Upload folder using huggingface_hub
Browse files- README.md +33 -0
- config.json +68 -0
- model.safetensors +3 -0
- optimizer.pt +3 -0
- preprocessor_config.json +15 -0
- rng_state.pth +3 -0
- scaler.pt +3 -0
- scheduler.pt +3 -0
- trainer_state.json +3046 -0
- training_args.bin +3 -0
README.md
ADDED
|
@@ -0,0 +1,33 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
language: multilingual
|
| 3 |
+
license: apache-2.0
|
| 4 |
+
tags:
|
| 5 |
+
- audio-classification
|
| 6 |
+
- whisper
|
| 7 |
+
- speech
|
| 8 |
+
- music
|
| 9 |
+
datasets:
|
| 10 |
+
- AIGenLab/speech-music-merge
|
| 11 |
+
base_model: openai/whisper-small
|
| 12 |
+
pipeline_tag: audio-classification
|
| 13 |
+
---
|
| 14 |
+
|
| 15 |
+
# Whisper Small - Speech/Music Classifier
|
| 16 |
+
|
| 17 |
+
Whisper Small fine-tuned for speech vs music classification.
|
| 18 |
+
|
| 19 |
+
## Usage
|
| 20 |
+
```python
|
| 21 |
+
from transformers import pipeline
|
| 22 |
+
|
| 23 |
+
classifier = pipeline("audio-classification", model="AIGenLab/whisper-small-speech-music-classifier")
|
| 24 |
+
result = classifier("audio.wav")
|
| 25 |
+
print(result)
|
| 26 |
+
```
|
| 27 |
+
|
| 28 |
+
## Details
|
| 29 |
+
|
| 30 |
+
- **Base:** openai/whisper-small
|
| 31 |
+
- **Dataset:** AIGenLab/speech-music-merge
|
| 32 |
+
- **Classes:** speech, music
|
| 33 |
+
- **Sample Rate:** 16kHz
|
config.json
ADDED
|
@@ -0,0 +1,68 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"activation_dropout": 0.0,
|
| 3 |
+
"activation_function": "gelu",
|
| 4 |
+
"apply_spec_augment": false,
|
| 5 |
+
"architectures": [
|
| 6 |
+
"WhisperForAudioClassification"
|
| 7 |
+
],
|
| 8 |
+
"attention_dropout": 0.0,
|
| 9 |
+
"begin_suppress_tokens": null,
|
| 10 |
+
"bos_token_id": 50257,
|
| 11 |
+
"classifier_proj_size": 256,
|
| 12 |
+
"d_model": 768,
|
| 13 |
+
"decoder_attention_heads": 12,
|
| 14 |
+
"decoder_ffn_dim": 3072,
|
| 15 |
+
"decoder_layerdrop": 0.0,
|
| 16 |
+
"decoder_layers": 12,
|
| 17 |
+
"decoder_start_token_id": 50258,
|
| 18 |
+
"dropout": 0.0,
|
| 19 |
+
"dtype": "float32",
|
| 20 |
+
"encoder_attention_heads": 12,
|
| 21 |
+
"encoder_ffn_dim": 3072,
|
| 22 |
+
"encoder_layerdrop": 0.0,
|
| 23 |
+
"encoder_layers": 12,
|
| 24 |
+
"eos_token_id": 50257,
|
| 25 |
+
"forced_decoder_ids": [
|
| 26 |
+
[
|
| 27 |
+
1,
|
| 28 |
+
50259
|
| 29 |
+
],
|
| 30 |
+
[
|
| 31 |
+
2,
|
| 32 |
+
50359
|
| 33 |
+
],
|
| 34 |
+
[
|
| 35 |
+
3,
|
| 36 |
+
50363
|
| 37 |
+
]
|
| 38 |
+
],
|
| 39 |
+
"id2label": {
|
| 40 |
+
"0": "music",
|
| 41 |
+
"1": "speech"
|
| 42 |
+
},
|
| 43 |
+
"init_std": 0.02,
|
| 44 |
+
"is_encoder_decoder": true,
|
| 45 |
+
"label2id": {
|
| 46 |
+
"music": 0,
|
| 47 |
+
"speech": 1
|
| 48 |
+
},
|
| 49 |
+
"mask_feature_length": 10,
|
| 50 |
+
"mask_feature_min_masks": 0,
|
| 51 |
+
"mask_feature_prob": 0.0,
|
| 52 |
+
"mask_time_length": 10,
|
| 53 |
+
"mask_time_min_masks": 2,
|
| 54 |
+
"mask_time_prob": 0.05,
|
| 55 |
+
"max_length": null,
|
| 56 |
+
"max_source_positions": 1500,
|
| 57 |
+
"max_target_positions": 448,
|
| 58 |
+
"median_filter_width": 7,
|
| 59 |
+
"model_type": "whisper",
|
| 60 |
+
"num_hidden_layers": 12,
|
| 61 |
+
"num_mel_bins": 80,
|
| 62 |
+
"pad_token_id": 50257,
|
| 63 |
+
"scale_embedding": false,
|
| 64 |
+
"transformers_version": "4.57.1",
|
| 65 |
+
"use_cache": true,
|
| 66 |
+
"use_weighted_layer_sum": false,
|
| 67 |
+
"vocab_size": 51865
|
| 68 |
+
}
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:77a8f21ae141afe3541404057b0097bfa813847bab66c58ffbb400029b89c1e2
|
| 3 |
+
size 353426360
|
optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9735d3e800e15f6237c8cddb02f120efb25b10840921ed6a28e13158b3e79567
|
| 3 |
+
size 697755403
|
preprocessor_config.json
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"chunk_length": 30,
|
| 3 |
+
"dither": 0.0,
|
| 4 |
+
"feature_extractor_type": "WhisperFeatureExtractor",
|
| 5 |
+
"feature_size": 80,
|
| 6 |
+
"hop_length": 160,
|
| 7 |
+
"n_fft": 400,
|
| 8 |
+
"n_samples": 480000,
|
| 9 |
+
"nb_max_frames": 3000,
|
| 10 |
+
"padding_side": "right",
|
| 11 |
+
"padding_value": 0.0,
|
| 12 |
+
"processor_class": "WhisperProcessor",
|
| 13 |
+
"return_attention_mask": false,
|
| 14 |
+
"sampling_rate": 16000
|
| 15 |
+
}
|
rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c20da6e1cee6dd1223f5a0e5be8b8689e8bb621abc6091198f1551010f7d93de
|
| 3 |
+
size 14645
|
scaler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:702984c680199a0e28a46b8d953b4d6a6c82386d777f96e7c9bdcfa92e49f34f
|
| 3 |
+
size 1383
|
scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:12d1f602d774c0a9625a71174bf4749c7d66c086f60aee402aae44dffd4becf4
|
| 3 |
+
size 1465
|
trainer_state.json
ADDED
|
@@ -0,0 +1,3046 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": 4284,
|
| 3 |
+
"best_metric": 0.9999124458258547,
|
| 4 |
+
"best_model_checkpoint": "./aynur_model3/checkpoint-4284",
|
| 5 |
+
"epoch": 1.0,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 4284,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.00023346757719021772,
|
| 14 |
+
"grad_norm": 3.623156785964966,
|
| 15 |
+
"learning_rate": 0.0,
|
| 16 |
+
"loss": 0.6427,
|
| 17 |
+
"step": 1
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.002334675771902177,
|
| 21 |
+
"grad_norm": 1.9220610857009888,
|
| 22 |
+
"learning_rate": 8.391608391608393e-06,
|
| 23 |
+
"loss": 0.6315,
|
| 24 |
+
"step": 10
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.004669351543804354,
|
| 28 |
+
"grad_norm": 2.1033425331115723,
|
| 29 |
+
"learning_rate": 1.7715617715617717e-05,
|
| 30 |
+
"loss": 0.4519,
|
| 31 |
+
"step": 20
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.007004027315706532,
|
| 35 |
+
"grad_norm": 0.2185792475938797,
|
| 36 |
+
"learning_rate": 2.7039627039627042e-05,
|
| 37 |
+
"loss": 0.0917,
|
| 38 |
+
"step": 30
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.009338703087608709,
|
| 42 |
+
"grad_norm": 0.013207816518843174,
|
| 43 |
+
"learning_rate": 3.6363636363636364e-05,
|
| 44 |
+
"loss": 0.0019,
|
| 45 |
+
"step": 40
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.011673378859510886,
|
| 49 |
+
"grad_norm": 0.005016999784857035,
|
| 50 |
+
"learning_rate": 4.568764568764569e-05,
|
| 51 |
+
"loss": 0.0002,
|
| 52 |
+
"step": 50
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.014008054631413063,
|
| 56 |
+
"grad_norm": 0.004350466653704643,
|
| 57 |
+
"learning_rate": 5.314685314685315e-05,
|
| 58 |
+
"loss": 0.0264,
|
| 59 |
+
"step": 60
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.01634273040331524,
|
| 63 |
+
"grad_norm": 0.0037658039946109056,
|
| 64 |
+
"learning_rate": 6.247086247086247e-05,
|
| 65 |
+
"loss": 0.0609,
|
| 66 |
+
"step": 70
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.018677406175217418,
|
| 70 |
+
"grad_norm": 0.003915839828550816,
|
| 71 |
+
"learning_rate": 7.17948717948718e-05,
|
| 72 |
+
"loss": 0.0008,
|
| 73 |
+
"step": 80
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.021012081947119593,
|
| 77 |
+
"grad_norm": 0.018606621772050858,
|
| 78 |
+
"learning_rate": 8.111888111888112e-05,
|
| 79 |
+
"loss": 0.0134,
|
| 80 |
+
"step": 90
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.023346757719021772,
|
| 84 |
+
"grad_norm": 0.005168155301362276,
|
| 85 |
+
"learning_rate": 9.044289044289046e-05,
|
| 86 |
+
"loss": 0.0012,
|
| 87 |
+
"step": 100
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.025681433490923947,
|
| 91 |
+
"grad_norm": 0.004493937361985445,
|
| 92 |
+
"learning_rate": 9.976689976689977e-05,
|
| 93 |
+
"loss": 0.0035,
|
| 94 |
+
"step": 110
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.028016109262826126,
|
| 98 |
+
"grad_norm": 0.006067783106118441,
|
| 99 |
+
"learning_rate": 0.00010909090909090909,
|
| 100 |
+
"loss": 0.0091,
|
| 101 |
+
"step": 120
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.030350785034728302,
|
| 105 |
+
"grad_norm": 0.017138389870524406,
|
| 106 |
+
"learning_rate": 0.00011841491841491842,
|
| 107 |
+
"loss": 0.0003,
|
| 108 |
+
"step": 130
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 0.03268546080663048,
|
| 112 |
+
"grad_norm": 0.06054692715406418,
|
| 113 |
+
"learning_rate": 0.00012773892773892774,
|
| 114 |
+
"loss": 0.0036,
|
| 115 |
+
"step": 140
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.03502013657853266,
|
| 119 |
+
"grad_norm": 0.09328058362007141,
|
| 120 |
+
"learning_rate": 0.00013706293706293706,
|
| 121 |
+
"loss": 0.1175,
|
| 122 |
+
"step": 150
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.037354812350434835,
|
| 126 |
+
"grad_norm": 0.006551404017955065,
|
| 127 |
+
"learning_rate": 0.00014638694638694638,
|
| 128 |
+
"loss": 0.0009,
|
| 129 |
+
"step": 160
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.03968948812233701,
|
| 133 |
+
"grad_norm": 0.08505500108003616,
|
| 134 |
+
"learning_rate": 0.0001557109557109557,
|
| 135 |
+
"loss": 0.0159,
|
| 136 |
+
"step": 170
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.042024163894239186,
|
| 140 |
+
"grad_norm": 0.007050831336528063,
|
| 141 |
+
"learning_rate": 0.00016503496503496504,
|
| 142 |
+
"loss": 0.0008,
|
| 143 |
+
"step": 180
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.04435883966614136,
|
| 147 |
+
"grad_norm": 0.031755417585372925,
|
| 148 |
+
"learning_rate": 0.00017435897435897436,
|
| 149 |
+
"loss": 0.05,
|
| 150 |
+
"step": 190
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 0.046693515438043544,
|
| 154 |
+
"grad_norm": 0.06047971546649933,
|
| 155 |
+
"learning_rate": 0.00018368298368298368,
|
| 156 |
+
"loss": 0.0186,
|
| 157 |
+
"step": 200
|
| 158 |
+
},
|
| 159 |
+
{
|
| 160 |
+
"epoch": 0.04902819120994572,
|
| 161 |
+
"grad_norm": 0.11913339048624039,
|
| 162 |
+
"learning_rate": 0.000193006993006993,
|
| 163 |
+
"loss": 0.0093,
|
| 164 |
+
"step": 210
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.051362866981847895,
|
| 168 |
+
"grad_norm": 0.006012667436152697,
|
| 169 |
+
"learning_rate": 0.00020233100233100232,
|
| 170 |
+
"loss": 0.0105,
|
| 171 |
+
"step": 220
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.05369754275375007,
|
| 175 |
+
"grad_norm": 0.006747289560735226,
|
| 176 |
+
"learning_rate": 0.00021165501165501164,
|
| 177 |
+
"loss": 0.0003,
|
| 178 |
+
"step": 230
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.05603221852565225,
|
| 182 |
+
"grad_norm": 5.616020202636719,
|
| 183 |
+
"learning_rate": 0.00022097902097902096,
|
| 184 |
+
"loss": 0.0484,
|
| 185 |
+
"step": 240
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.05836689429755443,
|
| 189 |
+
"grad_norm": 0.04582913592457771,
|
| 190 |
+
"learning_rate": 0.00023030303030303033,
|
| 191 |
+
"loss": 0.0243,
|
| 192 |
+
"step": 250
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 0.060701570069456603,
|
| 196 |
+
"grad_norm": 5.976644992828369,
|
| 197 |
+
"learning_rate": 0.00023962703962703965,
|
| 198 |
+
"loss": 0.0437,
|
| 199 |
+
"step": 260
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 0.06303624584135878,
|
| 203 |
+
"grad_norm": 0.006856445223093033,
|
| 204 |
+
"learning_rate": 0.00024895104895104897,
|
| 205 |
+
"loss": 0.0933,
|
| 206 |
+
"step": 270
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.06537092161326095,
|
| 210 |
+
"grad_norm": 0.031221158802509308,
|
| 211 |
+
"learning_rate": 0.0002582750582750583,
|
| 212 |
+
"loss": 0.015,
|
| 213 |
+
"step": 280
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.06770559738516313,
|
| 217 |
+
"grad_norm": 0.5157426595687866,
|
| 218 |
+
"learning_rate": 0.0002675990675990676,
|
| 219 |
+
"loss": 0.0367,
|
| 220 |
+
"step": 290
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.07004027315706532,
|
| 224 |
+
"grad_norm": 0.011151552200317383,
|
| 225 |
+
"learning_rate": 0.00027692307692307695,
|
| 226 |
+
"loss": 0.0501,
|
| 227 |
+
"step": 300
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 0.0723749489289675,
|
| 231 |
+
"grad_norm": 0.03010399080812931,
