rate": 1.273878627968338e-05,
|
| 1039 |
+
"loss": 0.5202,
|
| 1040 |
+
"step": 3440
|
| 1041 |
+
},
|
| 1042 |
+
{
|
| 1043 |
+
"epoch": 0.37,
|
| 1044 |
+
"learning_rate": 1.2696569920844328e-05,
|
| 1045 |
+
"loss": 0.472,
|
| 1046 |
+
"step": 3460
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 0.37,
|
| 1050 |
+
"learning_rate": 1.2654353562005276e-05,
|
| 1051 |
+
"loss": 0.6936,
|
| 1052 |
+
"step": 3480
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 0.37,
|
| 1056 |
+
"learning_rate": 1.2612137203166228e-05,
|
| 1057 |
+
"loss": 0.5366,
|
| 1058 |
+
"step": 3500
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 0.37,
|
| 1062 |
+
"learning_rate": 1.2569920844327176e-05,
|
| 1063 |
+
"loss": 0.6306,
|
| 1064 |
+
"step": 3520
|
| 1065 |
+
},
|
| 1066 |
+
{
|
| 1067 |
+
"epoch": 0.37,
|
| 1068 |
+
"learning_rate": 1.2527704485488128e-05,
|
| 1069 |
+
"loss": 0.6323,
|
| 1070 |
+
"step": 3540
|
| 1071 |
+
},
|
| 1072 |
+
{
|
| 1073 |
+
"epoch": 0.38,
|
| 1074 |
+
"learning_rate": 1.2485488126649078e-05,
|
| 1075 |
+
"loss": 0.5825,
|
| 1076 |
+
"step": 3560
|
| 1077 |
+
},
|
| 1078 |
+
{
|
| 1079 |
+
"epoch": 0.38,
|
| 1080 |
+
"learning_rate": 1.2443271767810027e-05,
|
| 1081 |
+
"loss": 0.5203,
|
| 1082 |
+
"step": 3580
|
| 1083 |
+
},
|
| 1084 |
+
{
|
| 1085 |
+
"epoch": 0.38,
|
| 1086 |
+
"learning_rate": 1.2401055408970977e-05,
|
| 1087 |
+
"loss": 0.461,
|
| 1088 |
+
"step": 3600
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 0.38,
|
| 1092 |
+
"learning_rate": 1.2358839050131927e-05,
|
| 1093 |
+
"loss": 0.4318,
|
| 1094 |
+
"step": 3620
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 0.38,
|
| 1098 |
+
"learning_rate": 1.2316622691292877e-05,
|
| 1099 |
+
"loss": 0.6699,
|
| 1100 |
+
"step": 3640
|
| 1101 |
+
},
|
| 1102 |
+
{
|
| 1103 |
+
"epoch": 0.39,
|
| 1104 |
+
"learning_rate": 1.2274406332453827e-05,
|
| 1105 |
+
"loss": 0.4229,
|
| 1106 |
+
"step": 3660
|
| 1107 |
+
},
|
| 1108 |
+
{
|
| 1109 |
+
"epoch": 0.39,
|
| 1110 |
+
"learning_rate": 1.2232189973614777e-05,
|
| 1111 |
+
"loss": 0.4691,
|
| 1112 |
+
"step": 3680
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 0.39,
|
| 1116 |
+
"learning_rate": 1.2189973614775727e-05,
|
| 1117 |
+
"loss": 0.5152,
|
| 1118 |
+
"step": 3700
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 0.39,
|
| 1122 |
+
"learning_rate": 1.2147757255936677e-05,
|
| 1123 |
+
"loss": 0.4918,
|
| 1124 |
+
"step": 3720
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 0.39,
|
| 1128 |
+
"learning_rate": 1.2105540897097628e-05,
|
| 1129 |
+
"loss": 0.7893,
|
| 1130 |
+
"step": 3740
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 0.4,
|
| 1134 |
+
"learning_rate": 1.2063324538258576e-05,
|
| 1135 |
+
"loss": 0.6167,
|
| 1136 |
+
"step": 3760
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 0.4,
|
| 1140 |
+
"learning_rate": 1.2021108179419525e-05,
|
| 1141 |
+
"loss": 0.4979,
|
| 1142 |
+
"step": 3780
|
| 1143 |
+
},
|
| 1144 |
+
{
|
| 1145 |
+
"epoch": 0.4,
|
| 1146 |
+
"learning_rate": 1.1978891820580476e-05,
|
| 1147 |
+
"loss": 0.5982,
|
| 1148 |
+
"step": 3800
|
| 1149 |
+
},
|
| 1150 |
+
{
|
| 1151 |
+
"epoch": 0.4,
|
| 1152 |
+
"learning_rate": 1.1936675461741424e-05,
|
| 1153 |
+
"loss": 0.5404,
|
| 1154 |
+
"step": 3820
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 0.41,
|
| 1158 |
+
"learning_rate": 1.1894459102902376e-05,
|
| 1159 |
+
"loss": 0.5102,
|
| 1160 |
+
"step": 3840
|
| 1161 |
+
},
|
| 1162 |
+
{
|
| 1163 |
+
"epoch": 0.41,
|
| 1164 |
+
"learning_rate": 1.1852242744063324e-05,
|
| 1165 |
+
"loss": 0.7537,
|
| 1166 |
+
"step": 3860
|
| 1167 |
+
},
|
| 1168 |
+
{
|
| 1169 |
+
"epoch": 0.41,
|
| 1170 |
+
"learning_rate": 1.1810026385224276e-05,
|
| 1171 |
+
"loss": 0.5364,
|
| 1172 |
+
"step": 3880
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 0.41,
|
| 1176 |
+
"learning_rate": 1.1767810026385225e-05,
|
| 1177 |
+
"loss": 0.5699,
|
| 1178 |
+
"step": 3900
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 0.41,
|
| 1182 |
+
"learning_rate": 1.1725593667546175e-05,
|
| 1183 |
+
"loss": 0.6618,
|
| 1184 |
+
"step": 3920
|
| 1185 |
+
},
|
| 1186 |
+
{
|
| 1187 |
+
"epoch": 0.42,
|
| 1188 |
+
"learning_rate": 1.1683377308707125e-05,
|
| 1189 |
+
"loss": 0.457,
|
| 1190 |
+
"step": 3940
|
| 1191 |
+
},
|
| 1192 |
+
{
|
| 1193 |
+
"epoch": 0.42,
|
| 1194 |
+
"learning_rate": 1.1641160949868075e-05,
|
| 1195 |
+
"loss": 0.3489,
|
| 1196 |
+
"step": 3960
|
| 1197 |
+
},
|
| 1198 |
+
{
|
| 1199 |
+
"epoch": 0.42,
|
| 1200 |
+
"learning_rate": 1.1598944591029025e-05,
|
| 1201 |
+
"loss": 0.2943,
|
| 1202 |
+
"step": 3980
|
| 1203 |
+
},
|
| 1204 |
+
{
|
| 1205 |
+
"epoch": 0.42,
|
| 1206 |
+
"learning_rate": 1.1556728232189975e-05,
|
| 1207 |
+
"loss": 0.4931,
|
| 1208 |
+
"step": 4000
|
| 1209 |
+
},
|
| 1210 |
+
{
|
| 1211 |
+
"epoch": 0.42,
|
| 1212 |
+
"learning_rate": 1.1514511873350925e-05,
|
| 1213 |
+
"loss": 0.5655,
|
| 1214 |
+
"step": 4020
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 0.43,
|
| 1218 |
+
"learning_rate": 1.1472295514511875e-05,
|
| 1219 |
+
"loss": 0.514,
|
| 1220 |
+
"step": 4040
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 0.43,
|
| 1224 |
+
"learning_rate": 1.1430079155672824e-05,
|
| 1225 |
+
"loss": 0.3397,
|
| 1226 |
+
"step": 4060
|
| 1227 |
+
},
|
| 1228 |
+
{
|
| 1229 |
+
"epoch": 0.43,
|
| 1230 |
+
"learning_rate": 1.1387862796833773e-05,
|
| 1231 |
+
"loss": 0.524,
|
| 1232 |
+
"step": 4080
|
| 1233 |
+
},
|
| 1234 |
+
{
|
| 1235 |
+
"epoch": 0.43,
|
| 1236 |
+
"learning_rate": 1.1345646437994724e-05,
|
| 1237 |
+
"loss": 0.6505,
|
| 1238 |
+
"step": 4100
|
| 1239 |
+
},
|
| 1240 |
+
{
|
| 1241 |
+
"epoch": 0.43,
|
| 1242 |
+
"learning_rate": 1.1303430079155672e-05,
|
| 1243 |
+
"loss": 0.425,
|
| 1244 |
+
"step": 4120
|
| 1245 |
+
},
|
| 1246 |
+
{
|
| 1247 |
+
"epoch": 0.44,
|
| 1248 |
+
"learning_rate": 1.1261213720316624e-05,
|
| 1249 |
+
"loss": 0.62,
|
| 1250 |
+
"step": 4140
|
| 1251 |
+
},
|
| 1252 |
+
{
|
| 1253 |
+
"epoch": 0.44,
|
| 1254 |
+
"learning_rate": 1.1218997361477572e-05,
|
| 1255 |
+
"loss": 0.5354,
|
| 1256 |
+
"step": 4160
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 0.44,
|
| 1260 |
+
"learning_rate": 1.1176781002638524e-05,
|
| 1261 |
+
"loss": 0.6713,
|
| 1262 |
+
"step": 4180
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 0.44,
|
| 1266 |
+