|
| 232 |
+
"learning_rate": 0.00028624708624708624,
|
| 233 |
+
"loss": 0.0472,
|
| 234 |
+
"step": 310
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.07470962470086967,
|
| 238 |
+
"grad_norm": 0.044886477291584015,
|
| 239 |
+
"learning_rate": 0.0002955710955710956,
|
| 240 |
+
"loss": 0.0259,
|
| 241 |
+
"step": 320
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 0.07704430047277185,
|
| 245 |
+
"grad_norm": 2.6834709644317627,
|
| 246 |
+
"learning_rate": 0.0003048951048951049,
|
| 247 |
+
"loss": 0.0336,
|
| 248 |
+
"step": 330
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.07937897624467402,
|
| 252 |
+
"grad_norm": 0.021105894818902016,
|
| 253 |
+
"learning_rate": 0.0003142191142191143,
|
| 254 |
+
"loss": 0.009,
|
| 255 |
+
"step": 340
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 0.0817136520165762,
|
| 259 |
+
"grad_norm": 0.0037551075220108032,
|
| 260 |
+
"learning_rate": 0.00032354312354312357,
|
| 261 |
+
"loss": 0.0003,
|
| 262 |
+
"step": 350
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.08404832778847837,
|
| 266 |
+
"grad_norm": 0.001763952779583633,
|
| 267 |
+
"learning_rate": 0.0003328671328671329,
|
| 268 |
+
"loss": 0.0001,
|
| 269 |
+
"step": 360
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 0.08638300356038055,
|
| 273 |
+
"grad_norm": 0.10732467472553253,
|
| 274 |
+
"learning_rate": 0.0003421911421911422,
|
| 275 |
+
"loss": 0.0161,
|
| 276 |
+
"step": 370
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.08871767933228272,
|
| 280 |
+
"grad_norm": 0.028179295361042023,
|
| 281 |
+
"learning_rate": 0.00035151515151515155,
|
| 282 |
+
"loss": 0.0233,
|
| 283 |
+
"step": 380
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 0.09105235510418491,
|
| 287 |
+
"grad_norm": 1.2709873914718628,
|
| 288 |
+
"learning_rate": 0.00036083916083916084,
|
| 289 |
+
"loss": 0.0281,
|
| 290 |
+
"step": 390
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.09338703087608709,
|
| 294 |
+
"grad_norm": 0.6180899143218994,
|
| 295 |
+
"learning_rate": 0.0003701631701631702,
|
| 296 |
+
"loss": 0.041,
|
| 297 |
+
"step": 400
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.09572170664798926,
|
| 301 |
+
"grad_norm": 2.0692641735076904,
|
| 302 |
+
"learning_rate": 0.0003794871794871795,
|
| 303 |
+
"loss": 0.1022,
|
| 304 |
+
"step": 410
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 0.09805638241989144,
|
| 308 |
+
"grad_norm": 0.06485776603221893,
|
| 309 |
+
"learning_rate": 0.0003888111888111888,
|
| 310 |
+
"loss": 0.0525,
|
| 311 |
+
"step": 420
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 0.10039105819179361,
|
| 315 |
+
"grad_norm": 0.03012872114777565,
|
| 316 |
+
"learning_rate": 0.00039813519813519817,
|
| 317 |
+
"loss": 0.011,
|
| 318 |
+
"step": 430
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 0.10272573396369579,
|
| 322 |
+
"grad_norm": 4.067958354949951,
|
| 323 |
+
"learning_rate": 0.0003991699092088197,
|
| 324 |
+
"loss": 0.0437,
|
| 325 |
+
"step": 440
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.10506040973559796,
|
| 329 |
+
"grad_norm": 0.11274830996990204,
|
| 330 |
+
"learning_rate": 0.0003981322957198444,
|
| 331 |
+
"loss": 0.0304,
|
| 332 |
+
"step": 450
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.10739508550750014,
|
| 336 |
+
"grad_norm": 1.0862525701522827,
|
| 337 |
+
"learning_rate": 0.00039709468223086904,
|
| 338 |
+
"loss": 0.0078,
|
| 339 |
+
"step": 460
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.10972976127940233,
|
| 343 |
+
"grad_norm": 1.328466773033142,
|
| 344 |
+
"learning_rate": 0.0003960570687418937,
|
| 345 |
+
"loss": 0.0373,
|
| 346 |
+
"step": 470
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 0.1120644370513045,
|
| 350 |
+
"grad_norm": 0.19725392758846283,
|
| 351 |
+
"learning_rate": 0.00039501945525291835,
|
| 352 |
+
"loss": 0.0802,
|
| 353 |
+
"step": 480
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 0.11439911282320668,
|
| 357 |
+
"grad_norm": 1.5255461931228638,
|
| 358 |
+
"learning_rate": 0.00039398184176394295,
|
| 359 |
+
"loss": 0.0357,
|
| 360 |
+
"step": 490
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 0.11673378859510886,
|
| 364 |
+
"grad_norm": 0.059472762048244476,
|
| 365 |
+
"learning_rate": 0.00039294422827496756,
|
| 366 |
+
"loss": 0.0269,
|
| 367 |
+
"step": 500
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 0.11906846436701103,
|
| 371 |
+
"grad_norm": 0.14000196754932404,
|
| 372 |
+
"learning_rate": 0.0003919066147859922,
|
| 373 |
+
"loss": 0.0142,
|
| 374 |
+
"step": 510
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 0.12140314013891321,
|
| 378 |
+
"grad_norm": 0.012143092229962349,
|
| 379 |
+
"learning_rate": 0.00039086900129701687,
|
| 380 |
+
"loss": 0.03,
|
| 381 |
+
"step": 520
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 0.12373781591081538,
|
| 385 |
+
"grad_norm": 0.5115292072296143,
|
| 386 |
+
"learning_rate": 0.00038983138780804153,
|
| 387 |
+
"loss": 0.0342,
|
| 388 |
+
"step": 530
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 0.12607249168271756,
|
| 392 |
+
"grad_norm": 5.99329137802124,
|
| 393 |
+
"learning_rate": 0.0003887937743190662,
|
| 394 |
+
"loss": 0.0446,
|
| 395 |
+
"step": 540
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 0.12840716745461975,
|
| 399 |
+
"grad_norm": 0.1291157752275467,
|
| 400 |
+
"learning_rate": 0.0003877561608300908,
|
| 401 |
+
"loss": 0.0161,
|
| 402 |
+
"step": 550
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 0.1307418432265219,
|
| 406 |
+
"grad_norm": 1.0032269954681396,
|
| 407 |
+
"learning_rate": 0.00038671854734111544,
|
| 408 |
+
"loss": 0.0166,
|
| 409 |
+
"step": 560
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 0.1330765189984241,
|
| 413 |
+
"grad_norm": 16.30780601501465,
|
| 414 |
+
"learning_rate": 0.0003856809338521401,
|
| 415 |
+
"loss": 0.1032,
|
| 416 |
+
"step": 570
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 0.13541119477032626,
|
| 420 |
+
"grad_norm": 4.544622421264648,
|
| 421 |
+
"learning_rate": 0.00038464332036316476,
|
| 422 |
+
"loss": 0.0279,
|
| 423 |
+
"step": 580
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 0.13774587054222845,
|
| 427 |
+
"grad_norm": 5.893352508544922,
|
| 428 |
+
"learning_rate": 0.0003836057068741894,
|
| 429 |
+
"loss": 0.0304,
|
| 430 |
+
"step": 590
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 0.14008054631413064,
|
| 434 |
+
"grad_norm": 0.32387709617614746,
|
| 435 |
+
"learning_rate": 0.000382568093385214,
|
| 436 |
+
"loss": 0.0404,
|
| 437 |
+
"step": 600
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 0.1424152220860328,
|
| 441 |
+
"grad_norm": 0.051759008318185806,
|
| 442 |
+
"learning_rate": 0.0003815304798962387,
|
| 443 |
+
"loss": 0.0182,
|
| 444 |
+
"step": 610
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 0.144749897857935,
|
| 448 |
+
"grad_norm": 0.09647126495838165,
|
| 449 |
+
"learning_rate": 0.0003804928664072633,
|
| 450 |
+
"loss": 0.019,
|
| 451 |
+
"step": 620
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 0.14708457362983715,
|
| 455 |
+
"grad_norm": 0.1588832288980484,
|
| 456 |
+
"learning_rate": 0.00037945525291828793,
|
| 457 |
+
"loss": 0.016,
|
| 458 |
+
"step": 630
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 0.14941924940173934,
|
| 462 |
+
"grad_norm": 0.008694116957485676,
|
| 463 |
+
"learning_rate": 0.0003784176394293126,
|
| 464 |
+
"loss": 0.0042,
|
| 465 |
+
"step": 640
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 0.1517539251736415,
|
| 469 |
+
"grad_norm": 0.040139373391866684,
|
| 470 |
+
"learning_rate": 0.00037738002594033725,
|
| 471 |
+
"loss": 0.0201,
|
| 472 |
+
"step": 650
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 0.1540886009455437,
|
| 476 |
+
"grad_norm": 0.032652173191308975,
|
| 477 |
+
"learning_rate": 0.0003763424124513619,
|
| 478 |
+
"loss": 0.0127,
|
| 479 |
+
"step": 660
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 0.15642327671744585,
|
| 483 |
+
"grad_norm": 0.027006104588508606,
|
| 484 |
+
"learning_rate": 0.0003753047989623865,
|
| 485 |
+
"loss": 0.0044,
|
| 486 |
+
"step": 670
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 0.15875795248934804,
|
| 490 |
+
"grad_norm": 2.665090799331665,
|
| 491 |
+
"learning_rate": 0.00037426718547341117,
|
| 492 |
+
"loss": 0.0179,
|
| 493 |
+
"step": 680
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 0.16109262826125023,
|
| 497 |
+
"grad_norm": 0.011558118276298046,
|
| 498 |
+
"learning_rate": 0.0003732295719844358,
|
| 499 |
+
"loss": 0.0132,
|
| 500 |
+
"step": 690
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 0.1634273040331524,
|
| 504 |
+
"grad_norm": 0.013224626891314983,
|
| 505 |
+
"learning_rate": 0.0003721919584954605,
|
| 506 |
+
"loss": 0.0044,
|
| 507 |
+
"step": 700
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 0.16576197980505458,
|
| 511 |
+
"grad_norm": 0.03626665472984314,
|
| 512 |
+
"learning_rate": 0.00037115434500648514,
|
| 513 |
+
"loss": 0.0019,
|
| 514 |
+
"step": 710
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 0.16809665557695674,
|
| 518 |
+
"grad_norm": 0.10942396521568298,
|
| 519 |
+
"learning_rate": 0.00037011673151750974,
|
| 520 |
+
"loss": 0.0472,
|
| 521 |
+
"step": 720
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 0.17043133134885893,
|
| 525 |
+
"grad_norm": 0.8933264017105103,
|
| 526 |
+
"learning_rate": 0.0003690791180285344,
|
| 527 |
+
"loss": 0.0148,
|
| 528 |
+
"step": 730
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 0.1727660071207611,
|
| 532 |
+
"grad_norm": 0.06506644189357758,
|
| 533 |
+
"learning_rate": 0.000368041504539559,
|
| 534 |
+
"loss": 0.0249,
|
| 535 |
+
"step": 740
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 0.17510068289266328,
|
| 539 |
+
"grad_norm": 0.16546858847141266,
|
| 540 |
+
"learning_rate": 0.00036700389105058366,
|
| 541 |
+
"loss": 0.0105,
|
| 542 |
+
"step": 750
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 0.17743535866456545,
|
| 546 |
+
"grad_norm": 0.032960060983896255,
|
| 547 |
+
"learning_rate": 0.0003659662775616083,
|
| 548 |
+
"loss": 0.0022,
|
| 549 |
+
"step": 760
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 0.17977003443646764,
|
| 553 |
+
"grad_norm": 0.04502630606293678,
|
| 554 |
+
"learning_rate": 0.00036492866407263297,
|
| 555 |
+
"loss": 0.0234,
|
| 556 |
+
"step": 770
|
| 557 |
+
},
|
| 558 |
+
{
|
| 559 |
+
"epoch": 0.18210471020836982,
|
| 560 |
+
"grad_norm": 0.018004219979047775,
|
| 561 |
+
"learning_rate": 0.0003638910505836576,
|
| 562 |
+
"loss": 0.0024,
|
| 563 |
+
"step": 780
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 0.18443938598027199,
|
| 567 |
+
"grad_norm": 0.026385951787233353,
|
| 568 |
+
"learning_rate": 0.00036285343709468223,
|
| 569 |
+
"loss": 0.0079,
|
| 570 |
+
"step": 790
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 0.18677406175217418,
|
| 574 |
+
"grad_norm": 0.02292817272245884,
|
| 575 |
+
"learning_rate": 0.0003618158236057069,
|
| 576 |
+
"loss": 0.0113,
|
| 577 |
+
"step": 800
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 0.18910873752407634,
|
| 581 |
+
"grad_norm": 0.01237889751791954,
|
| 582 |
+
"learning_rate": 0.00036077821011673154,
|
| 583 |
+
"loss": 0.01,
|
| 584 |
+
"step": 810
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 0.19144341329597853,
|
| 588 |
+
"grad_norm": 0.011882675811648369,
|
| 589 |
+
"learning_rate": 0.0003597405966277562,
|
| 590 |
+
"loss": 0.0067,
|
| 591 |
+
"step": 820
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 0.1937780890678807,
|
| 595 |
+
"grad_norm": 0.021467048674821854,
|
| 596 |
+
"learning_rate": 0.00035870298313878086,
|
| 597 |
+
"loss": 0.0138,
|
| 598 |
+
"step": 830
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 0.19611276483978288,
|
| 602 |
+
"grad_norm": 0.04117121547460556,
|
| 603 |
+
"learning_rate": 0.00035766536964980546,
|
| 604 |
+
"loss": 0.0131,
|
| 605 |
+
"step": 840