"learning_rate": 1.1134564643799472e-05,
|
| 1267 |
+
"loss": 0.4955,
|
| 1268 |
+
"step": 4200
|
| 1269 |
+
},
|
| 1270 |
+
{
|
| 1271 |
+
"epoch": 0.45,
|
| 1272 |
+
"learning_rate": 1.1092348284960423e-05,
|
| 1273 |
+
"loss": 0.5697,
|
| 1274 |
+
"step": 4220
|
| 1275 |
+
},
|
| 1276 |
+
{
|
| 1277 |
+
"epoch": 0.45,
|
| 1278 |
+
"learning_rate": 1.1050131926121372e-05,
|
| 1279 |
+
"loss": 0.5844,
|
| 1280 |
+
"step": 4240
|
| 1281 |
+
},
|
| 1282 |
+
{
|
| 1283 |
+
"epoch": 0.45,
|
| 1284 |
+
"learning_rate": 1.1007915567282323e-05,
|
| 1285 |
+
"loss": 0.7378,
|
| 1286 |
+
"step": 4260
|
| 1287 |
+
},
|
| 1288 |
+
{
|
| 1289 |
+
"epoch": 0.45,
|
| 1290 |
+
"learning_rate": 1.0965699208443273e-05,
|
| 1291 |
+
"loss": 0.5467,
|
| 1292 |
+
"step": 4280
|
| 1293 |
+
},
|
| 1294 |
+
{
|
| 1295 |
+
"epoch": 0.45,
|
| 1296 |
+
"learning_rate": 1.0923482849604223e-05,
|
| 1297 |
+
"loss": 0.4497,
|
| 1298 |
+
"step": 4300
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 0.46,
|
| 1302 |
+
"learning_rate": 1.0881266490765173e-05,
|
| 1303 |
+
"loss": 0.6415,
|
| 1304 |
+
"step": 4320
|
| 1305 |
+
},
|
| 1306 |
+
{
|
| 1307 |
+
"epoch": 0.46,
|
| 1308 |
+
"learning_rate": 1.0839050131926123e-05,
|
| 1309 |
+
"loss": 0.3529,
|
| 1310 |
+
"step": 4340
|
| 1311 |
+
},
|
| 1312 |
+
{
|
| 1313 |
+
"epoch": 0.46,
|
| 1314 |
+
"learning_rate": 1.0796833773087073e-05,
|
| 1315 |
+
"loss": 0.5863,
|
| 1316 |
+
"step": 4360
|
| 1317 |
+
},
|
| 1318 |
+
{
|
| 1319 |
+
"epoch": 0.46,
|
| 1320 |
+
"learning_rate": 1.0754617414248022e-05,
|
| 1321 |
+
"loss": 0.6605,
|
| 1322 |
+
"step": 4380
|
| 1323 |
+
},
|
| 1324 |
+
{
|
| 1325 |
+
"epoch": 0.46,
|
| 1326 |
+
"learning_rate": 1.0712401055408972e-05,
|
| 1327 |
+
"loss": 0.5001,
|
| 1328 |
+
"step": 4400
|
| 1329 |
+
},
|
| 1330 |
+
{
|
| 1331 |
+
"epoch": 0.47,
|
| 1332 |
+
"learning_rate": 1.067018469656992e-05,
|
| 1333 |
+
"loss": 0.5911,
|
| 1334 |
+
"step": 4420
|
| 1335 |
+
},
|
| 1336 |
+
{
|
| 1337 |
+
"epoch": 0.47,
|
| 1338 |
+
"learning_rate": 1.0627968337730872e-05,
|
| 1339 |
+
"loss": 0.4878,
|
| 1340 |
+
"step": 4440
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 0.47,
|
| 1344 |
+
"learning_rate": 1.058575197889182e-05,
|
| 1345 |
+
"loss": 0.3502,
|
| 1346 |
+
"step": 4460
|
| 1347 |
+
},
|
| 1348 |
+
{
|
| 1349 |
+
"epoch": 0.47,
|
| 1350 |
+
"learning_rate": 1.0543535620052772e-05,
|
| 1351 |
+
"loss": 0.5124,
|
| 1352 |
+
"step": 4480
|
| 1353 |
+
},
|
| 1354 |
+
{
|
| 1355 |
+
"epoch": 0.47,
|
| 1356 |
+
"learning_rate": 1.050131926121372e-05,
|
| 1357 |
+
"loss": 0.6404,
|
| 1358 |
+
"step": 4500
|
| 1359 |
+
},
|
| 1360 |
+
{
|
| 1361 |
+
"epoch": 0.48,
|
| 1362 |
+
"learning_rate": 1.0459102902374672e-05,
|
| 1363 |
+
"loss": 0.5595,
|
| 1364 |
+
"step": 4520
|
| 1365 |
+
},
|
| 1366 |
+
{
|
| 1367 |
+
"epoch": 0.48,
|
| 1368 |
+
"learning_rate": 1.041688654353562e-05,
|
| 1369 |
+
"loss": 0.4514,
|
| 1370 |
+
"step": 4540
|
| 1371 |
+
},
|
| 1372 |
+
{
|
| 1373 |
+
"epoch": 0.48,
|
| 1374 |
+
"learning_rate": 1.0374670184696571e-05,
|
| 1375 |
+
"loss": 0.3171,
|
| 1376 |
+
"step": 4560
|
| 1377 |
+
},
|
| 1378 |
+
{
|
| 1379 |
+
"epoch": 0.48,
|
| 1380 |
+
"learning_rate": 1.033245382585752e-05,
|
| 1381 |
+
"loss": 0.1881,
|
| 1382 |
+
"step": 4580
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 0.49,
|
| 1386 |
+
"learning_rate": 1.0290237467018471e-05,
|
| 1387 |
+
"loss": 0.6278,
|
| 1388 |
+
"step": 4600
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 0.49,
|
| 1392 |
+
"learning_rate": 1.024802110817942e-05,
|
| 1393 |
+
"loss": 0.6495,
|
| 1394 |
+
"step": 4620
|
| 1395 |
+
},
|
| 1396 |
+
{
|
| 1397 |
+
"epoch": 0.49,
|
| 1398 |
+
"learning_rate": 1.020580474934037e-05,
|
| 1399 |
+
"loss": 0.2513,
|
| 1400 |
+
"step": 4640
|
| 1401 |
+
},
|
| 1402 |
+
{
|
| 1403 |
+
"epoch": 0.49,
|
| 1404 |
+
"learning_rate": 1.016358839050132e-05,
|
| 1405 |
+
"loss": 0.5532,
|
| 1406 |
+
"step": 4660
|
| 1407 |
+
},
|
| 1408 |
+
{
|
| 1409 |
+
"epoch": 0.49,
|
| 1410 |
+
"learning_rate": 1.012137203166227e-05,
|
| 1411 |
+
"loss": 0.5993,
|
| 1412 |
+
"step": 4680
|
| 1413 |
+
},
|
| 1414 |
+
{
|
| 1415 |
+
"epoch": 0.5,
|
| 1416 |
+
"learning_rate": 1.007915567282322e-05,
|
| 1417 |
+
"loss": 0.4503,
|
| 1418 |
+
"step": 4700
|
| 1419 |
+
},
|
| 1420 |
+
{
|
| 1421 |
+
"epoch": 0.5,
|
| 1422 |
+
"learning_rate": 1.0036939313984169e-05,
|
| 1423 |
+
"loss": 0.4031,
|
| 1424 |
+
"step": 4720
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 0.5,
|
| 1428 |
+
"learning_rate": 9.99472295514512e-06,
|
| 1429 |
+
"loss": 0.7682,
|
| 1430 |
+
"step": 4740
|
| 1431 |
+
},
|
| 1432 |
+
{
|
| 1433 |
+
"epoch": 0.5,
|
| 1434 |
+
"learning_rate": 9.95250659630607e-06,
|
| 1435 |
+
"loss": 0.7421,
|
| 1436 |
+
"step": 4760
|
| 1437 |
+
},
|
| 1438 |
+
{
|
| 1439 |
+
"epoch": 0.5,
|
| 1440 |
+
"learning_rate": 9.91029023746702e-06,
|
| 1441 |
+
"loss": 0.4206,
|
| 1442 |
+
"step": 4780
|
| 1443 |
+
},
|
| 1444 |
+
{
|
| 1445 |
+
"epoch": 0.51,
|
| 1446 |
+
"learning_rate": 9.86807387862797e-06,
|
| 1447 |
+
"loss": 0.4319,
|
| 1448 |
+
"step": 4800
|
| 1449 |
+
},
|
| 1450 |
+
{
|
| 1451 |
+
"epoch": 0.51,
|
| 1452 |
+
"learning_rate": 9.82585751978892e-06,
|
| 1453 |
+
"loss": 0.5304,
|
| 1454 |
+
"step": 4820
|
| 1455 |
+
},
|
| 1456 |
+
{
|
| 1457 |
+
"epoch": 0.51,
|
| 1458 |
+
"learning_rate": 9.78364116094987e-06,
|
| 1459 |
+
"loss": 0.6692,
|
| 1460 |
+
"step": 4840
|
| 1461 |
+
},
|
| 1462 |
+
{
|
| 1463 |
+
"epoch": 0.51,
|
| 1464 |
+
"learning_rate": 9.741424802110818e-06,
|
| 1465 |
+
"loss": 0.595,
|
| 1466 |
+
"step": 4860
|
| 1467 |
+
},
|
| 1468 |
+
{
|
| 1469 |
+
"epoch": 0.52,
|
| 1470 |
+
"learning_rate": 9.699208443271768e-06,
|
| 1471 |
+
"loss": 0.496,
|
| 1472 |
+
"step": 4880
|
| 1473 |
+
},
|
| 1474 |
+
{
|
| 1475 |
+
"epoch": 0.52,
|
| 1476 |
+
"learning_rate": 9.656992084432717e-06,
|
| 1477 |
+
"loss": 0.6153,
|
| 1478 |
+
"step": 4900
|
| 1479 |
+
},
|
| 1480 |
+
{
|
| 1481 |
+
"epoch": 0.52,
|
| 1482 |
+
"learning_rate": 9.614775725593667e-06,
|
| 1483 |
+
"loss": 0.3291,
|
| 1484 |
+
"step": 4920
|
| 1485 |
+
},
|
| 1486 |
+
{
|
| 1487 |
+
"epoch": 0.52,
|
| 1488 |
+
"learning_rate": 9.572559366754617e-06,
|
| 1489 |
+
"loss": 0.5457,
|
| 1490 |
+
"step": 4940
|
| 1491 |
+
},
|
| 1492 |
+
{
|
| 1493 |
+
"epoch": 0.52,
|
| 1494 |
+
"learning_rate": 9.530343007915567e-06,
|
| 1495 |
+
"loss": 0.5569,
|
| 1496 |
+
"step": 4960
|
| 1497 |
+
},
|
| 1498 |
+
{
|
| 1499 |
+
"epoch": 0.53,
|
| 1500 |
+
"learning_rate": 9.488126649076517e-06,
|
| 1501 |
+
"loss": 0.6754,
|
| 1502 |
+
"step": 4980
|
| 1503 |
+
},
|
| 1504 |
+
{
|
| 1505 |
+
"epoch": 0.53,
|
| 1506 |
+
"learning_rate": 9.445910290237469e-06,
|
| 1507 |
+
"loss": 0.5545,
|
| 1508 |
+
"step": 5000
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"epoch": 0.53,
|
| 1512 |
+
"learning_rate": 9.403693931398418e-06,
|
| 1513 |
+
"loss": 0.6471,
|
| 1514 |
+
"step": 5020
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 0.53,
|
| 1518 |
+
"learning_rate": 9.361477572559368e-06,
|
| 1519 |
+
"loss": 0.5456,
|
| 1520 |
+
"step": 5040
|
| 1521 |
+
},
|
| 1522 |
+
{
|
| 1523 |
+
"epoch": 0.53,
|
| 1524 |
+
"learning_rate": 9.319261213720318e-06,
|
| 1525 |
+
"loss": 0.617,
|
| 1526 |
+
"step": 5060
|
| 1527 |
+
},
|
| 1528 |
+
{
|
| 1529 |
+
"epoch": 0.54,
|
| 1530 |
+
"learning_rate": 9.277044854881268e-06,
|
| 1531 |
+
"loss": 0.4465,
|
| 1532 |
+
"step": 5080
|
| 1533 |
+
},
|
| 1534 |
+
{
|
| 1535 |
+
"epoch": 0.54,
|
| 1536 |
+
"learning_rate": 9.234828496042218e-06,
|
| 1537 |
+
"loss": 0.5831,
|
| 1538 |
+
"step": 5100
|
| 1539 |
+
},
|
| 1540 |
+
{
|
| 1541 |
+
"epoch": 0.54,
|
| 1542 |
+
"learning_rate": 9.192612137203168e-06,
|
| 1543 |
+
"loss": 0.465,
|
| 1544 |
+
"step": 5120
|
| 1545 |
+
},
|
| 1546 |
+
{
|
| 1547 |
+
"epoch": 0.54,
|
| 1548 |
+
"learning_rate": 9.150395778364118e-06,
|
| 1549 |
+
"loss": 0.5952,
|
| 1550 |
+
"step": 5140
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 0.54,
|
| 1554 |
+
"learning_rate": 9.108179419525068e-06,
|
| 1555 |
+
"loss": 0.5419,
|
| 1556 |
+
"step": 5160
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 0.55,
|
| 1560 |
+
"learning_rate": 9.065963060686016e-06,
|
| 1561 |
+
"loss": 0.4287,
|
| 1562 |
+
"step": 5180
|
| 1563 |
+
},
|
| 1564 |
+
{
|
| 1565 |
+
"epoch": 0.55,
|
| 1566 |
+
"learning_rate": 9.023746701846966e-06,
|
| 1567 |
+
"loss": 0.5648,
|
| 1568 |
+
"step": 5200
|
| 1569 |
+
},
|
| 1570 |
+
{
|
| 1571 |
+
"epoch": 0.55,
|
| 1572 |
+
"learning_rate": 8.981530343007915e-06,
|
| 1573 |
+
"loss": 0.6586,
|
| 1574 |
+
"step": 5220
|
| 1575 |
+
},
|
| 1576 |
+
{
|
| 1577 |
+
"epoch": 0.55,
|
| 1578 |
+
"learning_rate": 8.939313984168865e-06,
|
| 1579 |
+
"loss": 0.6909,
|
| 1580 |
+
"step": 5240
|
| 1581 |
+
},
|
| 1582 |
+
{
|
| 1583 |
+
"epoch": 0.56,
|
| 1584 |
+
"learning_rate": 8.897097625329815e-06,
|
| 1585 |
+
"loss": 0.7338,
|
| 1586 |
+
"step": 5260
|
| 1587 |
+
},
|
| 1588 |
+
{
|
| 1589 |
+
"epoch": 0.56,
|
| 1590 |
+
"learning_rate": 8.854881266490765e-06,
|
| 1591 |
+
"loss": 0.6468,
|
| 1592 |
+
"step": 5280
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 0.56,
|
| 1596 |
+
"learning_rate": 8.812664907651715e-06,
|
| 1597 |
+
"loss": 0.4928,
|
| 1598 |
+
"step": 5300
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 0.56,
|
| 1602 |
+
"learning_rate": 8.770448548812665e-06,
|
| 1603 |
+
"loss": 0.5157,
|
| 1604 |
+
"step": 5320
|
| 1605 |
+
},
|
| 1606 |
+
{
|
| 1607 |
+
"epoch": 0.56,
|
| 1608 |
+
"learning_rate": 8.728232189973615e-06,
|
| 1609 |
+
"loss": 0.4733,
|
| 1610 |
+
"step": 5340
|
| 1611 |
+
},
|
| 1612 |
+
{
|
| 1613 |
+
"epoch": 0.57,
|
| 1614 |
+
"learning_rate": 8.686015831134566e-06,
|
| 1615 |
+
"loss": 0.6383,
|
| 1616 |
+
"step": 5360
|
| 1617 |
+
},
|
| 1618 |
+
{
|
| 1619 |
+
"epoch": 0.57,
|
| 1620 |
+
"learning_rate": 8.643799472295516e-06,
|
| 1621 |
+
"loss": 0.4141,
|
| 1622 |
+
"step": 5380
|
| 1623 |
+
},
|
| 1624 |
+
{
|
| 1625 |
+
"epoch": 0.57,
|
| 1626 |
+
"learning_rate": 8.601583113456466e-06,
|
| 1627 |
+
"loss": 0.5483,
|
| 1628 |
+
"step": 5400
|
| 1629 |
+
},
|
| 1630 |
+
{
|
| 1631 |
+
"epoch": 0.57,
|
| 1632 |
+
"learning_rate": 8.559366754617416e-06,
|
| 1633 |
+
"loss": 0.4199,
|
| 1634 |
+
"step": 5420
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 0.57,
|
| 1638 |
+
"learning_rate": 8.517150395778366e-06,
|
| 1639 |
+
"loss": 0.5284,
|
| 1640 |
+
"step": 5440
|
| 1641 |
+
},
|
| 1642 |
+
{
|
| 1643 |
+
"epoch": 0.58,
|
| 1644 |
+
"learning_rate": 8.474934036939316e-06,
|
| 1645 |
+
"loss": 0.5165,
|
| 1646 |
+
"step": 5460
|
| 1647 |
+
},
|
| 1648 |
+
{
|
| 1649 |
+
"epoch": 0.58,
|
| 1650 |
+
"learning_rate": 8.432717678100266e-06,
|
| 1651 |
+
"loss": 0.5744,
|
| 1652 |
+
"step": 5480
|
| 1653 |
+
},
|
| 1654 |
+
{
|
| 1655 |
+
"epoch": 0.58,
|
| 1656 |
+
"learning_rate": 8.390501319261214e-06,
|
| 1657 |
+
"loss": 0.5308,
|
| 1658 |
+
"step": 5500
|
| 1659 |
+
},
|
| 1660 |
+
{
|
| 1661 |
+
"epoch": 0.58,
|
| 1662 |
+
"learning_rate": 8.348284960422164e-06,
|
| 1663 |
+
"loss": 0.5786,
|
| 1664 |
+
"step": 5520
|
| 1665 |
+
},
|
| 1666 |
+
{
|
| 1667 |
+
"epoch": 0.58,
|
| 1668 |
+
"learning_rate": 8.306068601583113e-06,
|
| 1669 |
+
"loss": 0.5103,
|
| 1670 |
+
"step": 5540
|
| 1671 |
+
},
|
| 1672 |
+
{
|
| 1673 |
+
"epoch": 0.59,
|
| 1674 |
+
"learning_rate": 8.263852242744063e-06,
|
| 1675 |
+
"loss": 0.3575,
|
| 1676 |
+
"step": 5560
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"epoch": 0.59,
|
| 1680 |
+
"learning_rate": 8.221635883905013e-06,
|
| 1681 |
+
"loss": 0.6229,
|
| 1682 |
+
"step": 5580
|
| 1683 |
+
},
|
| 1684 |
+
{
|
| 1685 |
+
"epoch": 0.59,
|
| 1686 |
+
"learning_rate": 8.179419525065963e-06,
|
| 1687 |
+
"loss": 0.6278,
|
| 1688 |
+
"step": 5600
|
| 1689 |
+
},
|
| 1690 |
+
{
|
| 1691 |
+
"epoch": 0.59,
|
| 1692 |
+
"learning_rate": 8.137203166226913e-06,
|
| 1693 |
+
"loss": 0.714,
|
| 1694 |
+
"step": 5620
|
| 1695 |
+
},
|
| 1696 |
+
{
|
| 1697 |
+
"epoch": 0.6,
|
| 1698 |
+
"learning_rate": 8.094986807387863e-06,
|
| 1699 |
+
"loss": 0.425,
|
| 1700 |
+
"step": 5640
|
| 1701 |
+
},
|
| 1702 |
+
{
|
| 1703 |
+
"epoch": 0.6,
|
| 1704 |
+
"learning_rate": 8.052770448548813e-06,
|
| 1705 |
+
"loss": 0.4628,
|
| 1706 |
+
"step": 5660
|
| 1707 |
+
},
|
| 1708 |
+
{
|
| 1709 |
+
"epoch": 0.6,
|
| 1710 |
+
"learning_rate": 8.010554089709763e-06,
|
| 1711 |
+
"loss": 0.5427,
|
| 1712 |
+
"step": 5680
|
| 1713 |
+
},
|
| 1714 |
+
{
|
| 1715 |
+
"epoch": 0.6,
|
| 1716 |
+
"learning_rate": 7.968337730870712e-06,
|
| 1717 |
+
"loss": 0.5701,
|
| 1718 |
+
"step": 5700
|
| 1719 |
+
},
|
| 1720 |
+
{
|
| 1721 |
+
"epoch": 0.6,
|
| 1722 |
+
"learning_rate": 7.926121372031664e-06,
|
| 1723 |
+
"loss": 0.4103,
|
| 1724 |