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 0.19844744061168504,
|
| 609 |
+
"grad_norm": 0.04125780984759331,
|
| 610 |
+
"learning_rate": 0.0003566277561608301,
|
| 611 |
+
"loss": 0.0063,
|
| 612 |
+
"step": 850
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 0.20078211638358723,
|
| 616 |
+
"grad_norm": 0.01064964011311531,
|
| 617 |
+
"learning_rate": 0.0003555901426718547,
|
| 618 |
+
"loss": 0.0068,
|
| 619 |
+
"step": 860
|
| 620 |
+
},
|
| 621 |
+
{
|
| 622 |
+
"epoch": 0.20311679215548942,
|
| 623 |
+
"grad_norm": 0.18367743492126465,
|
| 624 |
+
"learning_rate": 0.0003545525291828794,
|
| 625 |
+
"loss": 0.0197,
|
| 626 |
+
"step": 870
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 0.20545146792739158,
|
| 630 |
+
"grad_norm": 0.017399262636899948,
|
| 631 |
+
"learning_rate": 0.00035351491569390403,
|
| 632 |
+
"loss": 0.0048,
|
| 633 |
+
"step": 880
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 0.20778614369929377,
|
| 637 |
+
"grad_norm": 0.04123668745160103,
|
| 638 |
+
"learning_rate": 0.0003524773022049287,
|
| 639 |
+
"loss": 0.012,
|
| 640 |
+
"step": 890
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 0.21012081947119593,
|
| 644 |
+
"grad_norm": 0.021881213411688805,
|
| 645 |
+
"learning_rate": 0.0003514396887159533,
|
| 646 |
+
"loss": 0.002,
|
| 647 |
+
"step": 900
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 0.21245549524309812,
|
| 651 |
+
"grad_norm": 0.7196763753890991,
|
| 652 |
+
"learning_rate": 0.00035040207522697795,
|
| 653 |
+
"loss": 0.0053,
|
| 654 |
+
"step": 910
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 0.21479017101500028,
|
| 658 |
+
"grad_norm": 0.006894146092236042,
|
| 659 |
+
"learning_rate": 0.0003493644617380026,
|
| 660 |
+
"loss": 0.0004,
|
| 661 |
+
"step": 920
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 0.21712484678690247,
|
| 665 |
+
"grad_norm": 0.02064809761941433,
|
| 666 |
+
"learning_rate": 0.00034832684824902726,
|
| 667 |
+
"loss": 0.0022,
|
| 668 |
+
"step": 930
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 0.21945952255880466,
|
| 672 |
+
"grad_norm": 0.0018330852035433054,
|
| 673 |
+
"learning_rate": 0.0003472892347600519,
|
| 674 |
+
"loss": 0.0008,
|
| 675 |
+
"step": 940
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 0.22179419833070682,
|
| 679 |
+
"grad_norm": 0.00754689471796155,
|
| 680 |
+
"learning_rate": 0.0003462516212710766,
|
| 681 |
+
"loss": 0.0356,
|
| 682 |
+
"step": 950
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 0.224128874102609,
|
| 686 |
+
"grad_norm": 0.0938534140586853,
|
| 687 |
+
"learning_rate": 0.0003452140077821012,
|
| 688 |
+
"loss": 0.0021,
|
| 689 |
+
"step": 960
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 0.22646354987451117,
|
| 693 |
+
"grad_norm": 0.043399691581726074,
|
| 694 |
+
"learning_rate": 0.00034417639429312584,
|
| 695 |
+
"loss": 0.0056,
|
| 696 |
+
"step": 970
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"epoch": 0.22879822564641336,
|
| 700 |
+
"grad_norm": 0.014040129259228706,
|
| 701 |
+
"learning_rate": 0.00034313878080415044,
|
| 702 |
+
"loss": 0.0006,
|
| 703 |
+
"step": 980
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 0.23113290141831552,
|
| 707 |
+
"grad_norm": 0.003576503833755851,
|
| 708 |
+
"learning_rate": 0.0003421011673151751,
|
| 709 |
+
"loss": 0.0007,
|
| 710 |
+
"step": 990
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 0.2334675771902177,
|
| 714 |
+
"grad_norm": 0.0051997085101902485,
|
| 715 |
+
"learning_rate": 0.00034106355382619976,
|
| 716 |
+
"loss": 0.0001,
|
| 717 |
+
"step": 1000
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 0.23580225296211987,
|
| 721 |
+
"grad_norm": 0.0035423666704446077,
|
| 722 |
+
"learning_rate": 0.0003400259403372244,
|
| 723 |
+
"loss": 0.011,
|
| 724 |
+
"step": 1010
|
| 725 |
+
},
|
| 726 |
+
{
|
| 727 |
+
"epoch": 0.23813692873402206,
|
| 728 |
+
"grad_norm": 0.009176980704069138,
|
| 729 |
+
"learning_rate": 0.000338988326848249,
|
| 730 |
+
"loss": 0.0129,
|
| 731 |
+
"step": 1020
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 0.24047160450592425,
|
| 735 |
+
"grad_norm": 1.0268243551254272,
|
| 736 |
+
"learning_rate": 0.00033795071335927367,
|
| 737 |
+
"loss": 0.0128,
|
| 738 |
+
"step": 1030
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"epoch": 0.24280628027782641,
|
| 742 |
+
"grad_norm": 0.15778960287570953,
|
| 743 |
+
"learning_rate": 0.00033691309987029833,
|
| 744 |
+
"loss": 0.0025,
|
| 745 |
+
"step": 1040
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 0.2451409560497286,
|
| 749 |
+
"grad_norm": 0.014147180132567883,
|
| 750 |
+
"learning_rate": 0.000335875486381323,
|
| 751 |
+
"loss": 0.0012,
|
| 752 |
+
"step": 1050
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 0.24747563182163076,
|
| 756 |
+
"grad_norm": 0.6503289341926575,
|
| 757 |
+
"learning_rate": 0.00033483787289234764,
|
| 758 |
+
"loss": 0.0077,
|
| 759 |
+
"step": 1060
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 0.24981030759353295,
|
| 763 |
+
"grad_norm": 0.036625299602746964,
|
| 764 |
+
"learning_rate": 0.0003338002594033723,
|
| 765 |
+
"loss": 0.0066,
|
| 766 |
+
"step": 1070
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"epoch": 0.2521449833654351,
|
| 770 |
+
"grad_norm": 0.05064311996102333,
|
| 771 |
+
"learning_rate": 0.0003327626459143969,
|
| 772 |
+
"loss": 0.0132,
|
| 773 |
+
"step": 1080
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 0.2544796591373373,
|
| 777 |
+
"grad_norm": 0.01040785014629364,
|
| 778 |
+
"learning_rate": 0.00033172503242542156,
|
| 779 |
+
"loss": 0.0022,
|
| 780 |
+
"step": 1090
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 0.2568143349092395,
|
| 784 |
+
"grad_norm": 0.0481790155172348,
|
| 785 |
+
"learning_rate": 0.00033068741893644616,
|
| 786 |
+
"loss": 0.0169,
|
| 787 |
+
"step": 1100
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 0.25914901068114166,
|
| 791 |
+
"grad_norm": 0.04906298220157623,
|
| 792 |
+
"learning_rate": 0.0003296498054474708,
|
| 793 |
+
"loss": 0.0122,
|
| 794 |
+
"step": 1110
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 0.2614836864530438,
|
| 798 |
+
"grad_norm": 0.010882526636123657,
|
| 799 |
+
"learning_rate": 0.0003286121919584955,
|
| 800 |
+
"loss": 0.01,
|
| 801 |
+
"step": 1120
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 0.26381836222494603,
|
| 805 |
+
"grad_norm": 0.05711141228675842,
|
| 806 |
+
"learning_rate": 0.00032757457846952013,
|
| 807 |
+
"loss": 0.0017,
|
| 808 |
+
"step": 1130
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"epoch": 0.2661530379968482,
|
| 812 |
+
"grad_norm": 0.578333854675293,
|
| 813 |
+
"learning_rate": 0.00032653696498054474,
|
| 814 |
+
"loss": 0.0067,
|
| 815 |
+
"step": 1140
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 0.26848771376875036,
|
| 819 |
+
"grad_norm": 0.01890755444765091,
|
| 820 |
+
"learning_rate": 0.0003254993514915694,
|
| 821 |
+
"loss": 0.004,
|
| 822 |
+
"step": 1150
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"epoch": 0.2708223895406525,
|
| 826 |
+
"grad_norm": 0.07548126578330994,
|
| 827 |
+
"learning_rate": 0.00032446173800259405,
|
| 828 |
+
"loss": 0.0018,
|
| 829 |
+
"step": 1160
|
| 830 |
+
},
|
| 831 |
+
{
|
| 832 |
+
"epoch": 0.27315706531255474,
|
| 833 |
+
"grad_norm": 0.13876883685588837,
|
| 834 |
+
"learning_rate": 0.0003234241245136187,
|
| 835 |
+
"loss": 0.0002,
|
| 836 |
+
"step": 1170
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 0.2754917410844569,
|
| 840 |
+
"grad_norm": 0.0030123014003038406,
|
| 841 |
+
"learning_rate": 0.00032238651102464336,
|
| 842 |
+
"loss": 0.0001,
|
| 843 |
+
"step": 1180
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 0.27782641685635906,
|
| 847 |
+
"grad_norm": 0.00535109406337142,
|
| 848 |
+
"learning_rate": 0.00032134889753566797,
|
| 849 |
+
"loss": 0.0072,
|
| 850 |
+
"step": 1190
|
| 851 |
+
},
|
| 852 |
+
{
|
| 853 |
+
"epoch": 0.2801610926282613,
|
| 854 |
+
"grad_norm": 0.05122077465057373,
|
| 855 |
+
"learning_rate": 0.0003203112840466926,
|
| 856 |
+
"loss": 0.0016,
|
| 857 |
+
"step": 1200
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 0.28249576840016344,
|
| 861 |
+
"grad_norm": 0.003601687727496028,
|
| 862 |
+
"learning_rate": 0.0003192736705577173,
|
| 863 |
+
"loss": 0.0003,
|
| 864 |
+
"step": 1210
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 0.2848304441720656,
|
| 868 |
+
"grad_norm": 0.001865709782578051,
|
| 869 |
+
"learning_rate": 0.0003182360570687419,
|
| 870 |
+
"loss": 0.0001,
|
| 871 |
+
"step": 1220
|
| 872 |
+
},
|
| 873 |
+
{
|
| 874 |
+
"epoch": 0.28716511994396776,
|
| 875 |
+
"grad_norm": 0.0040197898633778095,
|
| 876 |
+
"learning_rate": 0.00031719844357976654,
|
| 877 |
+
"loss": 0.0,
|
| 878 |
+
"step": 1230
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 0.28949979571587,
|
| 882 |
+
"grad_norm": 1.5420883893966675,
|
| 883 |
+
"learning_rate": 0.0003161608300907912,
|
| 884 |
+
"loss": 0.0048,
|
| 885 |
+
"step": 1240
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 0.29183447148777214,
|
| 889 |
+
"grad_norm": 0.6053478717803955,
|
| 890 |
+
"learning_rate": 0.00031512321660181585,
|
| 891 |
+
"loss": 0.0023,
|
| 892 |
+
"step": 1250
|
| 893 |
+
},
|
| 894 |
+
{
|
| 895 |
+
"epoch": 0.2941691472596743,
|
| 896 |
+
"grad_norm": 0.008091798983514309,
|
| 897 |
+
"learning_rate": 0.00031408560311284046,
|
| 898 |
+
"loss": 0.0011,
|
| 899 |
+
"step": 1260
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 0.29650382303157646,
|
| 903 |
+
"grad_norm": 0.027423491701483727,
|
| 904 |
+
"learning_rate": 0.0003130479896238651,
|
| 905 |
+
"loss": 0.002,
|
| 906 |
+
"step": 1270
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 0.2988384988034787,
|
| 910 |
+
"grad_norm": 0.008556556887924671,
|
| 911 |
+
"learning_rate": 0.00031201037613488977,
|
| 912 |
+
"loss": 0.0204,
|
| 913 |
+
"step": 1280
|
| 914 |
+
},
|
| 915 |
+
{
|
| 916 |
+
"epoch": 0.30117317457538084,
|
| 917 |
+
"grad_norm": 0.02086860127747059,
|
| 918 |
+
"learning_rate": 0.00031097276264591443,
|
| 919 |
+
"loss": 0.001,
|
| 920 |
+
"step": 1290
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 0.303507850347283,
|
| 924 |
+
"grad_norm": 0.016817396506667137,
|
| 925 |
+
"learning_rate": 0.0003099351491569391,
|
| 926 |
+
"loss": 0.0005,
|
| 927 |
+
"step": 1300
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 0.3058425261191852,
|
| 931 |
+
"grad_norm": 0.038000259548425674,
|
| 932 |
+
"learning_rate": 0.0003088975356679637,
|
| 933 |
+
"loss": 0.0136,
|
| 934 |
+
"step": 1310
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 0.3081772018910874,
|
| 938 |
+
"grad_norm": 0.08027376979589462,
|
| 939 |
+
"learning_rate": 0.00030785992217898834,
|
| 940 |
+
"loss": 0.0015,
|
| 941 |
+
"step": 1320
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 0.31051187766298954,
|
| 945 |
+
"grad_norm": 0.009664513170719147,
|
| 946 |
+
"learning_rate": 0.000306822308690013,
|
| 947 |
+
"loss": 0.0007,
|
| 948 |
+
"step": 1330
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 0.3128465534348917,
|
| 952 |
+
"grad_norm": 0.004226653836667538,
|
| 953 |
+
"learning_rate": 0.0003057846952010376,
|
| 954 |
+
"loss": 0.0002,
|
| 955 |
+
"step": 1340
|
| 956 |
+
},
|
| 957 |
+
{
|
| 958 |
+
"epoch": 0.3151812292067939,
|
| 959 |
+
"grad_norm": 0.0029185679741203785,
|
| 960 |
+
"learning_rate": 0.00030474708171206226,
|
| 961 |
+
"loss": 0.0006,
|
| 962 |
+
"step": 1350
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 0.3175159049786961,
|
| 966 |
+
"grad_norm": 0.13733680546283722,
|
| 967 |
+
"learning_rate": 0.0003037094682230869,
|
| 968 |
+
"loss": 0.0002,
|
| 969 |
+
"step": 1360
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 0.31985058075059825,
|
| 973 |
+
"grad_norm": 0.0015662899240851402,
|
| 974 |
+
"learning_rate": 0.0003026718547341116,
|
| 975 |
+
"loss": 0.0,
|
| 976 |
+
"step": 1370
|
| 977 |
+
},
|
| 978 |