+
"step": 5720
|
| 1725 |
+
},
|
| 1726 |
+
{
|
| 1727 |
+
"epoch": 0.61,
|
| 1728 |
+
"learning_rate": 7.883905013192614e-06,
|
| 1729 |
+
"loss": 0.3837,
|
| 1730 |
+
"step": 5740
|
| 1731 |
+
},
|
| 1732 |
+
{
|
| 1733 |
+
"epoch": 0.61,
|
| 1734 |
+
"learning_rate": 7.841688654353564e-06,
|
| 1735 |
+
"loss": 0.6341,
|
| 1736 |
+
"step": 5760
|
| 1737 |
+
},
|
| 1738 |
+
{
|
| 1739 |
+
"epoch": 0.61,
|
| 1740 |
+
"learning_rate": 7.799472295514514e-06,
|
| 1741 |
+
"loss": 0.3624,
|
| 1742 |
+
"step": 5780
|
| 1743 |
+
},
|
| 1744 |
+
{
|
| 1745 |
+
"epoch": 0.61,
|
| 1746 |
+
"learning_rate": 7.757255936675462e-06,
|
| 1747 |
+
"loss": 0.4841,
|
| 1748 |
+
"step": 5800
|
| 1749 |
+
},
|
| 1750 |
+
{
|
| 1751 |
+
"epoch": 0.61,
|
| 1752 |
+
"learning_rate": 7.715039577836412e-06,
|
| 1753 |
+
"loss": 0.4588,
|
| 1754 |
+
"step": 5820
|
| 1755 |
+
},
|
| 1756 |
+
{
|
| 1757 |
+
"epoch": 0.62,
|
| 1758 |
+
"learning_rate": 7.672823218997362e-06,
|
| 1759 |
+
"loss": 0.3238,
|
| 1760 |
+
"step": 5840
|
| 1761 |
+
},
|
| 1762 |
+
{
|
| 1763 |
+
"epoch": 0.62,
|
| 1764 |
+
"learning_rate": 7.630606860158311e-06,
|
| 1765 |
+
"loss": 0.4156,
|
| 1766 |
+
"step": 5860
|
| 1767 |
+
},
|
| 1768 |
+
{
|
| 1769 |
+
"epoch": 0.62,
|
| 1770 |
+
"learning_rate": 7.588390501319261e-06,
|
| 1771 |
+
"loss": 0.556,
|
| 1772 |
+
"step": 5880
|
| 1773 |
+
},
|
| 1774 |
+
{
|
| 1775 |
+
"epoch": 0.62,
|
| 1776 |
+
"learning_rate": 7.546174142480211e-06,
|
| 1777 |
+
"loss": 0.3624,
|
| 1778 |
+
"step": 5900
|
| 1779 |
+
},
|
| 1780 |
+
{
|
| 1781 |
+
"epoch": 0.62,
|
| 1782 |
+
"learning_rate": 7.503957783641161e-06,
|
| 1783 |
+
"loss": 0.6702,
|
| 1784 |
+
"step": 5920
|
| 1785 |
+
},
|
| 1786 |
+
{
|
| 1787 |
+
"epoch": 0.63,
|
| 1788 |
+
"learning_rate": 7.461741424802112e-06,
|
| 1789 |
+
"loss": 0.6105,
|
| 1790 |
+
"step": 5940
|
| 1791 |
+
},
|
| 1792 |
+
{
|
| 1793 |
+
"epoch": 0.63,
|
| 1794 |
+
"learning_rate": 7.419525065963062e-06,
|
| 1795 |
+
"loss": 0.3351,
|
| 1796 |
+
"step": 5960
|
| 1797 |
+
},
|
| 1798 |
+
{
|
| 1799 |
+
"epoch": 0.63,
|
| 1800 |
+
"learning_rate": 7.3773087071240116e-06,
|
| 1801 |
+
"loss": 0.5021,
|
| 1802 |
+
"step": 5980
|
| 1803 |
+
},
|
| 1804 |
+
{
|
| 1805 |
+
"epoch": 0.63,
|
| 1806 |
+
"learning_rate": 7.3350923482849614e-06,
|
| 1807 |
+
"loss": 0.6574,
|
| 1808 |
+
"step": 6000
|
| 1809 |
+
},
|
| 1810 |
+
{
|
| 1811 |
+
"epoch": 0.64,
|
| 1812 |
+
"learning_rate": 7.292875989445911e-06,
|
| 1813 |
+
"loss": 0.3253,
|
| 1814 |
+
"step": 6020
|
| 1815 |
+
},
|
| 1816 |
+
{
|
| 1817 |
+
"epoch": 0.64,
|
| 1818 |
+
"learning_rate": 7.250659630606861e-06,
|
| 1819 |
+
"loss": 0.6144,
|
| 1820 |
+
"step": 6040
|
| 1821 |
+
},
|
| 1822 |
+
{
|
| 1823 |
+
"epoch": 0.64,
|
| 1824 |
+
"learning_rate": 7.208443271767811e-06,
|
| 1825 |
+
"loss": 0.6338,
|
| 1826 |
+
"step": 6060
|
| 1827 |
+
},
|
| 1828 |
+
{
|
| 1829 |
+
"epoch": 0.64,
|
| 1830 |
+
"learning_rate": 7.166226912928761e-06,
|
| 1831 |
+
"loss": 0.5613,
|
| 1832 |
+
"step": 6080
|
| 1833 |
+
},
|
| 1834 |
+
{
|
| 1835 |
+
"epoch": 0.64,
|
| 1836 |
+
"learning_rate": 7.124010554089711e-06,
|
| 1837 |
+
"loss": 0.4417,
|
| 1838 |
+
"step": 6100
|
| 1839 |
+
},
|
| 1840 |
+
{
|
| 1841 |
+
"epoch": 0.65,
|
| 1842 |
+
"learning_rate": 7.08179419525066e-06,
|
| 1843 |
+
"loss": 0.6518,
|
| 1844 |
+
"step": 6120
|
| 1845 |
+
},
|
| 1846 |
+
{
|
| 1847 |
+
"epoch": 0.65,
|
| 1848 |
+
"learning_rate": 7.03957783641161e-06,
|
| 1849 |
+
"loss": 0.4236,
|
| 1850 |
+
"step": 6140
|
| 1851 |
+
},
|
| 1852 |
+
{
|
| 1853 |
+
"epoch": 0.65,
|
| 1854 |
+
"learning_rate": 6.99736147757256e-06,
|
| 1855 |
+
"loss": 0.7198,
|
| 1856 |
+
"step": 6160
|
| 1857 |
+
},
|
| 1858 |
+
{
|
| 1859 |
+
"epoch": 0.65,
|
| 1860 |
+
"learning_rate": 6.9551451187335095e-06,
|
| 1861 |
+
"loss": 0.5412,
|
| 1862 |
+
"step": 6180
|
| 1863 |
+
},
|
| 1864 |
+
{
|
| 1865 |
+
"epoch": 0.65,
|
| 1866 |
+
"learning_rate": 6.912928759894459e-06,
|
| 1867 |
+
"loss": 0.4579,
|
| 1868 |
+
"step": 6200
|
| 1869 |
+
},
|
| 1870 |
+
{
|
| 1871 |
+
"epoch": 0.66,
|
| 1872 |
+
"learning_rate": 6.870712401055409e-06,
|
| 1873 |
+
"loss": 0.5445,
|
| 1874 |
+
"step": 6220
|
| 1875 |
+
},
|
| 1876 |
+
{
|
| 1877 |
+
"epoch": 0.66,
|
| 1878 |
+
"learning_rate": 6.828496042216359e-06,
|
| 1879 |
+
"loss": 0.3513,
|
| 1880 |
+
"step": 6240
|
| 1881 |
+
},
|
| 1882 |
+
{
|
| 1883 |
+
"epoch": 0.66,
|
| 1884 |
+
"learning_rate": 6.786279683377309e-06,
|
| 1885 |
+
"loss": 0.4794,
|
| 1886 |
+
"step": 6260
|
| 1887 |
+
},
|
| 1888 |
+
{
|
| 1889 |
+
"epoch": 0.66,
|
| 1890 |
+
"learning_rate": 6.744063324538259e-06,
|
| 1891 |
+
"loss": 0.48,
|
| 1892 |
+
"step": 6280
|
| 1893 |
+
},
|
| 1894 |
+
{
|
| 1895 |
+
"epoch": 0.66,
|
| 1896 |
+
"learning_rate": 6.701846965699209e-06,
|
| 1897 |
+
"loss": 0.6672,
|
| 1898 |
+
"step": 6300
|
| 1899 |
+
},
|
| 1900 |
+
{
|
| 1901 |
+
"epoch": 0.67,
|
| 1902 |
+
"learning_rate": 6.659630606860159e-06,
|
| 1903 |
+
"loss": 0.3798,
|
| 1904 |
+
"step": 6320
|
| 1905 |
+
},
|
| 1906 |
+
{
|
| 1907 |
+
"epoch": 0.67,
|
| 1908 |
+
"learning_rate": 6.617414248021109e-06,
|
| 1909 |
+
"loss": 0.3502,
|
| 1910 |
+
"step": 6340
|
| 1911 |
+
},
|
| 1912 |
+
{
|
| 1913 |
+
"epoch": 0.67,
|
| 1914 |
+
"learning_rate": 6.575197889182059e-06,
|
| 1915 |
+
"loss": 0.5835,
|
| 1916 |
+
"step": 6360
|
| 1917 |
+
},
|
| 1918 |
+
{
|
| 1919 |
+
"epoch": 0.67,
|
| 1920 |
+
"learning_rate": 6.532981530343009e-06,
|
| 1921 |
+
"loss": 0.5271,
|
| 1922 |
+
"step": 6380
|
| 1923 |
+
},
|
| 1924 |
+
{
|
| 1925 |
+
"epoch": 0.68,
|
| 1926 |
+
"learning_rate": 6.490765171503959e-06,
|
| 1927 |
+
"loss": 0.4799,
|
| 1928 |
+
"step": 6400
|
| 1929 |
+
},
|
| 1930 |
+
{
|
| 1931 |
+
"epoch": 0.68,
|
| 1932 |
+
"learning_rate": 6.448548812664909e-06,
|
| 1933 |
+
"loss": 0.5692,
|
| 1934 |
+
"step": 6420
|
| 1935 |
+
},
|
| 1936 |
+
{
|
| 1937 |
+
"epoch": 0.68,
|
| 1938 |
+
"learning_rate": 6.406332453825858e-06,
|
| 1939 |
+
"loss": 0.5402,
|
| 1940 |
+
"step": 6440
|
| 1941 |
+
},
|
| 1942 |
+
{
|
| 1943 |
+
"epoch": 0.68,
|
| 1944 |
+
"learning_rate": 6.364116094986808e-06,
|
| 1945 |
+
"loss": 0.4828,
|
| 1946 |
+
"step": 6460
|
| 1947 |
+
},
|
| 1948 |
+
{
|
| 1949 |
+
"epoch": 0.68,
|
| 1950 |
+
"learning_rate": 6.3218997361477576e-06,
|
| 1951 |
+
"loss": 0.515,
|
| 1952 |
+
"step": 6480
|
| 1953 |
+
},
|
| 1954 |
+
{
|
| 1955 |
+
"epoch": 0.69,
|
| 1956 |
+
"learning_rate": 6.2796833773087074e-06,
|
| 1957 |
+
"loss": 0.5592,
|
| 1958 |
+
"step": 6500
|
| 1959 |
+
},
|
| 1960 |
+
{
|
| 1961 |
+
"epoch": 0.69,
|
| 1962 |
+
"learning_rate": 6.237467018469657e-06,
|
| 1963 |
+
"loss": 0.6036,
|
| 1964 |
+
"step": 6520
|
| 1965 |
+
},
|
| 1966 |
+
{
|
| 1967 |
+
"epoch": 0.69,
|
| 1968 |
+
"learning_rate": 6.195250659630607e-06,
|
| 1969 |
+
"loss": 0.6602,
|
| 1970 |
+
"step": 6540
|
| 1971 |
+
},
|
| 1972 |
+
{
|
| 1973 |
+
"epoch": 0.69,
|
| 1974 |
+
"learning_rate": 6.153034300791557e-06,
|
| 1975 |
+
"loss": 0.6557,
|
| 1976 |
+
"step": 6560
|
| 1977 |
+
},
|
| 1978 |
+
{
|
| 1979 |
+
"epoch": 0.69,
|
| 1980 |
+
"learning_rate": 6.110817941952507e-06,
|
| 1981 |
+
"loss": 0.6216,
|
| 1982 |
+
"step": 6580
|
| 1983 |
+
},
|
| 1984 |
+
{
|
| 1985 |
+
"epoch": 0.7,
|
| 1986 |
+
"learning_rate": 6.068601583113457e-06,
|
| 1987 |
+
"loss": 0.4562,
|
| 1988 |
+
"step": 6600
|
| 1989 |
+
},
|
| 1990 |
+
{
|
| 1991 |
+
"epoch": 0.7,
|
| 1992 |
+
"learning_rate": 6.026385224274407e-06,
|
| 1993 |
+
"loss": 0.6253,
|
| 1994 |
+
"step": 6620
|
| 1995 |
+
},
|
| 1996 |
+
{
|
| 1997 |
+
"epoch": 0.7,
|
| 1998 |
+
"learning_rate": 5.9841688654353566e-06,
|
| 1999 |
+
"loss": 0.4678,
|
| 2000 |
+
"step": 6640
|
| 2001 |
+
},
|
| 2002 |
+
{
|
| 2003 |
+
"epoch": 0.7,
|
| 2004 |
+
"learning_rate": 5.9419525065963064e-06,
|
| 2005 |
+
"loss": 0.7542,
|
| 2006 |
+
"step": 6660
|
| 2007 |
+
},
|
| 2008 |
+
{
|
| 2009 |
+
"epoch": 0.71,
|
| 2010 |
+
"learning_rate": 5.899736147757257e-06,
|
| 2011 |
+
"loss": 0.5984,
|
| 2012 |
+
"step": 6680
|
| 2013 |
+
},
|
| 2014 |
+
{
|
| 2015 |
+
"epoch": 0.71,
|
| 2016 |
+
"learning_rate": 5.857519788918207e-06,
|
| 2017 |
+
"loss": 0.4195,
|
| 2018 |
+
"step": 6700
|
| 2019 |
+
},
|
| 2020 |
+
{
|
| 2021 |
+
"epoch": 0.71,
|
| 2022 |
+
"learning_rate": 5.815303430079157e-06,
|
| 2023 |
+
"loss": 0.6876,
|
| 2024 |
+
"step": 6720
|
| 2025 |
+
},
|
| 2026 |
+
{
|
| 2027 |
+
"epoch": 0.71,
|
| 2028 |
+
"learning_rate": 5.773087071240105e-06,
|
| 2029 |
+
"loss": 0.486,
|
| 2030 |
+
"step": 6740
|
| 2031 |
+
},
|
| 2032 |
+
{
|
| 2033 |
+
"epoch": 0.71,
|
| 2034 |
+
"learning_rate": 5.730870712401056e-06,
|
| 2035 |
+
"loss": 0.706,
|
| 2036 |
+
"step": 6760
|
| 2037 |
+
},
|
| 2038 |
+
{
|
| 2039 |
+
"epoch": 0.72,
|
| 2040 |
+
"learning_rate": 5.688654353562006e-06,
|
| 2041 |
+
"loss": 0.5317,
|
| 2042 |
+
"step": 6780
|
| 2043 |
+
},
|
| 2044 |
+
{
|
| 2045 |
+
"epoch": 0.72,
|
| 2046 |
+
"learning_rate": 5.6464379947229556e-06,
|
| 2047 |
+
"loss": 0.4366,
|
| 2048 |
+
"step": 6800
|
| 2049 |
+
},
|
| 2050 |
+
{
|
| 2051 |
+
"epoch": 0.72,
|
| 2052 |
+
"learning_rate": 5.6042216358839054e-06,
|
| 2053 |
+
"loss": 0.3125,
|
| 2054 |
+
"step": 6820
|
| 2055 |
+
},
|
| 2056 |
+
{
|
| 2057 |
+
"epoch": 0.72,
|
| 2058 |
+
"learning_rate": 5.562005277044855e-06,
|
| 2059 |
+
"loss": 0.4181,
|
| 2060 |
+
"step": 6840
|
| 2061 |
+
},
|
| 2062 |
+
{
|
| 2063 |
+
"epoch": 0.72,
|
| 2064 |
+
"learning_rate": 5.519788918205805e-06,
|
| 2065 |
+
"loss": 0.6685,
|
| 2066 |
+
"step": 6860
|
| 2067 |
+
},
|
| 2068 |
+
{
|
| 2069 |
+
"epoch": 0.73,
|
| 2070 |
+
"learning_rate": 5.477572559366755e-06,
|
| 2071 |
+
"loss": 0.4767,
|
| 2072 |
+
"step": 6880
|
| 2073 |
+
},
|
| 2074 |
+
{
|
| 2075 |
+
"epoch": 0.73,
|
| 2076 |
+
"learning_rate": 5.435356200527705e-06,
|
| 2077 |
+
"loss": 0.5124,
|
| 2078 |
+
"step": 6900
|
| 2079 |
+
},
|
| 2080 |
+
{
|
| 2081 |
+
"epoch": 0.73,
|
| 2082 |
+
"learning_rate": 5.393139841688655e-06,
|
| 2083 |
+
"loss": 0.3873,
|
| 2084 |
+
"step": 6920
|
| 2085 |
+
},
|
| 2086 |
+
{
|
| 2087 |
+
"epoch": 0.73,
|
| 2088 |
+
"learning_rate": 5.350923482849605e-06,
|
| 2089 |
+
"loss": 0.6609,
|
| 2090 |
+
"step": 6940
|
| 2091 |
+
},
|
| 2092 |
+
{
|
| 2093 |
+
"epoch": 0.73,
|
| 2094 |
+
"learning_rate": 5.3087071240105546e-06,
|
| 2095 |
+
"loss": 0.5962,
|
| 2096 |
+
"step": 6960
|
| 2097 |
+
},
|
| 2098 |
+
{
|
| 2099 |
+
"epoch": 0.74,
|
| 2100 |
+
"learning_rate": 5.2664907651715044e-06,
|
| 2101 |
+
"loss": 0.4593,
|
| 2102 |
+
"step": 6980
|
| 2103 |
+
},
|
| 2104 |
+
{
|
| 2105 |
+
"epoch": 0.74,
|
| 2106 |
+
"learning_rate": 5.224274406332454e-06,
|
| 2107 |
+
"loss": 0.3752,
|
| 2108 |
+
"step": 7000
|
| 2109 |
+
},
|
| 2110 |
+
{
|
| 2111 |
+
"epoch": 0.74,
|
| 2112 |
+
"learning_rate": 5.182058047493404e-06,
|
| 2113 |
+
"loss": 0.6012,
|
| 2114 |
+
"step": 7020
|
| 2115 |
+
},
|
| 2116 |
+
{
|
| 2117 |
+
"epoch": 0.74,
|
| 2118 |
+
"learning_rate": 5.139841688654355e-06,
|
| 2119 |
+
"loss": 0.3253,
|
| 2120 |
+
"step": 7040
|
| 2121 |
+
},
|
| 2122 |
+
{
|
| 2123 |
+
"epoch": 0.75,
|
| 2124 |
+
"learning_rate": 5.097625329815303e-06,
|
| 2125 |
+
"loss": 0.2852,
|
| 2126 |
+
"step": 7060
|
| 2127 |
+
},
|
| 2128 |
+
{
|
| 2129 |
+
"epoch": 0.75,
|
| 2130 |
+
"learning_rate": 5.055408970976253e-06,
|
| 2131 |
+
"loss": 0.4875,
|
| 2132 |
+
"step": 7080
|
| 2133 |
+
},
|
| 2134 |
+
{
|
| 2135 |
+
"epoch": 0.75,
|
| 2136 |
+
"learning_rate": 5.013192612137203e-06,
|
| 2137 |
+
"loss": 0.6634,
|
| 2138 |
+
"step": 7100
|
| 2139 |
+
},
|
| 2140 |
+
{
|
| 2141 |
+
"epoch": 0.75,
|
| 2142 |
+
"learning_rate": 4.9709762532981536e-06,
|
| 2143 |
+
"loss": 0.5397,
|
| 2144 |
+
"step": 7120
|
| 2145 |
+
},
|
| 2146 |
+
{
|
| 2147 |
+
"epoch": 0.75,
|
| 2148 |
+
"learning_rate": 4.9287598944591034e-06,
|
| 2149 |
+
"loss": 0.4543,
|
| 2150 |
+
"step": 7140
|
| 2151 |
+
},
|
| 2152 |
+
{
|
| 2153 |
+
"epoch": 0.76,
|
| 2154 |
+
"learning_rate": 4.886543535620053e-06,
|
| 2155 |
+
"loss": 0.4,
|
| 2156 |
+
"step": 7160
|
| 2157 |
+
},
|
| 2158 |
+
{
|
| 2159 |
+
"epoch": 0.76,
|
| 2160 |
+
"learning_rate": 4.844327176781003e-06,
|
| 2161 |
+
"loss": 0.6291,
|
| 2162 |
+
"step": 7180
|
| 2163 |
+
},
|
| 2164 |
+
{
|
| 2165 |
+
"epoch": 0.76,
|
| 2166 |
+
"learning_rate": 4.802110817941953e-06,
|
| 2167 |
+
"loss": 0.5296,
|
| 2168 |
+
"step": 7200
|
| 2169 |
+
},
|
| 2170 |
+
{
|
| 2171 |
+
"epoch": 0.76,
|
| 2172 |
+
"learning_rate": 4.759894459102903e-06,
|
| 2173 |
+
"loss": 0.6181,
|
| 2174 |
+
"step": 7220
|
| 2175 |
+
},
|
| 2176 |
+
{
|
| 2177 |
+
"epoch": 0.76,
|
| 2178 |
+
"learning_rate": 4.717678100263853e-06,
|
| 2179 |
+
"loss": 0.4267,
|
| 2180 |
+
"step": 7240
|
| 2181 |
+
},
|
| 2182 |
+
{
|
| 2183 |
+
"epoch": 0.77,
|
| 2184 |
+