+
{
|
| 979 |
+
"epoch": 0.32218525652250046,
|
| 980 |
+
"grad_norm": 0.009845585562288761,
|
| 981 |
+
"learning_rate": 0.0003016342412451362,
|
| 982 |
+
"loss": 0.0038,
|
| 983 |
+
"step": 1380
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 0.3245199322944026,
|
| 987 |
+
"grad_norm": 0.0025016157887876034,
|
| 988 |
+
"learning_rate": 0.00030059662775616084,
|
| 989 |
+
"loss": 0.0023,
|
| 990 |
+
"step": 1390
|
| 991 |
+
},
|
| 992 |
+
{
|
| 993 |
+
"epoch": 0.3268546080663048,
|
| 994 |
+
"grad_norm": 0.0021448610350489616,
|
| 995 |
+
"learning_rate": 0.0002995590142671855,
|
| 996 |
+
"loss": 0.0014,
|
| 997 |
+
"step": 1400
|
| 998 |
+
},
|
| 999 |
+
{
|
| 1000 |
+
"epoch": 0.32918928383820695,
|
| 1001 |
+
"grad_norm": 0.0016592498868703842,
|
| 1002 |
+
"learning_rate": 0.00029852140077821015,
|
| 1003 |
+
"loss": 0.0031,
|
| 1004 |
+
"step": 1410
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 0.33152395961010916,
|
| 1008 |
+
"grad_norm": 0.0032906723208725452,
|
| 1009 |
+
"learning_rate": 0.0002974837872892348,
|
| 1010 |
+
"loss": 0.0088,
|
| 1011 |
+
"step": 1420
|
| 1012 |
+
},
|
| 1013 |
+
{
|
| 1014 |
+
"epoch": 0.3338586353820113,
|
| 1015 |
+
"grad_norm": 0.007907208986580372,
|
| 1016 |
+
"learning_rate": 0.0002964461738002594,
|
| 1017 |
+
"loss": 0.0045,
|
| 1018 |
+
"step": 1430
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"epoch": 0.3361933111539135,
|
| 1022 |
+
"grad_norm": 0.004995182156562805,
|
| 1023 |
+
"learning_rate": 0.00029540856031128407,
|
| 1024 |
+
"loss": 0.0009,
|
| 1025 |
+
"step": 1440
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 0.3385279869258157,
|
| 1029 |
+
"grad_norm": 0.004081010818481445,
|
| 1030 |
+
"learning_rate": 0.0002943709468223087,
|
| 1031 |
+
"loss": 0.0004,
|
| 1032 |
+
"step": 1450
|
| 1033 |
+
},
|
| 1034 |
+
{
|
| 1035 |
+
"epoch": 0.34086266269771787,
|
| 1036 |
+
"grad_norm": 0.003877257462590933,
|
| 1037 |
+
"learning_rate": 0.0002933333333333333,
|
| 1038 |
+
"loss": 0.0012,
|
| 1039 |
+
"step": 1460
|
| 1040 |
+
},
|
| 1041 |
+
{
|
| 1042 |
+
"epoch": 0.34319733846962003,
|
| 1043 |
+
"grad_norm": 0.002449814695864916,
|
| 1044 |
+
"learning_rate": 0.000292295719844358,
|
| 1045 |
+
"loss": 0.0002,
|
| 1046 |
+
"step": 1470
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 0.3455320142415222,
|
| 1050 |
+
"grad_norm": 0.002753973240032792,
|
| 1051 |
+
"learning_rate": 0.00029125810635538264,
|
| 1052 |
+
"loss": 0.0002,
|
| 1053 |
+
"step": 1480
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 0.3478666900134244,
|
| 1057 |
+
"grad_norm": 0.0017747861566022038,
|
| 1058 |
+
"learning_rate": 0.00029022049286640724,
|
| 1059 |
+
"loss": 0.0001,
|
| 1060 |
+
"step": 1490
|
| 1061 |
+
},
|
| 1062 |
+
{
|
| 1063 |
+
"epoch": 0.35020136578532657,
|
| 1064 |
+
"grad_norm": 0.0011892006732523441,
|
| 1065 |
+
"learning_rate": 0.0002891828793774319,
|
| 1066 |
+
"loss": 0.0004,
|
| 1067 |
+
"step": 1500
|
| 1068 |
+
},
|
| 1069 |
+
{
|
| 1070 |
+
"epoch": 0.35253604155722873,
|
| 1071 |
+
"grad_norm": 0.0011454072082415223,
|
| 1072 |
+
"learning_rate": 0.00028814526588845656,
|
| 1073 |
+
"loss": 0.0,
|
| 1074 |
+
"step": 1510
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 0.3548707173291309,
|
| 1078 |
+
"grad_norm": 0.0009849355556070805,
|
| 1079 |
+
"learning_rate": 0.0002871076523994812,
|
| 1080 |
+
"loss": 0.0,
|
| 1081 |
+
"step": 1520
|
| 1082 |
+
},
|
| 1083 |
+
{
|
| 1084 |
+
"epoch": 0.3572053931010331,
|
| 1085 |
+
"grad_norm": 0.0009979312308132648,
|
| 1086 |
+
"learning_rate": 0.00028607003891050587,
|
| 1087 |
+
"loss": 0.0,
|
| 1088 |
+
"step": 1530
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 0.35954006887293527,
|
| 1092 |
+
"grad_norm": 0.6208717226982117,
|
| 1093 |
+
"learning_rate": 0.00028503242542153053,
|
| 1094 |
+
"loss": 0.0012,
|
| 1095 |
+
"step": 1540
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 0.36187474464483743,
|
| 1099 |
+
"grad_norm": 1.135374665260315,
|
| 1100 |
+
"learning_rate": 0.00028399481193255513,
|
| 1101 |
+
"loss": 0.0008,
|
| 1102 |
+
"step": 1550
|
| 1103 |
+
},
|
| 1104 |
+
{
|
| 1105 |
+
"epoch": 0.36420942041673965,
|
| 1106 |
+
"grad_norm": 0.0006218306953087449,
|
| 1107 |
+
"learning_rate": 0.0002829571984435798,
|
| 1108 |
+
"loss": 0.0001,
|
| 1109 |
+
"step": 1560
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 0.3665440961886418,
|
| 1113 |
+
"grad_norm": 0.002007074421271682,
|
| 1114 |
+
"learning_rate": 0.00028191958495460444,
|
| 1115 |
+
"loss": 0.0013,
|
| 1116 |
+
"step": 1570
|
| 1117 |
+
},
|
| 1118 |
+
{
|
| 1119 |
+
"epoch": 0.36887877196054397,
|
| 1120 |
+
"grad_norm": 0.0011203576577827334,
|
| 1121 |
+
"learning_rate": 0.00028088197146562905,
|
| 1122 |
+
"loss": 0.0051,
|
| 1123 |
+
"step": 1580
|
| 1124 |
+
},
|
| 1125 |
+
{
|
| 1126 |
+
"epoch": 0.37121344773244613,
|
| 1127 |
+
"grad_norm": 0.004436762072145939,
|
| 1128 |
+
"learning_rate": 0.0002798443579766537,
|
| 1129 |
+
"loss": 0.0091,
|
| 1130 |
+
"step": 1590
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 0.37354812350434835,
|
| 1134 |
+
"grad_norm": 0.4702003598213196,
|
| 1135 |
+
"learning_rate": 0.00027880674448767836,
|
| 1136 |
+
"loss": 0.0009,
|
| 1137 |
+
"step": 1600
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 0.3758827992762505,
|
| 1141 |
+
"grad_norm": 0.013339284807443619,
|
| 1142 |
+
"learning_rate": 0.00027776913099870296,
|
| 1143 |
+
"loss": 0.0093,
|
| 1144 |
+
"step": 1610
|
| 1145 |
+
},
|
| 1146 |
+
{
|
| 1147 |
+
"epoch": 0.3782174750481527,
|
| 1148 |
+
"grad_norm": 0.0076319011859595776,
|
| 1149 |
+
"learning_rate": 0.0002767315175097276,
|
| 1150 |
+
"loss": 0.0025,
|
| 1151 |
+
"step": 1620
|
| 1152 |
+
},
|
| 1153 |
+
{
|
| 1154 |
+
"epoch": 0.3805521508200549,
|
| 1155 |
+
"grad_norm": 0.005932167172431946,
|
| 1156 |
+
"learning_rate": 0.0002756939040207523,
|
| 1157 |
+
"loss": 0.0004,
|
| 1158 |
+
"step": 1630
|
| 1159 |
+
},
|
| 1160 |
+
{
|
| 1161 |
+
"epoch": 0.38288682659195705,
|
| 1162 |
+
"grad_norm": 0.003560519078746438,
|
| 1163 |
+
"learning_rate": 0.00027465629053177693,
|
| 1164 |
+
"loss": 0.0003,
|
| 1165 |
+
"step": 1640
|
| 1166 |
+
},
|
| 1167 |
+
{
|
| 1168 |
+
"epoch": 0.3852215023638592,
|
| 1169 |
+
"grad_norm": 0.001825852901674807,
|
| 1170 |
+
"learning_rate": 0.0002736186770428016,
|
| 1171 |
+
"loss": 0.0003,
|
| 1172 |
+
"step": 1650
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 0.3875561781357614,
|
| 1176 |
+
"grad_norm": 0.11422229558229446,
|
| 1177 |
+
"learning_rate": 0.00027258106355382625,
|
| 1178 |
+
"loss": 0.0006,
|
| 1179 |
+
"step": 1660
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 0.3898908539076636,
|
| 1183 |
+
"grad_norm": 0.00336836208589375,
|
| 1184 |
+
"learning_rate": 0.00027154345006485085,
|
| 1185 |
+
"loss": 0.0002,
|
| 1186 |
+
"step": 1670
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 0.39222552967956575,
|
| 1190 |
+
"grad_norm": 0.020830854773521423,
|
| 1191 |
+
"learning_rate": 0.0002705058365758755,
|
| 1192 |
+
"loss": 0.0003,
|
| 1193 |
+
"step": 1680
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 0.3945602054514679,
|
| 1197 |
+
"grad_norm": 0.5437701940536499,
|
| 1198 |
+
"learning_rate": 0.00026946822308690017,
|
| 1199 |
+
"loss": 0.0006,
|
| 1200 |
+
"step": 1690
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"epoch": 0.3968948812233701,
|
| 1204 |
+
"grad_norm": 0.012015492655336857,
|
| 1205 |
+
"learning_rate": 0.00026843060959792477,
|
| 1206 |
+
"loss": 0.0001,
|
| 1207 |
+
"step": 1700
|
| 1208 |
+
},
|
| 1209 |
+
{
|
| 1210 |
+
"epoch": 0.3992295569952723,
|
| 1211 |
+
"grad_norm": 0.0010134581243619323,
|
| 1212 |
+
"learning_rate": 0.0002673929961089494,
|
| 1213 |
+
"loss": 0.0005,
|
| 1214 |
+
"step": 1710
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 0.40156423276717446,
|
| 1218 |
+
"grad_norm": 0.001305864891037345,
|
| 1219 |
+
"learning_rate": 0.0002663553826199741,
|
| 1220 |
+
"loss": 0.0006,
|
| 1221 |
+
"step": 1720
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 0.4038989085390766,
|
| 1225 |
+
"grad_norm": 0.0030054424423724413,
|
| 1226 |
+
"learning_rate": 0.0002653177691309987,
|
| 1227 |
+
"loss": 0.0,
|
| 1228 |
+
"step": 1730
|
| 1229 |
+
},
|
| 1230 |
+
{
|
| 1231 |
+
"epoch": 0.40623358431097883,
|
| 1232 |
+
"grad_norm": 0.0008168119820766151,
|
| 1233 |
+
"learning_rate": 0.00026428015564202334,
|
| 1234 |
+
"loss": 0.005,
|
| 1235 |
+
"step": 1740
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 0.408568260082881,
|
| 1239 |
+
"grad_norm": 0.0005433742771856487,
|
| 1240 |
+
"learning_rate": 0.000263242542153048,
|
| 1241 |
+
"loss": 0.0022,
|
| 1242 |
+
"step": 1750
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"epoch": 0.41090293585478316,
|
| 1246 |
+
"grad_norm": 0.03222297504544258,
|
| 1247 |
+
"learning_rate": 0.00026220492866407266,
|
| 1248 |
+
"loss": 0.0001,
|
| 1249 |
+
"step": 1760
|
| 1250 |
+
},
|
| 1251 |
+
{
|
| 1252 |
+
"epoch": 0.4132376116266853,
|
| 1253 |
+
"grad_norm": 0.001766858738847077,
|
| 1254 |
+
"learning_rate": 0.0002611673151750973,
|
| 1255 |
+
"loss": 0.0001,
|
| 1256 |
+
"step": 1770
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 0.41557228739858754,
|
| 1260 |
+
"grad_norm": 0.00046385781024582684,
|
| 1261 |
+
"learning_rate": 0.0002601297016861219,
|
| 1262 |
+
"loss": 0.0,
|
| 1263 |
+
"step": 1780
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 0.4179069631704897,
|
| 1267 |
+
"grad_norm": 0.00034939011675305665,
|
| 1268 |
+
"learning_rate": 0.00025909208819714657,
|
| 1269 |
+
"loss": 0.0,
|
| 1270 |
+
"step": 1790
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 0.42024163894239186,
|
| 1274 |
+
"grad_norm": 0.0006307600415311754,
|
| 1275 |
+
"learning_rate": 0.00025805447470817123,
|
| 1276 |
+
"loss": 0.0,
|
| 1277 |
+
"step": 1800
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 0.4225763147142941,
|
| 1281 |
+
"grad_norm": 0.0005922391428612173,
|
| 1282 |
+
"learning_rate": 0.0002570168612191959,
|
| 1283 |
+
"loss": 0.0,
|
| 1284 |
+
"step": 1810
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 0.42491099048619624,
|
| 1288 |
+
"grad_norm": 0.00035804559593088925,
|
| 1289 |
+
"learning_rate": 0.0002559792477302205,
|
| 1290 |
+
"loss": 0.0,
|
| 1291 |
+
"step": 1820
|
| 1292 |
+
},
|
| 1293 |
+
{
|
| 1294 |
+
"epoch": 0.4272456662580984,
|
| 1295 |
+
"grad_norm": 0.002265334827825427,
|
| 1296 |
+
"learning_rate": 0.00025494163424124515,
|
| 1297 |
+
"loss": 0.0,
|
| 1298 |
+
"step": 1830
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 0.42958034203000056,
|
| 1302 |
+
"grad_norm": 0.00030151245300658047,
|
| 1303 |
+
"learning_rate": 0.0002539040207522698,
|
| 1304 |
+
"loss": 0.0001,
|
| 1305 |
+
"step": 1840
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 0.4319150178019028,
|
| 1309 |
+
"grad_norm": 0.0007045480306260288,
|
| 1310 |
+
"learning_rate": 0.0002528664072632944,
|
| 1311 |
+
"loss": 0.0,
|
| 1312 |
+
"step": 1850
|
| 1313 |
+
},
|
| 1314 |
+
{
|
| 1315 |
+
"epoch": 0.43424969357380494,
|
| 1316 |
+
"grad_norm": 0.0061547341756522655,
|
| 1317 |
+
"learning_rate": 0.00025182879377431906,
|
| 1318 |
+
"loss": 0.0,
|
| 1319 |
+
"step": 1860
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 0.4365843693457071,
|
| 1323 |
+
"grad_norm": 0.0003125610819552094,
|
| 1324 |
+
"learning_rate": 0.0002507911802853437,
|
| 1325 |
+
"loss": 0.0,
|
| 1326 |
+
"step": 1870
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"epoch": 0.4389190451176093,
|
| 1330 |
+
"grad_norm": 0.0002632684481795877,
|
| 1331 |
+
"learning_rate": 0.0002497535667963684,
|
| 1332 |
+
"loss": 0.0,
|
| 1333 |
+
"step": 1880
|
| 1334 |
+
},
|
| 1335 |
+
{
|
| 1336 |
+
"epoch": 0.4412537208895115,
|
| 1337 |
+
"grad_norm": 0.0003021568991243839,
|
| 1338 |
+
"learning_rate": 0.00024871595330739303,
|
| 1339 |
+
"loss": 0.0,
|
| 1340 |
+
"step": 1890
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 0.44358839666141364,