"learning_rate": 4.675461741424803e-06,
|
| 2185 |
+
"loss": 0.55,
|
| 2186 |
+
"step": 7260
|
| 2187 |
+
},
|
| 2188 |
+
{
|
| 2189 |
+
"epoch": 0.77,
|
| 2190 |
+
"learning_rate": 4.6332453825857525e-06,
|
| 2191 |
+
"loss": 0.5503,
|
| 2192 |
+
"step": 7280
|
| 2193 |
+
},
|
| 2194 |
+
{
|
| 2195 |
+
"epoch": 0.77,
|
| 2196 |
+
"learning_rate": 4.5910290237467024e-06,
|
| 2197 |
+
"loss": 0.65,
|
| 2198 |
+
"step": 7300
|
| 2199 |
+
},
|
| 2200 |
+
{
|
| 2201 |
+
"epoch": 0.77,
|
| 2202 |
+
"learning_rate": 4.548812664907652e-06,
|
| 2203 |
+
"loss": 0.5358,
|
| 2204 |
+
"step": 7320
|
| 2205 |
+
},
|
| 2206 |
+
{
|
| 2207 |
+
"epoch": 0.77,
|
| 2208 |
+
"learning_rate": 4.506596306068602e-06,
|
| 2209 |
+
"loss": 0.5237,
|
| 2210 |
+
"step": 7340
|
| 2211 |
+
},
|
| 2212 |
+
{
|
| 2213 |
+
"epoch": 0.78,
|
| 2214 |
+
"learning_rate": 4.464379947229552e-06,
|
| 2215 |
+
"loss": 0.7353,
|
| 2216 |
+
"step": 7360
|
| 2217 |
+
},
|
| 2218 |
+
{
|
| 2219 |
+
"epoch": 0.78,
|
| 2220 |
+
"learning_rate": 4.422163588390502e-06,
|
| 2221 |
+
"loss": 0.454,
|
| 2222 |
+
"step": 7380
|
| 2223 |
+
},
|
| 2224 |
+
{
|
| 2225 |
+
"epoch": 0.78,
|
| 2226 |
+
"learning_rate": 4.379947229551452e-06,
|
| 2227 |
+
"loss": 0.508,
|
| 2228 |
+
"step": 7400
|
| 2229 |
+
},
|
| 2230 |
+
{
|
| 2231 |
+
"epoch": 0.78,
|
| 2232 |
+
"learning_rate": 4.337730870712402e-06,
|
| 2233 |
+
"loss": 0.6651,
|
| 2234 |
+
"step": 7420
|
| 2235 |
+
},
|
| 2236 |
+
{
|
| 2237 |
+
"epoch": 0.79,
|
| 2238 |
+
"learning_rate": 4.295514511873351e-06,
|
| 2239 |
+
"loss": 0.449,
|
| 2240 |
+
"step": 7440
|
| 2241 |
+
},
|
| 2242 |
+
{
|
| 2243 |
+
"epoch": 0.79,
|
| 2244 |
+
"learning_rate": 4.2532981530343006e-06,
|
| 2245 |
+
"loss": 0.6682,
|
| 2246 |
+
"step": 7460
|
| 2247 |
+
},
|
| 2248 |
+
{
|
| 2249 |
+
"epoch": 0.79,
|
| 2250 |
+
"learning_rate": 4.211081794195251e-06,
|
| 2251 |
+
"loss": 0.4796,
|
| 2252 |
+
"step": 7480
|
| 2253 |
+
},
|
| 2254 |
+
{
|
| 2255 |
+
"epoch": 0.79,
|
| 2256 |
+
"learning_rate": 4.168865435356201e-06,
|
| 2257 |
+
"loss": 0.4207,
|
| 2258 |
+
"step": 7500
|
| 2259 |
+
},
|
| 2260 |
+
{
|
| 2261 |
+
"epoch": 0.79,
|
| 2262 |
+
"learning_rate": 4.126649076517151e-06,
|
| 2263 |
+
"loss": 0.4697,
|
| 2264 |
+
"step": 7520
|
| 2265 |
+
},
|
| 2266 |
+
{
|
| 2267 |
+
"epoch": 0.8,
|
| 2268 |
+
"learning_rate": 4.084432717678101e-06,
|
| 2269 |
+
"loss": 0.4785,
|
| 2270 |
+
"step": 7540
|
| 2271 |
+
},
|
| 2272 |
+
{
|
| 2273 |
+
"epoch": 0.8,
|
| 2274 |
+
"learning_rate": 4.042216358839051e-06,
|
| 2275 |
+
"loss": 0.4247,
|
| 2276 |
+
"step": 7560
|
| 2277 |
+
},
|
| 2278 |
+
{
|
| 2279 |
+
"epoch": 0.8,
|
| 2280 |
+
"learning_rate": 4.000000000000001e-06,
|
| 2281 |
+
"loss": 0.3963,
|
| 2282 |
+
"step": 7580
|
| 2283 |
+
},
|
| 2284 |
+
{
|
| 2285 |
+
"epoch": 0.8,
|
| 2286 |
+
"learning_rate": 3.95778364116095e-06,
|
| 2287 |
+
"loss": 0.5378,
|
| 2288 |
+
"step": 7600
|
| 2289 |
+
},
|
| 2290 |
+
{
|
| 2291 |
+
"epoch": 0.8,
|
| 2292 |
+
"learning_rate": 3.9155672823218996e-06,
|
| 2293 |
+
"loss": 0.4826,
|
| 2294 |
+
"step": 7620
|
| 2295 |
+
},
|
| 2296 |
+
{
|
| 2297 |
+
"epoch": 0.81,
|
| 2298 |
+
"learning_rate": 3.8733509234828494e-06,
|
| 2299 |
+
"loss": 0.5258,
|
| 2300 |
+
"step": 7640
|
| 2301 |
+
},
|
| 2302 |
+
{
|
| 2303 |
+
"epoch": 0.81,
|
| 2304 |
+
"learning_rate": 3.831134564643799e-06,
|
| 2305 |
+
"loss": 0.7325,
|
| 2306 |
+
"step": 7660
|
| 2307 |
+
},
|
| 2308 |
+
{
|
| 2309 |
+
"epoch": 0.81,
|
| 2310 |
+
"learning_rate": 3.7889182058047496e-06,
|
| 2311 |
+
"loss": 0.6236,
|
| 2312 |
+
"step": 7680
|
| 2313 |
+
},
|
| 2314 |
+
{
|
| 2315 |
+
"epoch": 0.81,
|
| 2316 |
+
"learning_rate": 3.7467018469656995e-06,
|
| 2317 |
+
"loss": 0.5897,
|
| 2318 |
+
"step": 7700
|
| 2319 |
+
},
|
| 2320 |
+
{
|
| 2321 |
+
"epoch": 0.81,
|
| 2322 |
+
"learning_rate": 3.7044854881266494e-06,
|
| 2323 |
+
"loss": 0.4477,
|
| 2324 |
+
"step": 7720
|
| 2325 |
+
},
|
| 2326 |
+
{
|
| 2327 |
+
"epoch": 0.82,
|
| 2328 |
+
"learning_rate": 3.6622691292875997e-06,
|
| 2329 |
+
"loss": 0.5014,
|
| 2330 |
+
"step": 7740
|
| 2331 |
+
},
|
| 2332 |
+
{
|
| 2333 |
+
"epoch": 0.82,
|
| 2334 |
+
"learning_rate": 3.6200527704485487e-06,
|
| 2335 |
+
"loss": 0.6654,
|
| 2336 |
+
"step": 7760
|
| 2337 |
+
},
|
| 2338 |
+
{
|
| 2339 |
+
"epoch": 0.82,
|
| 2340 |
+
"learning_rate": 3.577836411609499e-06,
|
| 2341 |
+
"loss": 0.5581,
|
| 2342 |
+
"step": 7780
|
| 2343 |
+
},
|
| 2344 |
+
{
|
| 2345 |
+
"epoch": 0.82,
|
| 2346 |
+
"learning_rate": 3.535620052770449e-06,
|
| 2347 |
+
"loss": 0.5271,
|
| 2348 |
+
"step": 7800
|
| 2349 |
+
},
|
| 2350 |
+
{
|
| 2351 |
+
"epoch": 0.83,
|
| 2352 |
+
"learning_rate": 3.4934036939313987e-06,
|
| 2353 |
+
"loss": 0.6031,
|
| 2354 |
+
"step": 7820
|
| 2355 |
+
},
|
| 2356 |
+
{
|
| 2357 |
+
"epoch": 0.83,
|
| 2358 |
+
"learning_rate": 3.4511873350923486e-06,
|
| 2359 |
+
"loss": 0.3737,
|
| 2360 |
+
"step": 7840
|
| 2361 |
+
},
|
| 2362 |
+
{
|
| 2363 |
+
"epoch": 0.83,
|
| 2364 |
+
"learning_rate": 3.4089709762532985e-06,
|
| 2365 |
+
"loss": 0.5038,
|
| 2366 |
+
"step": 7860
|
| 2367 |
+
},
|
| 2368 |
+
{
|
| 2369 |
+
"epoch": 0.83,
|
| 2370 |
+
"learning_rate": 3.3667546174142484e-06,
|
| 2371 |
+
"loss": 0.5567,
|
| 2372 |
+
"step": 7880
|
| 2373 |
+
},
|
| 2374 |
+
{
|
| 2375 |
+
"epoch": 0.83,
|
| 2376 |
+
"learning_rate": 3.3245382585751982e-06,
|
| 2377 |
+
"loss": 0.6011,
|
| 2378 |
+
"step": 7900
|
| 2379 |
+
},
|
| 2380 |
+
{
|
| 2381 |
+
"epoch": 0.84,
|
| 2382 |
+
"learning_rate": 3.2823218997361477e-06,
|
| 2383 |
+
"loss": 0.5105,
|
| 2384 |
+
"step": 7920
|
| 2385 |
+
},
|
| 2386 |
+
{
|
| 2387 |
+
"epoch": 0.84,
|
| 2388 |
+
"learning_rate": 3.2401055408970976e-06,
|
| 2389 |
+
"loss": 0.3994,
|
| 2390 |
+
"step": 7940
|
| 2391 |
+
},
|
| 2392 |
+
{
|
| 2393 |
+
"epoch": 0.84,
|
| 2394 |
+
"learning_rate": 3.197889182058048e-06,
|
| 2395 |
+
"loss": 0.4226,
|
| 2396 |
+
"step": 7960
|
| 2397 |
+
},
|
| 2398 |
+
{
|
| 2399 |
+
"epoch": 0.84,
|
| 2400 |
+
"learning_rate": 3.1556728232189977e-06,
|
| 2401 |
+
"loss": 0.3077,
|
| 2402 |
+
"step": 7980
|
| 2403 |
+
},
|
| 2404 |
+
{
|
| 2405 |
+
"epoch": 0.84,
|
| 2406 |
+
"learning_rate": 3.1134564643799476e-06,
|
| 2407 |
+
"loss": 0.4729,
|
| 2408 |
+
"step": 8000
|
| 2409 |
+
},
|
| 2410 |
+
{
|
| 2411 |
+
"epoch": 0.85,
|
| 2412 |
+
"learning_rate": 3.0712401055408975e-06,
|
| 2413 |
+
"loss": 0.5839,
|
| 2414 |
+
"step": 8020
|
| 2415 |
+
},
|
| 2416 |
+
{
|
| 2417 |
+
"epoch": 0.85,
|
| 2418 |
+
"learning_rate": 3.0290237467018474e-06,
|
| 2419 |
+
"loss": 0.6405,
|
| 2420 |
+
"step": 8040
|
| 2421 |
+
},
|
| 2422 |
+
{
|
| 2423 |
+
"epoch": 0.85,
|
| 2424 |
+
"learning_rate": 2.9868073878627972e-06,
|
| 2425 |
+
"loss": 0.6151,
|
| 2426 |
+
"step": 8060
|
| 2427 |
+
},
|
| 2428 |
+
{
|
| 2429 |
+
"epoch": 0.85,
|
| 2430 |
+
"learning_rate": 2.9445910290237467e-06,
|
| 2431 |
+
"loss": 0.3509,
|
| 2432 |
+
"step": 8080
|
| 2433 |
+
},
|
| 2434 |
+
{
|
| 2435 |
+
"epoch": 0.85,
|
| 2436 |
+
"learning_rate": 2.9023746701846966e-06,
|
| 2437 |
+
"loss": 0.4864,
|
| 2438 |
+
"step": 8100
|
| 2439 |
+
},
|
| 2440 |
+
{
|
| 2441 |
+
"epoch": 0.86,
|
| 2442 |
+
"learning_rate": 2.8601583113456464e-06,
|
| 2443 |
+
"loss": 0.5188,
|
| 2444 |
+
"step": 8120
|
| 2445 |
+
},
|
| 2446 |
+
{
|
| 2447 |
+
"epoch": 0.86,
|
| 2448 |
+
"learning_rate": 2.8179419525065967e-06,
|
| 2449 |
+
"loss": 0.602,
|
| 2450 |
+
"step": 8140
|
| 2451 |
+
},
|
| 2452 |
+
{
|
| 2453 |
+
"epoch": 0.86,
|
| 2454 |
+
"learning_rate": 2.7757255936675466e-06,
|
| 2455 |
+
"loss": 0.4617,
|
| 2456 |
+
"step": 8160
|
| 2457 |
+
},
|
| 2458 |
+
{
|
| 2459 |
+
"epoch": 0.86,
|
| 2460 |
+
"learning_rate": 2.7335092348284965e-06,
|
| 2461 |
+
"loss": 0.2785,
|
| 2462 |
+
"step": 8180
|
| 2463 |
+
},
|
| 2464 |
+
{
|
| 2465 |
+
"epoch": 0.87,
|
| 2466 |
+
"learning_rate": 2.6912928759894464e-06,
|
| 2467 |
+
"loss": 0.3212,
|
| 2468 |
+
"step": 8200
|
| 2469 |
+
},
|
| 2470 |
+
{
|
| 2471 |
+
"epoch": 0.87,
|
| 2472 |
+
"learning_rate": 2.6490765171503962e-06,
|
| 2473 |
+
"loss": 0.5852,
|
| 2474 |
+
"step": 8220
|
| 2475 |
+
},
|
| 2476 |
+
{
|
| 2477 |
+
"epoch": 0.87,
|
| 2478 |
+
"learning_rate": 2.6068601583113457e-06,
|
| 2479 |
+
"loss": 0.5594,
|
| 2480 |
+
"step": 8240
|
| 2481 |
+
},
|
| 2482 |
+
{
|
| 2483 |
+
"epoch": 0.87,
|
| 2484 |
+
"learning_rate": 2.5646437994722956e-06,
|
| 2485 |
+
"loss": 0.3948,
|
| 2486 |
+
"step": 8260
|
| 2487 |
+
},
|
| 2488 |
+
{
|
| 2489 |
+
"epoch": 0.87,
|
| 2490 |
+
"learning_rate": 2.5224274406332454e-06,
|
| 2491 |
+
"loss": 0.5771,
|
| 2492 |
+
"step": 8280
|
| 2493 |
+
},
|
| 2494 |
+
{
|
| 2495 |
+
"epoch": 0.88,
|
| 2496 |
+
"learning_rate": 2.4802110817941953e-06,
|
| 2497 |
+
"loss": 0.3874,
|
| 2498 |
+
"step": 8300
|
| 2499 |
+
},
|
| 2500 |
+
{
|
| 2501 |
+
"epoch": 0.88,
|
| 2502 |
+
"learning_rate": 2.4379947229551456e-06,
|
| 2503 |
+
"loss": 0.6091,
|
| 2504 |
+
"step": 8320
|
| 2505 |
+
},
|
| 2506 |
+
{
|
| 2507 |
+
"epoch": 0.88,
|
| 2508 |
+
"learning_rate": 2.395778364116095e-06,
|
| 2509 |
+
"loss": 0.471,
|
| 2510 |
+
"step": 8340
|
| 2511 |
+
},
|
| 2512 |
+
{
|
| 2513 |
+
"epoch": 0.88,
|
| 2514 |
+
"learning_rate": 2.353562005277045e-06,
|
| 2515 |
+
"loss": 0.4263,
|
| 2516 |
+
"step": 8360
|
| 2517 |
+
},
|
| 2518 |
+
{
|
| 2519 |
+
"epoch": 0.88,
|
| 2520 |
+
"learning_rate": 2.311345646437995e-06,
|
| 2521 |
+
"loss": 0.4406,
|
| 2522 |
+
"step": 8380
|
| 2523 |
+
},
|
| 2524 |
+
{
|
| 2525 |
+
"epoch": 0.89,
|
| 2526 |
+
"learning_rate": 2.2691292875989447e-06,
|
| 2527 |
+
"loss": 0.3412,
|
| 2528 |
+
"step": 8400
|
| 2529 |
+
},
|
| 2530 |
+
{
|
| 2531 |
+
"epoch": 0.89,
|
| 2532 |
+
"learning_rate": 2.2269129287598945e-06,
|
| 2533 |
+
"loss": 0.4294,
|
| 2534 |
+
"step": 8420
|
| 2535 |
+
},
|
| 2536 |
+
{
|
| 2537 |
+
"epoch": 0.89,
|
| 2538 |
+
"learning_rate": 2.1846965699208444e-06,
|
| 2539 |
+
"loss": 0.6936,
|
| 2540 |
+
"step": 8440
|
| 2541 |
+
},
|
| 2542 |
+
{
|
| 2543 |
+
"epoch": 0.89,
|
| 2544 |
+
"learning_rate": 2.1424802110817943e-06,
|
| 2545 |
+
"loss": 0.5937,
|
| 2546 |
+
"step": 8460
|
| 2547 |
+
},
|
| 2548 |
+
{
|
| 2549 |
+
"epoch": 0.89,
|
| 2550 |
+
"learning_rate": 2.100263852242744e-06,
|
| 2551 |
+
"loss": 0.5567,
|
| 2552 |
+
"step": 8480
|
| 2553 |
+
},
|
| 2554 |
+
{
|
| 2555 |
+
"epoch": 0.9,
|
| 2556 |
+
"learning_rate": 2.058047493403694e-06,
|
| 2557 |
+
"loss": 0.4481,
|
| 2558 |
+
"step": 8500
|
| 2559 |
+
},
|
| 2560 |
+
{
|
| 2561 |
+
"epoch": 0.9,
|
| 2562 |
+
"learning_rate": 2.015831134564644e-06,
|
| 2563 |
+
"loss": 0.42,
|
| 2564 |
+
"step": 8520
|
| 2565 |
+
},
|
| 2566 |
+
{
|
| 2567 |
+
"epoch": 0.9,
|
| 2568 |
+
"learning_rate": 1.973614775725594e-06,
|
| 2569 |
+
"loss": 0.5064,
|
| 2570 |
+
"step": 8540
|
| 2571 |
+
},
|
| 2572 |
+
{
|
| 2573 |
+
"epoch": 0.9,
|
| 2574 |
+
"learning_rate": 1.9313984168865437e-06,
|
| 2575 |
+
"loss": 0.7567,
|
| 2576 |
+
"step": 8560
|
| 2577 |
+
},
|
| 2578 |
+
{
|
| 2579 |
+
"epoch": 0.91,
|
| 2580 |
+
"learning_rate": 1.8891820580474935e-06,
|
| 2581 |
+
"loss": 0.411,
|
| 2582 |
+
"step": 8580
|
| 2583 |
+
},
|
| 2584 |
+
{
|
| 2585 |
+
"epoch": 0.91,
|
| 2586 |
+
"learning_rate": 1.8469656992084434e-06,
|
| 2587 |
+
"loss": 0.4306,
|
| 2588 |
+
"step": 8600
|
| 2589 |
+
},
|
| 2590 |
+
{
|
| 2591 |
+
"epoch": 0.91,
|
| 2592 |
+
"learning_rate": 1.8047493403693933e-06,
|
| 2593 |
+
"loss": 0.6867,
|
| 2594 |
+
"step": 8620
|
| 2595 |
+
},
|
| 2596 |
+
{
|
| 2597 |
+
"epoch": 0.91,
|
| 2598 |
+
"learning_rate": 1.7625329815303432e-06,
|
| 2599 |
+
"loss": 0.6504,
|
| 2600 |
+
"step": 8640
|
| 2601 |
+
},
|
| 2602 |
+
{
|
| 2603 |
+
"epoch": 0.91,
|
| 2604 |
+
"learning_rate": 1.7203166226912928e-06,
|
| 2605 |
+
"loss": 0.4825,
|
| 2606 |
+
"step": 8660
|
| 2607 |
+
},
|
| 2608 |
+
{
|
| 2609 |
+
"epoch": 0.92,
|
| 2610 |
+
"learning_rate": 1.678100263852243e-06,
|
| 2611 |
+
"loss": 0.6795,
|
| 2612 |
+
"step": 8680
|
| 2613 |
+
},
|
| 2614 |
+
{
|
| 2615 |
+
"epoch": 0.92,
|
| 2616 |
+
"learning_rate": 1.6358839050131928e-06,
|
| 2617 |
+
"loss": 0.395,
|
| 2618 |
+
"step": 8700
|
| 2619 |
+
},
|
| 2620 |
+
{
|
| 2621 |
+
"epoch": 0.92,
|
| 2622 |
+
"learning_rate": 1.5936675461741427e-06,
|
| 2623 |
+
"loss": 0.4376,
|
| 2624 |
+
"step": 8720
|
| 2625 |
+
},
|
| 2626 |
+
{
|
| 2627 |
+
"epoch": 0.92,
|
| 2628 |
+
"learning_rate": 1.5514511873350923e-06,
|
| 2629 |
+
"loss": 0.5183,
|
| 2630 |
+
"step": 8740
|
| 2631 |
+
},
|
| 2632 |
+
{
|
| 2633 |
+
"epoch": 0.92,
|
| 2634 |
+
"learning_rate": 1.5092348284960422e-06,
|
| 2635 |
+
"loss": 0.5289,
|
| 2636 |
+
"step": 8760
|
| 2637 |
+
},
|
| 2638 |
+
{
|
| 2639 |
+
"epoch": 0.93,
|
| 2640 |
+
"learning_rate": 1.4670184696569923e-06,
|
| 2641 |
+