|
| 1344 |
+
"grad_norm": 0.0002902685955632478,
|
| 1345 |
+
"learning_rate": 0.00024767833981841764,
|
| 1346 |
+
"loss": 0.0,
|
| 1347 |
+
"step": 1900
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 0.4459230724333158,
|
| 1351 |
+
"grad_norm": 0.000261983135715127,
|
| 1352 |
+
"learning_rate": 0.0002466407263294423,
|
| 1353 |
+
"loss": 0.0001,
|
| 1354 |
+
"step": 1910
|
| 1355 |
+
},
|
| 1356 |
+
{
|
| 1357 |
+
"epoch": 0.448257748205218,
|
| 1358 |
+
"grad_norm": 0.00039503935840912163,
|
| 1359 |
+
"learning_rate": 0.00024560311284046695,
|
| 1360 |
+
"loss": 0.0,
|
| 1361 |
+
"step": 1920
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 0.4505924239771202,
|
| 1365 |
+
"grad_norm": 0.00022091201390139759,
|
| 1366 |
+
"learning_rate": 0.0002445654993514916,
|
| 1367 |
+
"loss": 0.0,
|
| 1368 |
+
"step": 1930
|
| 1369 |
+
},
|
| 1370 |
+
{
|
| 1371 |
+
"epoch": 0.45292709974902234,
|
| 1372 |
+
"grad_norm": 2.493230104446411,
|
| 1373 |
+
"learning_rate": 0.0002435278858625162,
|
| 1374 |
+
"loss": 0.0371,
|
| 1375 |
+
"step": 1940
|
| 1376 |
+
},
|
| 1377 |
+
{
|
| 1378 |
+
"epoch": 0.4552617755209245,
|
| 1379 |
+
"grad_norm": 0.07945345342159271,
|
| 1380 |
+
"learning_rate": 0.00024249027237354084,
|
| 1381 |
+
"loss": 0.0034,
|
| 1382 |
+
"step": 1950
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 0.4575964512928267,
|
| 1386 |
+
"grad_norm": 0.0050026909448206425,
|
| 1387 |
+
"learning_rate": 0.0002414526588845655,
|
| 1388 |
+
"loss": 0.0043,
|
| 1389 |
+
"step": 1960
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 0.4599311270647289,
|
| 1393 |
+
"grad_norm": 0.006855088286101818,
|
| 1394 |
+
"learning_rate": 0.00024041504539559015,
|
| 1395 |
+
"loss": 0.0065,
|
| 1396 |
+
"step": 1970
|
| 1397 |
+
},
|
| 1398 |
+
{
|
| 1399 |
+
"epoch": 0.46226580283663105,
|
| 1400 |
+
"grad_norm": 0.007564285770058632,
|
| 1401 |
+
"learning_rate": 0.00023937743190661478,
|
| 1402 |
+
"loss": 0.025,
|
| 1403 |
+
"step": 1980
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 0.46460047860853326,
|
| 1407 |
+
"grad_norm": 0.13361288607120514,
|
| 1408 |
+
"learning_rate": 0.00023833981841763944,
|
| 1409 |
+
"loss": 0.0058,
|
| 1410 |
+
"step": 1990
|
| 1411 |
+
},
|
| 1412 |
+
{
|
| 1413 |
+
"epoch": 0.4669351543804354,
|
| 1414 |
+
"grad_norm": 0.011933702044188976,
|
| 1415 |
+
"learning_rate": 0.0002373022049286641,
|
| 1416 |
+
"loss": 0.0034,
|
| 1417 |
+
"step": 2000
|
| 1418 |
+
},
|
| 1419 |
+
{
|
| 1420 |
+
"epoch": 0.4692698301523376,
|
| 1421 |
+
"grad_norm": 0.0047375899739563465,
|
| 1422 |
+
"learning_rate": 0.00023626459143968873,
|
| 1423 |
+
"loss": 0.0051,
|
| 1424 |
+
"step": 2010
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 0.47160450592423975,
|
| 1428 |
+
"grad_norm": 0.012455107644200325,
|
| 1429 |
+
"learning_rate": 0.00023522697795071338,
|
| 1430 |
+
"loss": 0.0001,
|
| 1431 |
+
"step": 2020
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"epoch": 0.47393918169614196,
|
| 1435 |
+
"grad_norm": 0.003286924911662936,
|
| 1436 |
+
"learning_rate": 0.00023418936446173801,
|
| 1437 |
+
"loss": 0.0026,
|
| 1438 |
+
"step": 2030
|
| 1439 |
+
},
|
| 1440 |
+
{
|
| 1441 |
+
"epoch": 0.4762738574680441,
|
| 1442 |
+
"grad_norm": 0.004305190406739712,
|
| 1443 |
+
"learning_rate": 0.00023315175097276267,
|
| 1444 |
+
"loss": 0.0034,
|
| 1445 |
+
"step": 2040
|
| 1446 |
+
},
|
| 1447 |
+
{
|
| 1448 |
+
"epoch": 0.4786085332399463,
|
| 1449 |
+
"grad_norm": 0.003077031811699271,
|
| 1450 |
+
"learning_rate": 0.00023211413748378733,
|
| 1451 |
+
"loss": 0.0007,
|
| 1452 |
+
"step": 2050
|
| 1453 |
+
},
|
| 1454 |
+
{
|
| 1455 |
+
"epoch": 0.4809432090118485,
|
| 1456 |
+
"grad_norm": 0.003154961857944727,
|
| 1457 |
+
"learning_rate": 0.00023107652399481193,
|
| 1458 |
+
"loss": 0.0003,
|
| 1459 |
+
"step": 2060
|
| 1460 |
+
},
|
| 1461 |
+
{
|
| 1462 |
+
"epoch": 0.48327788478375067,
|
| 1463 |
+
"grad_norm": 0.0026587171014398336,
|
| 1464 |
+
"learning_rate": 0.00023003891050583656,
|
| 1465 |
+
"loss": 0.0001,
|
| 1466 |
+
"step": 2070
|
| 1467 |
+
},
|
| 1468 |
+
{
|
| 1469 |
+
"epoch": 0.48561256055565283,
|
| 1470 |
+
"grad_norm": 0.018204033374786377,
|
| 1471 |
+
"learning_rate": 0.00022900129701686122,
|
| 1472 |
+
"loss": 0.0002,
|
| 1473 |
+
"step": 2080
|
| 1474 |
+
},
|
| 1475 |
+
{
|
| 1476 |
+
"epoch": 0.487947236327555,
|
| 1477 |
+
"grad_norm": 0.013659532181918621,
|
| 1478 |
+
"learning_rate": 0.00022796368352788588,
|
| 1479 |
+
"loss": 0.0001,
|
| 1480 |
+
"step": 2090
|
| 1481 |
+
},
|
| 1482 |
+
{
|
| 1483 |
+
"epoch": 0.4902819120994572,
|
| 1484 |
+
"grad_norm": 0.0016216342337429523,
|
| 1485 |
+
"learning_rate": 0.0002269260700389105,
|
| 1486 |
+
"loss": 0.0092,
|
| 1487 |
+
"step": 2100
|
| 1488 |
+
},
|
| 1489 |
+
{
|
| 1490 |
+
"epoch": 0.49261658787135937,
|
| 1491 |
+
"grad_norm": 0.13890917599201202,
|
| 1492 |
+
"learning_rate": 0.00022588845654993516,
|
| 1493 |
+
"loss": 0.0006,
|
| 1494 |
+
"step": 2110
|
| 1495 |
+
},
|
| 1496 |
+
{
|
| 1497 |
+
"epoch": 0.49495126364326153,
|
| 1498 |
+
"grad_norm": 0.004591196309775114,
|
| 1499 |
+
"learning_rate": 0.0002248508430609598,
|
| 1500 |
+
"loss": 0.0014,
|
| 1501 |
+
"step": 2120
|
| 1502 |
+
},
|
| 1503 |
+
{
|
| 1504 |
+
"epoch": 0.49728593941516375,
|
| 1505 |
+
"grad_norm": 0.0019425478531047702,
|
| 1506 |
+
"learning_rate": 0.00022381322957198445,
|
| 1507 |
+
"loss": 0.0001,
|
| 1508 |
+
"step": 2130
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"epoch": 0.4996206151870659,
|
| 1512 |
+
"grad_norm": 0.016683168709278107,
|
| 1513 |
+
"learning_rate": 0.0002227756160830091,
|
| 1514 |
+
"loss": 0.0042,
|
| 1515 |
+
"step": 2140
|
| 1516 |
+
},
|
| 1517 |
+
{
|
| 1518 |
+
"epoch": 0.5019552909589681,
|
| 1519 |
+
"grad_norm": 0.0052938396111130714,
|
| 1520 |
+
"learning_rate": 0.00022173800259403374,
|
| 1521 |
+
"loss": 0.0005,
|
| 1522 |
+
"step": 2150
|
| 1523 |
+
},
|
| 1524 |
+
{
|
| 1525 |
+
"epoch": 0.5042899667308702,
|
| 1526 |
+
"grad_norm": 0.0712481215596199,
|
| 1527 |
+
"learning_rate": 0.0002207003891050584,
|
| 1528 |
+
"loss": 0.0003,
|
| 1529 |
+
"step": 2160
|
| 1530 |
+
},
|
| 1531 |
+
{
|
| 1532 |
+
"epoch": 0.5066246425027724,
|
| 1533 |
+
"grad_norm": 0.002925017150118947,
|
| 1534 |
+
"learning_rate": 0.00021966277561608305,
|
| 1535 |
+
"loss": 0.005,
|
| 1536 |
+
"step": 2170
|
| 1537 |
+
},
|
| 1538 |
+
{
|
| 1539 |
+
"epoch": 0.5089593182746746,
|
| 1540 |
+
"grad_norm": 0.06688928604125977,
|
| 1541 |
+
"learning_rate": 0.00021862516212710765,
|
| 1542 |
+
"loss": 0.0108,
|
| 1543 |
+
"step": 2180
|
| 1544 |
+
},
|
| 1545 |
+
{
|
| 1546 |
+
"epoch": 0.5112939940465768,
|
| 1547 |
+
"grad_norm": 0.007370030973106623,
|
| 1548 |
+
"learning_rate": 0.00021758754863813228,
|
| 1549 |
+
"loss": 0.0005,
|
| 1550 |
+
"step": 2190
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 0.513628669818479,
|
| 1554 |
+
"grad_norm": 0.006824078969657421,
|
| 1555 |
+
"learning_rate": 0.00021654993514915694,
|
| 1556 |
+
"loss": 0.0002,
|
| 1557 |
+
"step": 2200
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 0.5159633455903812,
|
| 1561 |
+
"grad_norm": 0.004564494825899601,
|
| 1562 |
+
"learning_rate": 0.00021551232166018157,
|
| 1563 |
+
"loss": 0.0001,
|
| 1564 |
+
"step": 2210
|
| 1565 |
+
},
|
| 1566 |
+
{
|
| 1567 |
+
"epoch": 0.5182980213622833,
|
| 1568 |
+
"grad_norm": 0.0024187033995985985,
|
| 1569 |
+
"learning_rate": 0.00021447470817120623,
|
| 1570 |
+
"loss": 0.0003,
|
| 1571 |
+
"step": 2220
|
| 1572 |
+
},
|
| 1573 |
+
{
|
| 1574 |
+
"epoch": 0.5206326971341855,
|
| 1575 |
+
"grad_norm": 0.0019623206462711096,
|
| 1576 |
+
"learning_rate": 0.00021343709468223088,
|
| 1577 |
+
"loss": 0.0004,
|
| 1578 |
+
"step": 2230
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 0.5229673729060876,
|
| 1582 |
+
"grad_norm": 0.006479791831225157,
|
| 1583 |
+
"learning_rate": 0.0002123994811932555,
|
| 1584 |
+
"loss": 0.0009,
|
| 1585 |
+
"step": 2240
|
| 1586 |
+
},
|
| 1587 |
+
{
|
| 1588 |
+
"epoch": 0.5253020486779898,
|
| 1589 |
+
"grad_norm": 0.001472059520892799,
|
| 1590 |
+
"learning_rate": 0.00021136186770428017,
|
| 1591 |
+
"loss": 0.0022,
|
| 1592 |
+
"step": 2250
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 0.5276367244498921,
|
| 1596 |
+
"grad_norm": 0.0011262299958616495,
|
| 1597 |
+
"learning_rate": 0.00021032425421530483,
|
| 1598 |
+
"loss": 0.0,
|
| 1599 |
+
"step": 2260
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 0.5299714002217942,
|
| 1603 |
+
"grad_norm": 0.0010004049399867654,
|
| 1604 |
+
"learning_rate": 0.00020928664072632946,
|
| 1605 |
+
"loss": 0.0034,
|
| 1606 |
+
"step": 2270
|
| 1607 |
+
},
|
| 1608 |
+
{
|
| 1609 |
+
"epoch": 0.5323060759936964,
|
| 1610 |
+
"grad_norm": 0.010012038052082062,
|
| 1611 |
+
"learning_rate": 0.00020824902723735411,
|
| 1612 |
+
"loss": 0.0003,
|
| 1613 |
+
"step": 2280
|
| 1614 |
+
},
|
| 1615 |
+
{
|
| 1616 |
+
"epoch": 0.5346407517655986,
|
| 1617 |
+
"grad_norm": 0.002231718273833394,
|
| 1618 |
+
"learning_rate": 0.00020721141374837877,
|
| 1619 |
+
"loss": 0.0003,
|
| 1620 |
+
"step": 2290
|
| 1621 |
+
},
|
| 1622 |
+
{
|
| 1623 |
+
"epoch": 0.5369754275375007,
|
| 1624 |
+
"grad_norm": 0.0018004688899964094,
|
| 1625 |
+
"learning_rate": 0.00020617380025940337,
|
| 1626 |
+
"loss": 0.0001,
|
| 1627 |
+
"step": 2300
|
| 1628 |
+
},
|
| 1629 |
+
{
|
| 1630 |
+
"epoch": 0.5393101033094029,
|
| 1631 |
+
"grad_norm": 0.0008523733704350889,
|
| 1632 |
+
"learning_rate": 0.000205136186770428,
|
| 1633 |
+
"loss": 0.0001,
|
| 1634 |
+
"step": 2310
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 0.541644779081305,
|
| 1638 |
+
"grad_norm": 0.0011259455932304263,
|
| 1639 |
+
"learning_rate": 0.00020409857328145266,
|
| 1640 |
+
"loss": 0.0,
|
| 1641 |
+
"step": 2320
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 0.5439794548532073,
|
| 1645 |
+
"grad_norm": 0.0006843574810773134,
|
| 1646 |
+
"learning_rate": 0.0002030609597924773,
|
| 1647 |
+
"loss": 0.0,
|
| 1648 |
+
"step": 2330
|
| 1649 |
+
},
|
| 1650 |
+
{
|
| 1651 |
+
"epoch": 0.5463141306251095,
|
| 1652 |
+
"grad_norm": 0.0007879248005338013,
|
| 1653 |
+
"learning_rate": 0.00020202334630350195,
|
| 1654 |
+
"loss": 0.0,
|
| 1655 |
+
"step": 2340
|
| 1656 |
+
},
|
| 1657 |
+
{
|
| 1658 |
+
"epoch": 0.5486488063970116,
|
| 1659 |
+
"grad_norm": 0.0011782748624682426,
|
| 1660 |
+
"learning_rate": 0.0002009857328145266,
|
| 1661 |
+
"loss": 0.0001,
|
| 1662 |
+
"step": 2350
|
| 1663 |
+
},
|
| 1664 |
+
{
|
| 1665 |
+
"epoch": 0.5509834821689138,
|
| 1666 |
+
"grad_norm": 0.004246284253895283,
|
| 1667 |
+
"learning_rate": 0.00019994811932555123,
|
| 1668 |
+
"loss": 0.0163,
|
| 1669 |
+
"step": 2360
|
| 1670 |
+
},
|
| 1671 |
+
{
|
| 1672 |
+
"epoch": 0.553318157940816,
|
| 1673 |
+
"grad_norm": 0.7899481058120728,
|
| 1674 |
+
"learning_rate": 0.0001989105058365759,
|
| 1675 |
+
"loss": 0.0054,
|
| 1676 |
+
"step": 2370
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"epoch": 0.5556528337127181,
|
| 1680 |
+
"grad_norm": 0.0048600370064377785,
|
| 1681 |
+
"learning_rate": 0.00019787289234760055,
|
| 1682 |
+
"loss": 0.0001,
|
| 1683 |
+
"step": 2380
|
| 1684 |
+
},
|
| 1685 |
+
{
|
| 1686 |
+
"epoch": 0.5579875094846203,
|
| 1687 |
+
"grad_norm": 0.2455766797065735,
|
| 1688 |
+
"learning_rate": 0.00019683527885862515,
|
| 1689 |
+
"loss": 0.0009,
|
| 1690 |
+
"step": 2390
|
| 1691 |
+
},
|
| 1692 |
+
{
|
| 1693 |
+
"epoch": 0.5603221852565226,
|
| 1694 |
+
"grad_norm": 0.004527187906205654,
|
| 1695 |
+
"learning_rate": 0.0001957976653696498,
|
| 1696 |
+
"loss": 0.0005,
|
| 1697 |
+
"step": 2400
|
| 1698 |
+
},
|
| 1699 |
+
{
|
| 1700 |
+
"epoch": 0.5626568610284247,
|
| 1701 |
+
"grad_norm": 0.003127218456938863,
|
| 1702 |
+
"learning_rate": 0.00019476005188067446,
|
| 1703 |
+
"loss": 0.0001,
|
| 1704 |
+
"step": 2410
|
| 1705 |
+
},
|
| 1706 |
+
{
|
| 1707 |
+
"epoch": 0.5649915368003269,
|
| 1708 |