"loss": 0.6665,
|
| 2642 |
+
"step": 8780
|
| 2643 |
+
},
|
| 2644 |
+
{
|
| 2645 |
+
"epoch": 0.93,
|
| 2646 |
+
"learning_rate": 1.4248021108179422e-06,
|
| 2647 |
+
"loss": 0.5317,
|
| 2648 |
+
"step": 8800
|
| 2649 |
+
},
|
| 2650 |
+
{
|
| 2651 |
+
"epoch": 0.93,
|
| 2652 |
+
"learning_rate": 1.3825857519788918e-06,
|
| 2653 |
+
"loss": 0.2245,
|
| 2654 |
+
"step": 8820
|
| 2655 |
+
},
|
| 2656 |
+
{
|
| 2657 |
+
"epoch": 0.93,
|
| 2658 |
+
"learning_rate": 1.3403693931398417e-06,
|
| 2659 |
+
"loss": 0.44,
|
| 2660 |
+
"step": 8840
|
| 2661 |
+
},
|
| 2662 |
+
{
|
| 2663 |
+
"epoch": 0.94,
|
| 2664 |
+
"learning_rate": 1.2981530343007918e-06,
|
| 2665 |
+
"loss": 0.4556,
|
| 2666 |
+
"step": 8860
|
| 2667 |
+
},
|
| 2668 |
+
{
|
| 2669 |
+
"epoch": 0.94,
|
| 2670 |
+
"learning_rate": 1.2559366754617417e-06,
|
| 2671 |
+
"loss": 0.5804,
|
| 2672 |
+
"step": 8880
|
| 2673 |
+
},
|
| 2674 |
+
{
|
| 2675 |
+
"epoch": 0.94,
|
| 2676 |
+
"learning_rate": 1.2137203166226915e-06,
|
| 2677 |
+
"loss": 0.4209,
|
| 2678 |
+
"step": 8900
|
| 2679 |
+
},
|
| 2680 |
+
{
|
| 2681 |
+
"epoch": 0.94,
|
| 2682 |
+
"learning_rate": 1.1715039577836412e-06,
|
| 2683 |
+
"loss": 0.3838,
|
| 2684 |
+
"step": 8920
|
| 2685 |
+
},
|
| 2686 |
+
{
|
| 2687 |
+
"epoch": 0.94,
|
| 2688 |
+
"learning_rate": 1.129287598944591e-06,
|
| 2689 |
+
"loss": 0.4607,
|
| 2690 |
+
"step": 8940
|
| 2691 |
+
},
|
| 2692 |
+
{
|
| 2693 |
+
"epoch": 0.95,
|
| 2694 |
+
"learning_rate": 1.087071240105541e-06,
|
| 2695 |
+
"loss": 0.5023,
|
| 2696 |
+
"step": 8960
|
| 2697 |
+
},
|
| 2698 |
+
{
|
| 2699 |
+
"epoch": 0.95,
|
| 2700 |
+
"learning_rate": 1.0448548812664908e-06,
|
| 2701 |
+
"loss": 0.3096,
|
| 2702 |
+
"step": 8980
|
| 2703 |
+
},
|
| 2704 |
+
{
|
| 2705 |
+
"epoch": 0.95,
|
| 2706 |
+
"learning_rate": 1.0026385224274407e-06,
|
| 2707 |
+
"loss": 0.6846,
|
| 2708 |
+
"step": 9000
|
| 2709 |
+
},
|
| 2710 |
+
{
|
| 2711 |
+
"epoch": 0.95,
|
| 2712 |
+
"learning_rate": 9.604221635883906e-07,
|
| 2713 |
+
"loss": 0.3319,
|
| 2714 |
+
"step": 9020
|
| 2715 |
+
},
|
| 2716 |
+
{
|
| 2717 |
+
"epoch": 0.95,
|
| 2718 |
+
"learning_rate": 9.182058047493404e-07,
|
| 2719 |
+
"loss": 0.5508,
|
| 2720 |
+
"step": 9040
|
| 2721 |
+
},
|
| 2722 |
+
{
|
| 2723 |
+
"epoch": 0.96,
|
| 2724 |
+
"learning_rate": 8.759894459102903e-07,
|
| 2725 |
+
"loss": 0.6869,
|
| 2726 |
+
"step": 9060
|
| 2727 |
+
},
|
| 2728 |
+
{
|
| 2729 |
+
"epoch": 0.96,
|
| 2730 |
+
"learning_rate": 8.337730870712402e-07,
|
| 2731 |
+
"loss": 0.4138,
|
| 2732 |
+
"step": 9080
|
| 2733 |
+
},
|
| 2734 |
+
{
|
| 2735 |
+
"epoch": 0.96,
|
| 2736 |
+
"learning_rate": 7.915567282321901e-07,
|
| 2737 |
+
"loss": 0.6574,
|
| 2738 |
+
"step": 9100
|
| 2739 |
+
},
|
| 2740 |
+
{
|
| 2741 |
+
"epoch": 0.96,
|
| 2742 |
+
"learning_rate": 7.493403693931399e-07,
|
| 2743 |
+
"loss": 0.716,
|
| 2744 |
+
"step": 9120
|
| 2745 |
+
},
|
| 2746 |
+
{
|
| 2747 |
+
"epoch": 0.96,
|
| 2748 |
+
"learning_rate": 7.071240105540898e-07,
|
| 2749 |
+
"loss": 0.4334,
|
| 2750 |
+
"step": 9140
|
| 2751 |
+
},
|
| 2752 |
+
{
|
| 2753 |
+
"epoch": 0.97,
|
| 2754 |
+
"learning_rate": 6.649076517150396e-07,
|
| 2755 |
+
"loss": 0.481,
|
| 2756 |
+
"step": 9160
|
| 2757 |
+
},
|
| 2758 |
+
{
|
| 2759 |
+
"epoch": 0.97,
|
| 2760 |
+
"learning_rate": 6.226912928759895e-07,
|
| 2761 |
+
"loss": 0.6728,
|
| 2762 |
+
"step": 9180
|
| 2763 |
+
},
|
| 2764 |
+
{
|
| 2765 |
+
"epoch": 0.97,
|
| 2766 |
+
"learning_rate": 5.804749340369393e-07,
|
| 2767 |
+
"loss": 0.6311,
|
| 2768 |
+
"step": 9200
|
| 2769 |
+
},
|
| 2770 |
+
{
|
| 2771 |
+
"epoch": 0.97,
|
| 2772 |
+
"learning_rate": 5.382585751978892e-07,
|
| 2773 |
+
"loss": 0.4732,
|
| 2774 |
+
"step": 9220
|
| 2775 |
+
},
|
| 2776 |
+
{
|
| 2777 |
+
"epoch": 0.98,
|
| 2778 |
+
"learning_rate": 4.960422163588391e-07,
|
| 2779 |
+
"loss": 0.4456,
|
| 2780 |
+
"step": 9240
|
| 2781 |
+
},
|
| 2782 |
+
{
|
| 2783 |
+
"epoch": 0.98,
|
| 2784 |
+
"learning_rate": 4.5382585751978896e-07,
|
| 2785 |
+
"loss": 0.7213,
|
| 2786 |
+
"step": 9260
|
| 2787 |
+
},
|
| 2788 |
+
{
|
| 2789 |
+
"epoch": 0.98,
|
| 2790 |
+
"learning_rate": 4.1160949868073883e-07,
|
| 2791 |
+
"loss": 0.6193,
|
| 2792 |
+
"step": 9280
|
| 2793 |
+
},
|
| 2794 |
+
{
|
| 2795 |
+
"epoch": 0.98,
|
| 2796 |
+
"learning_rate": 3.693931398416887e-07,
|
| 2797 |
+
"loss": 0.4499,
|
| 2798 |
+
"step": 9300
|
| 2799 |
+
},
|
| 2800 |
+
{
|
| 2801 |
+
"epoch": 0.98,
|
| 2802 |
+
"learning_rate": 3.271767810026386e-07,
|
| 2803 |
+
"loss": 0.5332,
|
| 2804 |
+
"step": 9320
|
| 2805 |
+
},
|
| 2806 |
+
{
|
| 2807 |
+
"epoch": 0.99,
|
| 2808 |
+
"learning_rate": 2.849604221635884e-07,
|
| 2809 |
+
"loss": 0.5398,
|
| 2810 |
+
"step": 9340
|
| 2811 |
+
},
|
| 2812 |
+
{
|
| 2813 |
+
"epoch": 0.99,
|
| 2814 |
+
"learning_rate": 2.427440633245383e-07,
|
| 2815 |
+
"loss": 0.5923,
|
| 2816 |
+
"step": 9360
|
| 2817 |
+
},
|
| 2818 |
+
{
|
| 2819 |
+
"epoch": 0.99,
|
| 2820 |
+
"learning_rate": 2.0052770448548815e-07,
|
| 2821 |
+
"loss": 0.4648,
|
| 2822 |
+
"step": 9380
|
| 2823 |
+
},
|
| 2824 |
+
{
|
| 2825 |
+
"epoch": 0.99,
|
| 2826 |
+
"learning_rate": 1.5831134564643802e-07,
|
| 2827 |
+
"loss": 0.396,
|
| 2828 |
+
"step": 9400
|
| 2829 |
+
},
|
| 2830 |
+
{
|
| 2831 |
+
"epoch": 0.99,
|
| 2832 |
+
"learning_rate": 1.1609498680738787e-07,
|
| 2833 |
+
"loss": 0.4959,
|
| 2834 |
+
"step": 9420
|
| 2835 |
+
},
|
| 2836 |
+
{
|
| 2837 |
+
"epoch": 1.0,
|
| 2838 |
+
"learning_rate": 7.387862796833773e-08,
|
| 2839 |
+
"loss": 0.5742,
|
| 2840 |
+
"step": 9440
|
| 2841 |
+
},
|
| 2842 |
+
{
|
| 2843 |
+
"epoch": 1.0,
|
| 2844 |
+
"learning_rate": 3.16622691292876e-08,
|
| 2845 |
+
"loss": 0.7434,
|
| 2846 |
+
"step": 9460
|
| 2847 |
+
},
|
| 2848 |
+
{
|
| 2849 |
+
"epoch": 1.0,
|
| 2850 |
+
"step": 9475,
|
| 2851 |
+
"total_flos": 18716049801216.0,
|
| 2852 |
+
"train_loss": 0.5445125763019975,
|
| 2853 |
+
"train_runtime": 21957.5944,
|
| 2854 |
+
"train_samples_per_second": 0.432,
|
| 2855 |
+
"train_steps_per_second": 0.432
|
| 2856 |
+
}
|
| 2857 |
+
],
|
| 2858 |
+
"max_steps": 9475,
|
| 2859 |
+
"num_train_epochs": 1,
|
| 2860 |
+
"total_flos": 18716049801216.0,
|
| 2861 |
+
"trial_name": null,
|
| 2862 |
+
"trial_params": null
|
| 2863 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c0dc454463943c929e6b9a49b3c99312d40e96ac1abb2be54f10de686dbfd799
|
| 3 |
+
size 4859
|