+
"grad_norm": 0.0033744657412171364,
|
| 1709 |
+
"learning_rate": 0.0001937224383916991,
|
| 1710 |
+
"loss": 0.0008,
|
| 1711 |
+
"step": 2420
|
| 1712 |
+
},
|
| 1713 |
+
{
|
| 1714 |
+
"epoch": 0.567326212572229,
|
| 1715 |
+
"grad_norm": 0.0021291917655617,
|
| 1716 |
+
"learning_rate": 0.00019268482490272375,
|
| 1717 |
+
"loss": 0.0001,
|
| 1718 |
+
"step": 2430
|
| 1719 |
+
},
|
| 1720 |
+
{
|
| 1721 |
+
"epoch": 0.5696608883441312,
|
| 1722 |
+
"grad_norm": 0.002303266664966941,
|
| 1723 |
+
"learning_rate": 0.0001916472114137484,
|
| 1724 |
+
"loss": 0.0001,
|
| 1725 |
+
"step": 2440
|
| 1726 |
+
},
|
| 1727 |
+
{
|
| 1728 |
+
"epoch": 0.5719955641160334,
|
| 1729 |
+
"grad_norm": 0.5721760392189026,
|
| 1730 |
+
"learning_rate": 0.000190609597924773,
|
| 1731 |
+
"loss": 0.011,
|
| 1732 |
+
"step": 2450
|
| 1733 |
+
},
|
| 1734 |
+
{
|
| 1735 |
+
"epoch": 0.5743302398879355,
|
| 1736 |
+
"grad_norm": 1.1442689895629883,
|
| 1737 |
+
"learning_rate": 0.00018957198443579767,
|
| 1738 |
+
"loss": 0.0065,
|
| 1739 |
+
"step": 2460
|
| 1740 |
+
},
|
| 1741 |
+
{
|
| 1742 |
+
"epoch": 0.5766649156598377,
|
| 1743 |
+
"grad_norm": 0.03165394440293312,
|
| 1744 |
+
"learning_rate": 0.00018853437094682233,
|
| 1745 |
+
"loss": 0.001,
|
| 1746 |
+
"step": 2470
|
| 1747 |
+
},
|
| 1748 |
+
{
|
| 1749 |
+
"epoch": 0.57899959143174,
|
| 1750 |
+
"grad_norm": 0.007602803409099579,
|
| 1751 |
+
"learning_rate": 0.00018749675745784696,
|
| 1752 |
+
"loss": 0.0132,
|
| 1753 |
+
"step": 2480
|
| 1754 |
+
},
|
| 1755 |
+
{
|
| 1756 |
+
"epoch": 0.5813342672036421,
|
| 1757 |
+
"grad_norm": 0.026837633922696114,
|
| 1758 |
+
"learning_rate": 0.0001864591439688716,
|
| 1759 |
+
"loss": 0.001,
|
| 1760 |
+
"step": 2490
|
| 1761 |
+
},
|
| 1762 |
+
{
|
| 1763 |
+
"epoch": 0.5836689429755443,
|
| 1764 |
+
"grad_norm": 0.024656491354107857,
|
| 1765 |
+
"learning_rate": 0.00018542153047989624,
|
| 1766 |
+
"loss": 0.0018,
|
| 1767 |
+
"step": 2500
|
| 1768 |
+
},
|
| 1769 |
+
{
|
| 1770 |
+
"epoch": 0.5860036187474464,
|
| 1771 |
+
"grad_norm": 0.011152198538184166,
|
| 1772 |
+
"learning_rate": 0.0001843839169909209,
|
| 1773 |
+
"loss": 0.0004,
|
| 1774 |
+
"step": 2510
|
| 1775 |
+
},
|
| 1776 |
+
{
|
| 1777 |
+
"epoch": 0.5883382945193486,
|
| 1778 |
+
"grad_norm": 0.010260018520057201,
|
| 1779 |
+
"learning_rate": 0.00018334630350194553,
|
| 1780 |
+
"loss": 0.0002,
|
| 1781 |
+
"step": 2520
|
| 1782 |
+
},
|
| 1783 |
+
{
|
| 1784 |
+
"epoch": 0.5906729702912508,
|
| 1785 |
+
"grad_norm": 0.003675712738186121,
|
| 1786 |
+
"learning_rate": 0.00018230869001297019,
|
| 1787 |
+
"loss": 0.0025,
|
| 1788 |
+
"step": 2530
|
| 1789 |
+
},
|
| 1790 |
+
{
|
| 1791 |
+
"epoch": 0.5930076460631529,
|
| 1792 |
+
"grad_norm": 0.00440176110714674,
|
| 1793 |
+
"learning_rate": 0.00018127107652399482,
|
| 1794 |
+
"loss": 0.0001,
|
| 1795 |
+
"step": 2540
|
| 1796 |
+
},
|
| 1797 |
+
{
|
| 1798 |
+
"epoch": 0.5953423218350552,
|
| 1799 |
+
"grad_norm": 0.004901622422039509,
|
| 1800 |
+
"learning_rate": 0.00018023346303501947,
|
| 1801 |
+
"loss": 0.0019,
|
| 1802 |
+
"step": 2550
|
| 1803 |
+
},
|
| 1804 |
+
{
|
| 1805 |
+
"epoch": 0.5976769976069574,
|
| 1806 |
+
"grad_norm": 0.0028110845014452934,
|
| 1807 |
+
"learning_rate": 0.0001791958495460441,
|
| 1808 |
+
"loss": 0.0009,
|
| 1809 |
+
"step": 2560
|
| 1810 |
+
},
|
| 1811 |
+
{
|
| 1812 |
+
"epoch": 0.6000116733788595,
|
| 1813 |
+
"grad_norm": 0.0025059175677597523,
|
| 1814 |
+
"learning_rate": 0.00017815823605706876,
|
| 1815 |
+
"loss": 0.0,
|
| 1816 |
+
"step": 2570
|
| 1817 |
+
},
|
| 1818 |
+
{
|
| 1819 |
+
"epoch": 0.6023463491507617,
|
| 1820 |
+
"grad_norm": 0.0019852565601468086,
|
| 1821 |
+
"learning_rate": 0.0001771206225680934,
|
| 1822 |
+
"loss": 0.0,
|
| 1823 |
+
"step": 2580
|
| 1824 |
+
},
|
| 1825 |
+
{
|
| 1826 |
+
"epoch": 0.6046810249226638,
|
| 1827 |
+
"grad_norm": 0.0014332541031762958,
|
| 1828 |
+
"learning_rate": 0.00017608300907911802,
|
| 1829 |
+
"loss": 0.0004,
|
| 1830 |
+
"step": 2590
|
| 1831 |
+
},
|
| 1832 |
+
{
|
| 1833 |
+
"epoch": 0.607015700694566,
|
| 1834 |
+
"grad_norm": 0.32902491092681885,
|
| 1835 |
+
"learning_rate": 0.00017504539559014268,
|
| 1836 |
+
"loss": 0.0005,
|
| 1837 |
+
"step": 2600
|
| 1838 |
+
},
|
| 1839 |
+
{
|
| 1840 |
+
"epoch": 0.6093503764664682,
|
| 1841 |
+
"grad_norm": 0.0036561412271112204,
|
| 1842 |
+
"learning_rate": 0.00017400778210116733,
|
| 1843 |
+
"loss": 0.0011,
|
| 1844 |
+
"step": 2610
|
| 1845 |
+
},
|
| 1846 |
+
{
|
| 1847 |
+
"epoch": 0.6116850522383704,
|
| 1848 |
+
"grad_norm": 0.06626530736684799,
|
| 1849 |
+
"learning_rate": 0.00017297016861219196,
|
| 1850 |
+
"loss": 0.0001,
|
| 1851 |
+
"step": 2620
|
| 1852 |
+
},
|
| 1853 |
+
{
|
| 1854 |
+
"epoch": 0.6140197280102726,
|
| 1855 |
+
"grad_norm": 0.0011569494381546974,
|
| 1856 |
+
"learning_rate": 0.00017193255512321662,
|
| 1857 |
+
"loss": 0.0041,
|
| 1858 |
+
"step": 2630
|
| 1859 |
+
},
|
| 1860 |
+
{
|
| 1861 |
+
"epoch": 0.6163544037821748,
|
| 1862 |
+
"grad_norm": 0.002706947736442089,
|
| 1863 |
+
"learning_rate": 0.00017089494163424125,
|
| 1864 |
+
"loss": 0.0111,
|
| 1865 |
+
"step": 2640
|
| 1866 |
+
},
|
| 1867 |
+
{
|
| 1868 |
+
"epoch": 0.6186890795540769,
|
| 1869 |
+
"grad_norm": 0.0033384524285793304,
|
| 1870 |
+
"learning_rate": 0.00016985732814526588,
|
| 1871 |
+
"loss": 0.0003,
|
| 1872 |
+
"step": 2650
|
| 1873 |
+
},
|
| 1874 |
+
{
|
| 1875 |
+
"epoch": 0.6210237553259791,
|
| 1876 |
+
"grad_norm": 0.0037459495943039656,
|
| 1877 |
+
"learning_rate": 0.00016881971465629054,
|
| 1878 |
+
"loss": 0.0001,
|
| 1879 |
+
"step": 2660
|
| 1880 |
+
},
|
| 1881 |
+
{
|
| 1882 |
+
"epoch": 0.6233584310978812,
|
| 1883 |
+
"grad_norm": 0.0036509244237095118,
|
| 1884 |
+
"learning_rate": 0.0001677821011673152,
|
| 1885 |
+
"loss": 0.0001,
|
| 1886 |
+
"step": 2670
|
| 1887 |
+
},
|
| 1888 |
+
{
|
| 1889 |
+
"epoch": 0.6256931068697834,
|
| 1890 |
+
"grad_norm": 0.004108482040464878,
|
| 1891 |
+
"learning_rate": 0.00016674448767833982,
|
| 1892 |
+
"loss": 0.0002,
|
| 1893 |
+
"step": 2680
|
| 1894 |
+
},
|
| 1895 |
+
{
|
| 1896 |
+
"epoch": 0.6280277826416857,
|
| 1897 |
+
"grad_norm": 0.003080847905948758,
|
| 1898 |
+
"learning_rate": 0.00016570687418936448,
|
| 1899 |
+
"loss": 0.0001,
|
| 1900 |
+
"step": 2690
|
| 1901 |
+
},
|
| 1902 |
+
{
|
| 1903 |
+
"epoch": 0.6303624584135878,
|
| 1904 |
+
"grad_norm": 0.0028391852974891663,
|
| 1905 |
+
"learning_rate": 0.0001646692607003891,
|
| 1906 |
+
"loss": 0.0002,
|
| 1907 |
+
"step": 2700
|
| 1908 |
+
},
|
| 1909 |
+
{
|
| 1910 |
+
"epoch": 0.63269713418549,
|
| 1911 |
+
"grad_norm": 0.001830106251873076,
|
| 1912 |
+
"learning_rate": 0.00016363164721141374,
|
| 1913 |
+
"loss": 0.0004,
|
| 1914 |
+
"step": 2710
|
| 1915 |
+
},
|
| 1916 |
+
{
|
| 1917 |
+
"epoch": 0.6350318099573922,
|
| 1918 |
+
"grad_norm": 0.0024860044941306114,
|
| 1919 |
+
"learning_rate": 0.0001625940337224384,
|
| 1920 |
+
"loss": 0.0001,
|
| 1921 |
+
"step": 2720
|
| 1922 |
+
},
|
| 1923 |
+
{
|
| 1924 |
+
"epoch": 0.6373664857292943,
|
| 1925 |
+
"grad_norm": 0.0016177381621673703,
|
| 1926 |
+
"learning_rate": 0.00016155642023346305,
|
| 1927 |
+
"loss": 0.0003,
|
| 1928 |
+
"step": 2730
|
| 1929 |
+
},
|
| 1930 |
+
{
|
| 1931 |
+
"epoch": 0.6397011615011965,
|
| 1932 |
+
"grad_norm": 0.0024260838981717825,
|
| 1933 |
+
"learning_rate": 0.00016051880674448768,
|
| 1934 |
+
"loss": 0.0004,
|
| 1935 |
+
"step": 2740
|
| 1936 |
+
},
|
| 1937 |
+
{
|
| 1938 |
+
"epoch": 0.6420358372730987,
|
| 1939 |
+
"grad_norm": 0.038342151790857315,
|
| 1940 |
+
"learning_rate": 0.00015948119325551234,
|
| 1941 |
+
"loss": 0.0003,
|
| 1942 |
+
"step": 2750
|
| 1943 |
+
},
|
| 1944 |
+
{
|
| 1945 |
+
"epoch": 0.6443705130450009,
|
| 1946 |
+
"grad_norm": 0.0013671324122697115,
|
| 1947 |
+
"learning_rate": 0.00015844357976653697,
|
| 1948 |
+
"loss": 0.0055,
|
| 1949 |
+
"step": 2760
|
| 1950 |
+
},
|
| 1951 |
+
{
|
| 1952 |
+
"epoch": 0.6467051888169031,
|
| 1953 |
+
"grad_norm": 0.0012879414716735482,
|
| 1954 |
+
"learning_rate": 0.0001574059662775616,
|
| 1955 |
+
"loss": 0.0,
|
| 1956 |
+
"step": 2770
|
| 1957 |
+
},
|
| 1958 |
+
{
|
| 1959 |
+
"epoch": 0.6490398645888052,
|
| 1960 |
+
"grad_norm": 0.0014536501839756966,
|
| 1961 |
+
"learning_rate": 0.00015636835278858626,
|
| 1962 |
+
"loss": 0.0,
|
| 1963 |
+
"step": 2780
|
| 1964 |
+
},
|
| 1965 |
+
{
|
| 1966 |
+
"epoch": 0.6513745403607074,
|
| 1967 |
+
"grad_norm": 0.8653482794761658,
|
| 1968 |
+
"learning_rate": 0.00015533073929961092,
|
| 1969 |
+
"loss": 0.002,
|
| 1970 |
+
"step": 2790
|
| 1971 |
+
},
|
| 1972 |
+
{
|
| 1973 |
+
"epoch": 0.6537092161326096,
|
| 1974 |
+
"grad_norm": 0.0016417702427133918,
|
| 1975 |
+
"learning_rate": 0.00015429312581063555,
|
| 1976 |
+
"loss": 0.0003,
|
| 1977 |
+
"step": 2800
|
| 1978 |
+
},
|
| 1979 |
+
{
|
| 1980 |
+
"epoch": 0.6560438919045117,
|
| 1981 |
+
"grad_norm": 0.07089340686798096,
|
| 1982 |
+
"learning_rate": 0.0001532555123216602,
|
| 1983 |
+
"loss": 0.0131,
|
| 1984 |
+
"step": 2810
|
| 1985 |
+
},
|
| 1986 |
+
{
|
| 1987 |
+
"epoch": 0.6583785676764139,
|
| 1988 |
+
"grad_norm": 0.0033747514244168997,
|
| 1989 |
+
"learning_rate": 0.00015221789883268483,
|
| 1990 |
+
"loss": 0.0028,
|
| 1991 |
+
"step": 2820
|
| 1992 |
+
},
|
| 1993 |
+
{
|
| 1994 |
+
"epoch": 0.6607132434483162,
|
| 1995 |
+
"grad_norm": 0.006389171350747347,
|
| 1996 |
+
"learning_rate": 0.00015118028534370946,
|
| 1997 |
+
"loss": 0.0002,
|
| 1998 |
+
"step": 2830
|
| 1999 |
+
},
|
| 2000 |
+
{
|
| 2001 |
+
"epoch": 0.6630479192202183,
|
| 2002 |
+
"grad_norm": 0.0026964943390339613,
|
| 2003 |
+
"learning_rate": 0.00015014267185473412,
|
| 2004 |
+
"loss": 0.0002,
|
| 2005 |
+
"step": 2840
|
| 2006 |
+
},
|
| 2007 |
+
{
|
| 2008 |
+
"epoch": 0.6653825949921205,
|
| 2009 |
+
"grad_norm": 0.002625943860039115,
|
| 2010 |
+
"learning_rate": 0.00014910505836575878,
|
| 2011 |
+
"loss": 0.0004,
|
| 2012 |
+
"step": 2850
|
| 2013 |
+
},
|
| 2014 |
+
{
|
| 2015 |
+
"epoch": 0.6677172707640227,
|
| 2016 |
+
"grad_norm": 0.003076399676501751,
|
| 2017 |
+
"learning_rate": 0.0001480674448767834,
|
| 2018 |
+
"loss": 0.0001,
|
| 2019 |
+
"step": 2860
|
| 2020 |
+
},
|
| 2021 |
+
{
|
| 2022 |
+
"epoch": 0.6700519465359248,
|
| 2023 |
+
"grad_norm": 0.004514554515480995,
|
| 2024 |
+
"learning_rate": 0.00014702983138780806,
|
| 2025 |
+
"loss": 0.0001,
|
| 2026 |
+
"step": 2870
|
| 2027 |
+
},
|
| 2028 |
+
{
|
| 2029 |
+
"epoch": 0.672386622307827,
|
| 2030 |
+
"grad_norm": 0.0014996561221778393,
|
| 2031 |
+
"learning_rate": 0.0001459922178988327,
|
| 2032 |
+
"loss": 0.0002,
|
| 2033 |
+
"step": 2880
|
| 2034 |
+
},
|
| 2035 |
+
{
|
| 2036 |
+
"epoch": 0.6747212980797291,
|
| 2037 |
+
"grad_norm": 0.0015088297659531236,
|
| 2038 |
+
"learning_rate": 0.00014495460440985732,
|
| 2039 |
+
"loss": 0.0003,
|
| 2040 |
+
"step": 2890
|
| 2041 |
+
},
|
| 2042 |
+
{
|
| 2043 |
+
"epoch": 0.6770559738516314,
|
| 2044 |
+
"grad_norm": 0.0016345715848729014,
|
| 2045 |
+
"learning_rate": 0.00014391699092088198,
|
| 2046 |
+
"loss": 0.0,
|
| 2047 |
+
"step": 2900
|
| 2048 |
+
},
|
| 2049 |
+
{
|
| 2050 |
+
"epoch": 0.6793906496235336,
|
| 2051 |
+
"grad_norm": 0.0027825534343719482,
|
| 2052 |
+
"learning_rate": 0.00014287937743190664,
|
| 2053 |
+
"loss": 0.0008,
|
| 2054 |
+
"step": 2910
|
| 2055 |
+
},
|
| 2056 |
+
{
|
| 2057 |
+
"epoch": 0.6817253253954357,
|
| 2058 |
+
"grad_norm": 0.0012849323684349656,
|
| 2059 |
+
"learning_rate": 0.00014184176394293127,
|
| 2060 |
+
"loss": 0.0,
|
| 2061 |
+
"step": 2920
|
| 2062 |
+
},
|
| 2063 |
+
{
|
| 2064 |
+
"epoch": 0.6840600011673379,
|
| 2065 |
+
"grad_norm": 0.0021583992056548595,
|
| 2066 |
+
"learning_rate": 0.00014080415045395592,
|
| 2067 |
+
"loss": 0.0,
|
| 2068 |
+
"step": 2930
|
| 2069 |
+
},
|
| 2070 |
+
{
|
| 2071 |
+
"epoch": 0.6863946769392401,
|
| 2072 |
+
"grad_norm": 0.0012560015311464667,
|
| 2073 |
+
"learning_rate": 0.00013976653696498055,
|
| 2074 |
+
"loss": 0.0001,
|
| 2075 |
+
"step": 2940
|
| 2076 |
+
},
|
| 2077 |
+
{
|
| 2078 |
+
"epoch": 0.6887293527111422,
|
| 2079 |
+
"grad_norm": 0.0009112095576710999,
|
| 2080 |
+
"learning_rate": 0.00013872892347600518,
|
| 2081 |
+
"loss": 0.0001,
|
| 2082 |
+
"step": 2950
|
| 2083 |
+
},
|
| 2084 |
+
{
|
| 2085 |
+
"epoch": 0.6910640284830444,
|
| 2086 |
+
"grad_norm": 0.0013899313053116202,
|
| 2087 |
+
"learning_rate": 0.00013769130998702984,
|
| 2088 |
+
"loss": 0.0,
|
| 2089 |
+
"step": 2960
|
| 2090 |
+
},
|
| 2091 |
+
{
|
| 2092 |
+
"epoch": 0.6933987042549465,
|
| 2093 |
+
"grad_norm": 1.1169312000274658,
|
| 2094 |
+
"learning_rate": 0.0001366536964980545,
|
| 2095 |
+
"loss": 0.005,
|
| 2096 |
+
"step": 2970
|
| 2097 |
+
},
|
| 2098 |
+
{
|
| 2099 |
+
"epoch": 0.6957333800268488,
|
| 2100 |
+
"grad_norm": 0.0009174107108265162,
|
| 2101 |
+
"learning_rate": 0.00013561608300907913,
|
| 2102 |
+
"loss": 0.0009,
|
| 2103 |
+
"step": 2980
|
| 2104 |
+
},
|
| 2105 |
+
{
|
| 2106 |
+
"epoch": 0.698068055798751,
|
| 2107 |
+
"grad_norm": 0.0038010200951248407,
|
| 2108 |
+
"learning_rate": 0.00013457846952010378,
|
| 2109 |
+
"loss": 0.0071,
|
| 2110 |
+
"step": 2990
|
| 2111 |
+
},
|
| 2112 |
+
{
|
| 2113 |
+
"epoch": 0.7004027315706531,
|
| 2114 |
+
"grad_norm": 0.002235995838418603,
|
| 2115 |
+
"learning_rate": 0.00013354085603112841,
|
| 2116 |
+
"loss": 0.0002,
|
| 2117 |
+
"step": 3000
|
| 2118 |
+
},
|
| 2119 |
+
{
|
| 2120 |
+
"epoch": 0.7027374073425553,
|
| 2121 |
+
"grad_norm": 0.039830174297094345,
|
| 2122 |
+
"learning_rate": 0.00013250324254215304,
|
| 2123 |
+
"loss": 0.0004,
|
| 2124 |
+
"step": 3010
|
| 2125 |
+
},
|
| 2126 |
+
{
|
| 2127 |
+
"epoch": 0.7050720831144575,
|
| 2128 |
+
"grad_norm": 0.30222392082214355,
|
| 2129 |
+
"learning_rate": 0.0001314656290531777,
|
| 2130 |
+
"loss": 0.0008,
|
| 2131 |
+
"step": 3020
|
| 2132 |
+
},
|
| 2133 |
+
{
|
| 2134 |
+
"epoch": 0.7074067588863596,
|
| 2135 |
+
"grad_norm": 0.47657474875450134,
|
| 2136 |
+
"learning_rate": 0.00013042801556420233,
|
| 2137 |
+
"loss": 0.0008,
|
| 2138 |
+
"step": 3030
|
| 2139 |
+
},
|
| 2140 |
+
{
|
| 2141 |
+
"epoch": 0.7097414346582618,
|
| 2142 |
+
"grad_norm": 0.0009529945673421025,
|
| 2143 |
+
"learning_rate": 0.000129390402075227,
|
| 2144 |
+
"loss": 0.0,
|
| 2145 |
+
"step": 3040
|
| 2146 |
+
},
|
| 2147 |
+
{
|
| 2148 |
+
"epoch": 0.712076110430164,
|
| 2149 |
+
"grad_norm": 0.00109247793443501,
|
| 2150 |
+
"learning_rate": 0.00012835278858625164,
|
| 2151 |
+
"loss": 0.0004,
|
| 2152 |
+
"step": 3050
|
| 2153 |
+
},
|
| 2154 |
+
{
|
| 2155 |
+
"epoch": 0.7144107862020662,
|
| 2156 |
+
"grad_norm": 0.0016496065072715282,
|
| 2157 |
+
"learning_rate": 0.00012731517509727627,
|
| 2158 |
+
"loss": 0.0014,
|
| 2159 |
+
"step": 3060
|
| 2160 |
+
},
|
| 2161 |
+
{
|
| 2162 |
+
"epoch": 0.7167454619739684,
|
| 2163 |
+
"grad_norm": 0.0007458662148565054,
|
| 2164 |
+
"learning_rate": 0.0001262775616083009,
|
| 2165 |
+
"loss": 0.0,
|
| 2166 |
+
"step": 3070
|
| 2167 |
+
},
|
| 2168 |
+
{
|
| 2169 |
+
"epoch": 0.7190801377458705,
|
| 2170 |
+
"grad_norm": 0.0010477920295670629,
|
| 2171 |
+
"learning_rate": 0.00012523994811932556,
|
| 2172 |
+
"loss": 0.0,
|
| 2173 |
+
"step": 3080
|
| 2174 |
+
},
|
| 2175 |
+
{
|
| 2176 |
+
"epoch": 0.7214148135177727,
|
| 2177 |
+
"grad_norm": 0.0039003838319331408,
|
| 2178 |
+
"learning_rate": 0.0001242023346303502,
|
| 2179 |
+
"loss": 0.0098,
|
| 2180 |
+
"step": 3090
|
| 2181 |
+
},
|
| 2182 |
+
{
|
| 2183 |
+
"epoch": 0.7237494892896749,
|
| 2184 |
+
"grad_norm": 0.6328915953636169,
|
| 2185 |
+
"learning_rate": 0.00012316472114137485,
|
| 2186 |
+
"loss": 0.0048,
|
| 2187 |
+
"step": 3100
|
| 2188 |
+
},
|
| 2189 |
+
{
|
| 2190 |
+
"epoch": 0.726084165061577,
|
| 2191 |
+
"grad_norm": 0.0023845217656344175,
|
| 2192 |
+
"learning_rate": 0.0001221271076523995,
|
| 2193 |
+
"loss": 0.0001,
|
| 2194 |
+
"step": 3110
|
| 2195 |
+
},
|
| 2196 |
+
{
|
| 2197 |
+
"epoch": 0.7284188408334793,
|
| 2198 |
+
"grad_norm": 0.005935149732977152,
|
| 2199 |
+
"learning_rate": 0.00012108949416342412,
|
| 2200 |
+
"loss": 0.0003,
|
| 2201 |
+
"step": 3120
|
| 2202 |
+
},
|
| 2203 |
+
{
|
| 2204 |
+
"epoch": 0.7307535166053815,
|
| 2205 |
+
"grad_norm": 0.002948681591078639,
|
| 2206 |
+
"learning_rate": 0.00012005188067444876,
|
| 2207 |
+
"loss": 0.0002,
|
| 2208 |
+
"step": 3130
|
| 2209 |
+
},
|
| 2210 |
+
{
|
| 2211 |
+
"epoch": 0.7330881923772836,
|
| 2212 |
+
"grad_norm": 1.1137011051177979,
|
| 2213 |
+
"learning_rate": 0.00011901426718547342,
|
| 2214 |
+
"loss": 0.0072,
|
| 2215 |
+
"step": 3140
|
| 2216 |
+
},
|
| 2217 |
+
{
|
| 2218 |
+
"epoch": 0.7354228681491858,
|
| 2219 |
+
"grad_norm": 0.03960300236940384,
|
| 2220 |
+
"learning_rate": 0.00011797665369649807,
|
| 2221 |
+
"loss": 0.0064,
|
| 2222 |
+
"step": 3150
|
| 2223 |
+
},
|
| 2224 |
+
{
|
| 2225 |
+
"epoch": 0.7377575439210879,
|
| 2226 |
+
"grad_norm": 0.004956856369972229,
|
| 2227 |
+
"learning_rate": 0.00011693904020752271,
|
| 2228 |
+
"loss": 0.0048,
|
| 2229 |
+
"step": 3160
|
| 2230 |
+
},
|
| 2231 |
+
{
|
| 2232 |
+
"epoch": 0.7400922196929901,
|
| 2233 |
+
"grad_norm": 0.0028774456586688757,
|
| 2234 |
+
"learning_rate": 0.00011590142671854735,
|
| 2235 |
+
"loss": 0.001,
|
| 2236 |
+
"step": 3170
|
| 2237 |
+
},
|
| 2238 |
+
{
|
| 2239 |
+
"epoch": 0.7424268954648923,
|
| 2240 |
+
"grad_norm": 0.05007918179035187,
|
| 2241 |
+
"learning_rate": 0.00011486381322957198,
|
| 2242 |
+
"loss": 0.0014,
|
| 2243 |
+
"step": 3180
|
| 2244 |
+
},
|
| 2245 |
+
{
|
| 2246 |
+
"epoch": 0.7447615712367945,
|
| 2247 |
+
"grad_norm": 0.004933805204927921,
|
| 2248 |
+
"learning_rate": 0.00011382619974059663,
|
| 2249 |
+
"loss": 0.0001,
|
| 2250 |
+
"step": 3190
|
| 2251 |
+
},
|
| 2252 |
+
{
|
| 2253 |
+
"epoch": 0.7470962470086967,
|
| 2254 |
+
"grad_norm": 0.0028584490064531565,
|
| 2255 |
+
"learning_rate": 0.00011278858625162127,
|
| 2256 |
+
"loss": 0.011,
|
| 2257 |
+
"step": 3200
|
| 2258 |
+
},
|
| 2259 |
+
{
|
| 2260 |
+
"epoch": 0.7494309227805989,
|
| 2261 |
+
"grad_norm": 0.015388348139822483,
|
| 2262 |
+
"learning_rate": 0.00011175097276264593,
|
| 2263 |
+
"loss": 0.0004,
|
| 2264 |
+
"step": 3210
|
| 2265 |
+
},
|
| 2266 |
+
{
|
| 2267 |
+
"epoch": 0.751765598552501,
|
| 2268 |
+
"grad_norm": 0.04148218780755997,
|
| 2269 |
+
"learning_rate": 0.00011071335927367057,
|
| 2270 |
+
"loss": 0.0015,
|
| 2271 |
+
"step": 3220
|
| 2272 |
+
},
|
| 2273 |
+
{
|
| 2274 |
+
"epoch": 0.7541002743244032,
|
| 2275 |
+
"grad_norm": 0.6182008981704712,
|
| 2276 |
+
"learning_rate": 0.00010967574578469521,
|
| 2277 |
+
"loss": 0.0036,
|
| 2278 |
+
"step": 3230
|
| 2279 |
+
},
|
| 2280 |
+
{
|
| 2281 |
+
"epoch": 0.7564349500963053,
|
| 2282 |
+
"grad_norm": 0.0034724888391792774,
|
| 2283 |
+
"learning_rate": 0.00010863813229571984,
|
| 2284 |
+
"loss": 0.0002,
|
| 2285 |
+
"step": 3240
|
| 2286 |
+
},
|
| 2287 |
+
{
|
| 2288 |
+
"epoch": 0.7587696258682075,
|
| 2289 |
+
"grad_norm": 0.019181331619620323,
|
| 2290 |
+
"learning_rate": 0.00010760051880674449,
|
| 2291 |
+
"loss": 0.0002,
|
| 2292 |
+
"step": 3250
|
| 2293 |
+
},
|
| 2294 |
+
{
|
| 2295 |
+
"epoch": 0.7611043016401098,
|
| 2296 |
+
"grad_norm": 0.007054260466247797,
|
| 2297 |
+
"learning_rate": 0.00010656290531776913,
|
| 2298 |
+
"loss": 0.0004,
|
| 2299 |
+
"step": 3260
|
| 2300 |
+
},
|
| 2301 |
+
{
|
| 2302 |
+
"epoch": 0.7634389774120119,
|
| 2303 |
+
"grad_norm": 0.0026721367612481117,
|
| 2304 |
+
"learning_rate": 0.00010552529182879379,
|
| 2305 |
+
"loss": 0.0001,
|
| 2306 |
+
"step": 3270
|
| 2307 |
+
},
|
| 2308 |
+
{
|
| 2309 |
+
"epoch": 0.7657736531839141,
|
| 2310 |
+
"grad_norm": 0.001824371051043272,
|
| 2311 |
+
"learning_rate": 0.00010448767833981843,
|
| 2312 |
+
"loss": 0.0002,
|
| 2313 |
+
"step": 3280
|
| 2314 |
+
},
|
| 2315 |
+
{
|
| 2316 |
+
"epoch": 0.7681083289558163,
|
| 2317 |
+
"grad_norm": 0.2344302535057068,
|
| 2318 |
+
"learning_rate": 0.00010345006485084307,
|
| 2319 |
+
"loss": 0.0004,
|
| 2320 |
+
"step": 3290
|
| 2321 |
+
},
|
| 2322 |
+
{
|
| 2323 |
+
"epoch": 0.7704430047277184,
|
| 2324 |
+
"grad_norm": 0.0015703398967161775,
|
| 2325 |
+
"learning_rate": 0.0001024124513618677,
|
| 2326 |
+
"loss": 0.0,
|
| 2327 |
+
"step": 3300
|
| 2328 |
+
},
|
| 2329 |
+
{
|
| 2330 |
+
"epoch": 0.7727776804996206,
|
| 2331 |
+
"grad_norm": 0.0013199965469539165,
|
| 2332 |
+
"learning_rate": 0.00010137483787289235,
|
| 2333 |
+
"loss": 0.0,
|
| 2334 |
+
"step": 3310
|
| 2335 |
+
},
|
| 2336 |
+
{
|
| 2337 |
+
"epoch": 0.7751123562715228,
|
| 2338 |
+
"grad_norm": 0.001341565977782011,
|
| 2339 |
+
"learning_rate": 0.00010033722438391699,
|
| 2340 |
+
"loss": 0.0,
|
| 2341 |
+
"step": 3320
|
| 2342 |
+
},
|
| 2343 |
+
{
|
| 2344 |
+
"epoch": 0.777447032043425,
|
| 2345 |
+
"grad_norm": 0.0012060283916071057,
|
| 2346 |
+
"learning_rate": 9.929961089494165e-05,
|
| 2347 |
+
"loss": 0.0001,
|
| 2348 |
+
"step": 3330
|
| 2349 |
+
},
|
| 2350 |
+
{
|
| 2351 |
+
"epoch": 0.7797817078153272,
|
| 2352 |
+
"grad_norm": 0.0020445692352950573,
|
| 2353 |
+
"learning_rate": 9.826199740596628e-05,
|
| 2354 |
+
"loss": 0.0,
|
| 2355 |
+
"step": 3340
|
| 2356 |
+
},
|
| 2357 |
+
{
|
| 2358 |
+
"epoch": 0.7821163835872293,
|
| 2359 |
+
"grad_norm": 0.0010797139257192612,
|
| 2360 |
+
"learning_rate": 9.722438391699092e-05,
|
| 2361 |
+
"loss": 0.0,
|
| 2362 |
+
"step": 3350
|
| 2363 |
+
},
|
| 2364 |
+
{
|
| 2365 |
+
"epoch": 0.7844510593591315,
|
| 2366 |
+
"grad_norm": 0.0013513348530977964,
|
| 2367 |
+
"learning_rate": 9.618677042801558e-05,
|
| 2368 |
+
"loss": 0.0,
|
| 2369 |
+
"step": 3360
|
| 2370 |
+
},
|
| 2371 |
+
{
|
| 2372 |
+
"epoch": 0.7867857351310337,
|
| 2373 |
+
"grad_norm": 0.0008800049545243382,
|
| 2374 |
+
"learning_rate": 9.514915693904021e-05,
|
| 2375 |
+
"loss": 0.0,
|
| 2376 |
+
"step": 3370
|
| 2377 |
+
},
|
| 2378 |
+
{
|
| 2379 |
+
"epoch": 0.7891204109029358,
|
| 2380 |
+
"grad_norm": 0.001039789873175323,
|
| 2381 |
+
"learning_rate": 9.411154345006485e-05,
|
| 2382 |
+
"loss": 0.0,
|
| 2383 |
+
"step": 3380
|
| 2384 |
+
},
|
| 2385 |
+
{
|
| 2386 |
+
"epoch": 0.791455086674838,
|
| 2387 |
+
"grad_norm": 0.0011056034127250314,
|
| 2388 |
+
"learning_rate": 9.307392996108951e-05,
|
| 2389 |
+
"loss": 0.0,
|
| 2390 |
+
"step": 3390
|
| 2391 |
+
},
|
| 2392 |
+
{
|
| 2393 |
+
"epoch": 0.7937897624467402,
|
| 2394 |
+
"grad_norm": 0.00087336590513587,
|
| 2395 |
+
"learning_rate": 9.203631647211414e-05,
|
| 2396 |
+
"loss": 0.0005,
|
| 2397 |
+
"step": 3400
|
| 2398 |
+
},
|
| 2399 |
+
{
|
| 2400 |
+
"epoch": 0.7961244382186424,
|
| 2401 |
+
"grad_norm": 0.0016204583225771785,
|
| 2402 |
+
"learning_rate": 9.099870298313878e-05,
|
| 2403 |
+
"loss": 0.0,
|
| 2404 |
+
"step": 3410
|
| 2405 |
+
},
|
| 2406 |
+
{
|
| 2407 |
+
"epoch": 0.7984591139905446,
|
| 2408 |
+
"grad_norm": 0.0010950096184387803,
|
| 2409 |
+
"learning_rate": 8.996108949416342e-05,
|
| 2410 |
+
"loss": 0.0,
|
| 2411 |
+
"step": 3420
|
| 2412 |
+
},
|
| 2413 |
+
{
|
| 2414 |
+
"epoch": 0.8007937897624468,
|
| 2415 |
+
"grad_norm": 0.0011948348255828023,
|
| 2416 |
+
"learning_rate": 8.892347600518807e-05,
|
| 2417 |
+
"loss": 0.0037,
|
| 2418 |
+
"step": 3430
|
| 2419 |
+
},
|
| 2420 |
+
{
|
| 2421 |
+
"epoch": 0.8031284655343489,
|
| 2422 |
+
"grad_norm": 0.009840068407356739,
|
| 2423 |
+
"learning_rate": 8.788586251621271e-05,
|
| 2424 |
+
"loss": 0.0001,
|
| 2425 |
+
"step": 3440
|
| 2426 |
+
},
|
| 2427 |
+
{
|
| 2428 |
+
"epoch": 0.8054631413062511,
|
| 2429 |
+
"grad_norm": 0.0017067514127120376,
|
| 2430 |
+
"learning_rate": 8.684824902723735e-05,
|
| 2431 |
+
"loss": 0.0001,
|
| 2432 |
+
"step": 3450
|
| 2433 |
+
},
|
| 2434 |
+
{
|
| 2435 |
+
"epoch": 0.8077978170781532,
|
| 2436 |
+
"grad_norm": 0.0011140963761135936,
|
| 2437 |
+
"learning_rate": 8.5810635538262e-05,
|
| 2438 |
+
"loss": 0.004,
|
| 2439 |
+
"step": 3460
|
| 2440 |
+
},
|
| 2441 |
+
{
|
| 2442 |
+
"epoch": 0.8101324928500554,
|
| 2443 |
+
"grad_norm": 0.7195191979408264,
|
| 2444 |
+
"learning_rate": 8.477302204928664e-05,
|
| 2445 |
+
"loss": 0.0035,
|
| 2446 |
+
"step": 3470
|
| 2447 |
+
},
|
| 2448 |
+
{
|
| 2449 |
+
"epoch": 0.8124671686219577,
|
| 2450 |
+
"grad_norm": 0.0012634329032152891,
|
| 2451 |
+
"learning_rate": 8.373540856031128e-05,
|
| 2452 |
+
"loss": 0.0001,
|
| 2453 |
+
"step": 3480
|
| 2454 |
+
},
|
| 2455 |
+
{
|
| 2456 |
+
"epoch": 0.8148018443938598,
|
| 2457 |
+
"grad_norm": 0.0016726938774809241,
|
| 2458 |
+
"learning_rate": 8.269779507133593e-05,
|
| 2459 |
+
"loss": 0.0006,
|
| 2460 |
+
"step": 3490
|
| 2461 |
+
},
|
| 2462 |
+
{
|
| 2463 |
+
"epoch": 0.817136520165762,
|
| 2464 |
+
"grad_norm": 0.0019955493044108152,
|
| 2465 |
+
"learning_rate": 8.166018158236057e-05,
|
| 2466 |
+
"loss": 0.0006,
|
| 2467 |
+
"step": 3500
|
| 2468 |
+
},
|
| 2469 |
+
{
|
| 2470 |
+
"epoch": 0.8194711959376642,
|
| 2471 |
+
"grad_norm": 0.0008943151333369315,
|
| 2472 |
+
"learning_rate": 8.062256809338522e-05,
|
| 2473 |
+
"loss": 0.0006,
|
| 2474 |
+
"step": 3510
|
| 2475 |
+
},
|
| 2476 |
+
{
|
| 2477 |
+
"epoch": 0.8218058717095663,
|
| 2478 |
+
"grad_norm": 0.0013045528903603554,
|
| 2479 |
+
"learning_rate": 7.958495460440986e-05,
|
| 2480 |
+
"loss": 0.0002,
|
| 2481 |
+
"step": 3520
|
| 2482 |
+
},
|
| 2483 |
+
{
|
| 2484 |
+
"epoch": 0.8241405474814685,
|
| 2485 |
+
"grad_norm": 0.0010028982069343328,
|
| 2486 |
+
"learning_rate": 7.85473411154345e-05,
|
| 2487 |
+
"loss": 0.0,
|
| 2488 |
+
"step": 3530
|
| 2489 |
+
},
|
| 2490 |
+
{
|
| 2491 |
+
"epoch": 0.8264752232533706,
|
| 2492 |
+
"grad_norm": 0.0007102734525687993,
|
| 2493 |
+
"learning_rate": 7.750972762645915e-05,
|
| 2494 |
+
"loss": 0.0001,
|
| 2495 |
+
"step": 3540
|
| 2496 |
+
},
|
| 2497 |
+
{
|
| 2498 |
+
"epoch": 0.8288098990252729,
|
| 2499 |
+
"grad_norm": 0.0014275240246206522,
|
| 2500 |
+
"learning_rate": 7.647211413748379e-05,
|
| 2501 |
+
"loss": 0.0,
|
| 2502 |
+
"step": 3550
|
| 2503 |
+
},
|
| 2504 |
+
{
|
| 2505 |
+
"epoch": 0.8311445747971751,
|
| 2506 |
+
"grad_norm": 0.0009326430154033005,
|
| 2507 |
+
"learning_rate": 7.543450064850843e-05,
|
| 2508 |
+
"loss": 0.0,
|
| 2509 |
+
"step": 3560
|
| 2510 |
+
},
|
| 2511 |
+
{
|
| 2512 |
+
"epoch": 0.8334792505690772,
|
| 2513 |
+
"grad_norm": 0.0008573593804612756,
|
| 2514 |
+
"learning_rate": 7.439688715953308e-05,
|
| 2515 |
+
"loss": 0.0001,
|
| 2516 |
+
"step": 3570
|
| 2517 |
+
},
|
| 2518 |
+
{
|
| 2519 |
+
"epoch": 0.8358139263409794,
|
| 2520 |
+
"grad_norm": 0.0015563720371574163,
|
| 2521 |
+
"learning_rate": 7.335927367055772e-05,
|
| 2522 |
+
"loss": 0.0002,
|
| 2523 |
+
"step": 3580
|
| 2524 |
+
},
|
| 2525 |
+
{
|
| 2526 |
+
"epoch": 0.8381486021128816,
|
| 2527 |
+
"grad_norm": 0.0008948877803049982,
|
| 2528 |
+
"learning_rate": 7.232166018158236e-05,
|
| 2529 |
+
"loss": 0.0,
|
| 2530 |
+
"step": 3590
|
| 2531 |
+
},
|
| 2532 |
+
{
|
| 2533 |
+
"epoch": 0.8404832778847837,
|
| 2534 |
+
"grad_norm": 0.0015601961640641093,
|
| 2535 |
+
"learning_rate": 7.1284046692607e-05,
|
| 2536 |
+
"loss": 0.0,
|
| 2537 |
+
"step": 3600
|
| 2538 |
+
},
|
| 2539 |
+
{
|
| 2540 |
+
"epoch": 0.8428179536566859,
|
| 2541 |
+
"grad_norm": 0.0013114233734086156,
|
| 2542 |
+
"learning_rate": 7.024643320363165e-05,
|
| 2543 |
+
"loss": 0.0073,
|
| 2544 |
+
"step": 3610
|
| 2545 |
+
},
|
| 2546 |
+
{
|
| 2547 |
+
"epoch": 0.8451526294285882,
|
| 2548 |
+
"grad_norm": 0.00176639249548316,
|
| 2549 |
+
"learning_rate": 6.920881971465629e-05,
|
| 2550 |
+
"loss": 0.0037,
|
| 2551 |
+
"step": 3620
|
| 2552 |
+
},
|
| 2553 |
+
{
|
| 2554 |
+
"epoch": 0.8474873052004903,
|
| 2555 |
+
"grad_norm": 0.002710576867684722,
|
| 2556 |
+
"learning_rate": 6.817120622568094e-05,
|
| 2557 |
+
"loss": 0.0001,
|
| 2558 |
+
"step": 3630
|
| 2559 |
+
},
|
| 2560 |
+
{
|
| 2561 |
+
"epoch": 0.8498219809723925,
|
| 2562 |
+
"grad_norm": 0.13074593245983124,
|
| 2563 |
+
"learning_rate": 6.713359273670558e-05,
|
| 2564 |
+
"loss": 0.001,
|
| 2565 |
+
"step": 3640
|
| 2566 |
+
},
|
| 2567 |
+
{
|
| 2568 |
+
"epoch": 0.8521566567442946,
|
| 2569 |
+
"grad_norm": 0.002523267176002264,
|
| 2570 |
+
"learning_rate": 6.609597924773022e-05,
|
| 2571 |
+
"loss": 0.0001,
|
| 2572 |
+
"step": 3650
|
| 2573 |
+
},
|
| 2574 |
+
{
|
| 2575 |
+
"epoch": 0.8544913325161968,
|
| 2576 |
+
"grad_norm": 0.002858164021745324,
|
| 2577 |
+
"learning_rate": 6.505836575875487e-05,
|
| 2578 |
+
"loss": 0.0003,
|
| 2579 |
+
"step": 3660
|
| 2580 |
+
},
|
| 2581 |
+
{
|
| 2582 |
+
"epoch": 0.856826008288099,
|
| 2583 |
+
"grad_norm": 0.002222646027803421,
|
| 2584 |
+
"learning_rate": 6.402075226977951e-05,
|
| 2585 |
+
"loss": 0.0006,
|
| 2586 |
+
"step": 3670
|
| 2587 |
+
},
|
| 2588 |
+
{
|
| 2589 |
+
"epoch": 0.8591606840600011,
|
| 2590 |
+
"grad_norm": 0.03722568228840828,
|
| 2591 |
+
"learning_rate": 6.298313878080415e-05,
|
| 2592 |
+
"loss": 0.0001,
|
| 2593 |
+
"step": 3680
|
| 2594 |
+
},
|
| 2595 |
+
{
|
| 2596 |
+
"epoch": 0.8614953598319034,
|
| 2597 |
+
"grad_norm": 0.0012012380175292492,
|
| 2598 |
+
"learning_rate": 6.19455252918288e-05,
|
| 2599 |
+
"loss": 0.0,
|
| 2600 |
+
"step": 3690
|
| 2601 |
+
},
|
| 2602 |
+
{
|
| 2603 |
+
"epoch": 0.8638300356038056,
|
| 2604 |
+
"grad_norm": 0.0019116230541840196,
|
| 2605 |
+
"learning_rate": 6.0907911802853433e-05,
|
| 2606 |
+
"loss": 0.0,
|
| 2607 |
+
"step": 3700
|
| 2608 |
+
},
|
| 2609 |
+
{
|
| 2610 |
+
"epoch": 0.8661647113757077,
|
| 2611 |
+
"grad_norm": 0.0011818850180134177,
|
| 2612 |
+
"learning_rate": 5.9870298313878084e-05,
|
| 2613 |
+
"loss": 0.0001,
|
| 2614 |
+
"step": 3710
|
| 2615 |
+
},
|
| 2616 |
+
{
|
| 2617 |
+
"epoch": 0.8684993871476099,
|
| 2618 |
+
"grad_norm": 0.0008876454085111618,
|
| 2619 |
+
"learning_rate": 5.883268482490273e-05,
|
| 2620 |
+
"loss": 0.0001,
|
| 2621 |
+
"step": 3720
|
| 2622 |
+
},
|
| 2623 |
+
{
|
| 2624 |
+
"epoch": 0.870834062919512,
|
| 2625 |
+
"grad_norm": 0.0011559055419638753,
|
| 2626 |
+
"learning_rate": 5.7795071335927364e-05,
|
| 2627 |
+
"loss": 0.0013,
|
| 2628 |
+
"step": 3730
|
| 2629 |
+
},
|
| 2630 |
+
{
|
| 2631 |
+
"epoch": 0.8731687386914142,
|
| 2632 |
+
"grad_norm": 0.0008210024680010974,
|
| 2633 |
+
"learning_rate": 5.6757457846952014e-05,
|
| 2634 |
+
"loss": 0.0001,
|
| 2635 |
+
"step": 3740
|
| 2636 |
+
},
|
| 2637 |
+
{
|
| 2638 |
+
"epoch": 0.8755034144633164,
|
| 2639 |
+
"grad_norm": 0.0019268837058916688,
|
| 2640 |
+
"learning_rate": 5.571984435797666e-05,
|
| 2641 |
+
"loss": 0.0001,
|
| 2642 |
+
"step": 3750
|
| 2643 |
+
},
|
| 2644 |
+
{
|
| 2645 |
+
"epoch": 0.8778380902352186,
|
| 2646 |
+
"grad_norm": 0.40103089809417725,
|
| 2647 |
+
"learning_rate": 5.4682230869001294e-05,
|
| 2648 |
+
"loss": 0.0006,
|
| 2649 |
+
"step": 3760
|
| 2650 |
+
},
|
| 2651 |
+
{
|
| 2652 |
+
"epoch": 0.8801727660071208,
|
| 2653 |
+
"grad_norm": 0.002693564398214221,
|
| 2654 |
+
"learning_rate": 5.3644617380025944e-05,
|
| 2655 |
+
"loss": 0.002,
|
| 2656 |
+
"step": 3770
|
| 2657 |
+
},
|
| 2658 |
+
{
|
| 2659 |
+
"epoch": 0.882507441779023,
|
| 2660 |
+
"grad_norm": 0.11337973922491074,
|
| 2661 |
+
"learning_rate": 5.260700389105059e-05,
|
| 2662 |
+
"loss": 0.0002,
|
| 2663 |
+
"step": 3780
|
| 2664 |
+
},
|
| 2665 |
+
{
|
| 2666 |
+
"epoch": 0.8848421175509251,
|
| 2667 |
+
"grad_norm": 0.000948163156863302,
|
| 2668 |
+
"learning_rate": 5.156939040207524e-05,
|
| 2669 |
+
"loss": 0.01,
|
| 2670 |
+
"step": 3790
|
| 2671 |
+
},
|
| 2672 |
+
{
|
| 2673 |
+
"epoch": 0.8871767933228273,
|
| 2674 |
+
"grad_norm": 0.0012289845617488027,
|
| 2675 |
+
"learning_rate": 5.053177691309987e-05,
|
| 2676 |
+
"loss": 0.0051,
|
| 2677 |
+
"step": 3800
|
| 2678 |
+
},
|
| 2679 |
+
{
|
| 2680 |
+
"epoch": 0.8895114690947294,
|
| 2681 |
+
"grad_norm": 1.0166712999343872,
|
| 2682 |
+
"learning_rate": 4.949416342412452e-05,
|
| 2683 |
+
"loss": 0.0009,
|
| 2684 |
+
"step": 3810
|
| 2685 |
+
},
|
| 2686 |
+
{
|
| 2687 |
+
"epoch": 0.8918461448666316,
|
| 2688 |
+
"grad_norm": 0.0020474784541875124,
|
| 2689 |
+
"learning_rate": 4.845654993514916e-05,
|
| 2690 |
+
"loss": 0.0003,
|
| 2691 |
+
"step": 3820
|
| 2692 |
+
},
|
| 2693 |
+
{
|
| 2694 |
+
"epoch": 0.8941808206385338,
|
| 2695 |
+
"grad_norm": 0.0022713476791977882,
|
| 2696 |
+
"learning_rate": 4.74189364461738e-05,
|
| 2697 |
+
"loss": 0.0001,
|
| 2698 |
+
"step": 3830
|
| 2699 |
+
},
|
| 2700 |
+
{
|
| 2701 |
+
"epoch": 0.896515496410436,
|
| 2702 |
+
"grad_norm": 0.004341310355812311,
|
| 2703 |
+
"learning_rate": 4.638132295719845e-05,
|
| 2704 |
+
"loss": 0.0,
|
| 2705 |
+
"step": 3840
|
| 2706 |
+
},
|
| 2707 |
+
{
|
| 2708 |
+
"epoch": 0.8988501721823382,
|
| 2709 |
+
"grad_norm": 0.0015770102618262172,
|
| 2710 |
+
"learning_rate": 4.534370946822309e-05,
|
| 2711 |
+
"loss": 0.0001,
|
| 2712 |
+
"step": 3850
|
| 2713 |
+
},
|
| 2714 |
+
{
|
| 2715 |
+
"epoch": 0.9011848479542404,
|
| 2716 |
+
"grad_norm": 0.012255331501364708,
|
| 2717 |
+
"learning_rate": 4.430609597924773e-05,
|
| 2718 |
+
"loss": 0.0001,
|
| 2719 |
+
"step": 3860
|
| 2720 |
+
},
|
| 2721 |
+
{
|
| 2722 |
+
"epoch": 0.9035195237261425,
|
| 2723 |
+
"grad_norm": 0.02712065726518631,
|
| 2724 |
+
"learning_rate": 4.326848249027238e-05,
|
| 2725 |
+
"loss": 0.0001,
|
| 2726 |
+
"step": 3870
|
| 2727 |
+
},
|
| 2728 |
+
{
|
| 2729 |
+
"epoch": 0.9058541994980447,
|
| 2730 |
+
"grad_norm": 0.001247554668225348,
|
| 2731 |
+
"learning_rate": 4.223086900129702e-05,
|
| 2732 |
+
"loss": 0.0,
|
| 2733 |
+
"step": 3880
|
| 2734 |
+
},
|
| 2735 |
+
{
|
| 2736 |
+
"epoch": 0.9081888752699468,
|
| 2737 |
+
"grad_norm": 0.0011685139033943415,
|
| 2738 |
+
"learning_rate": 4.119325551232166e-05,
|
| 2739 |
+
"loss": 0.0,
|
| 2740 |
+
"step": 3890
|
| 2741 |
+
},
|
| 2742 |
+
{
|
| 2743 |
+
"epoch": 0.910523551041849,
|
| 2744 |
+
"grad_norm": 0.0018253360176458955,
|
| 2745 |
+
"learning_rate": 4.015564202334631e-05,
|
| 2746 |
+
"loss": 0.0,
|
| 2747 |
+
"step": 3900
|
| 2748 |
+
},
|
| 2749 |
+
{
|
| 2750 |
+
"epoch": 0.9128582268137513,
|
| 2751 |
+
"grad_norm": 0.0008875974453985691,
|
| 2752 |
+
"learning_rate": 3.9118028534370945e-05,
|
| 2753 |
+
"loss": 0.0,
|
| 2754 |
+
"step": 3910
|
| 2755 |
+
},
|
| 2756 |
+
{
|
| 2757 |
+
"epoch": 0.9151929025856534,
|
| 2758 |
+
"grad_norm": 0.0011628433130681515,
|
| 2759 |
+
"learning_rate": 3.808041504539559e-05,
|
| 2760 |
+
"loss": 0.0001,
|
| 2761 |
+
"step": 3920
|
| 2762 |
+
},
|
| 2763 |
+
{
|
| 2764 |
+
"epoch": 0.9175275783575556,
|
| 2765 |
+
"grad_norm": 0.001058564055711031,
|
| 2766 |
+
"learning_rate": 3.704280155642024e-05,
|
| 2767 |
+
"loss": 0.0,
|
| 2768 |
+
"step": 3930
|
| 2769 |
+
},
|
| 2770 |
+
{
|
| 2771 |
+
"epoch": 0.9198622541294578,
|
| 2772 |
+
"grad_norm": 0.0010234726360067725,
|
| 2773 |
+
"learning_rate": 3.6005188067444876e-05,
|
| 2774 |
+
"loss": 0.0,
|
| 2775 |
+
"step": 3940
|
| 2776 |
+
},
|
| 2777 |
+
{
|
| 2778 |
+
"epoch": 0.9221969299013599,
|
| 2779 |
+
"grad_norm": 0.0009812922216951847,
|
| 2780 |
+
"learning_rate": 3.496757457846952e-05,
|
| 2781 |
+
"loss": 0.0004,
|
| 2782 |
+
"step": 3950
|
| 2783 |
+
},
|
| 2784 |
+
{
|
| 2785 |
+
"epoch": 0.9245316056732621,
|
| 2786 |
+
"grad_norm": 0.0009394401567988098,
|
| 2787 |
+
"learning_rate": 3.392996108949417e-05,
|
| 2788 |
+
"loss": 0.0,
|
| 2789 |
+
"step": 3960
|
| 2790 |
+
},
|
| 2791 |
+
{
|
| 2792 |
+
"epoch": 0.9268662814451643,
|
| 2793 |
+
"grad_norm": 0.0009512811666354537,
|
| 2794 |
+
"learning_rate": 3.2892347600518806e-05,
|
| 2795 |
+
"loss": 0.0001,
|
| 2796 |
+
"step": 3970
|
| 2797 |
+
},
|
| 2798 |
+
{
|
| 2799 |
+
"epoch": 0.9292009572170665,
|
| 2800 |
+
"grad_norm": 0.0007261955761350691,
|
| 2801 |
+
"learning_rate": 3.185473411154345e-05,
|
| 2802 |
+
"loss": 0.0,
|
| 2803 |
+
"step": 3980
|
| 2804 |
+
},
|
| 2805 |
+
{
|
| 2806 |
+
"epoch": 0.9315356329889687,
|
| 2807 |
+
"grad_norm": 0.0010610457975417376,
|
| 2808 |
+
"learning_rate": 3.08171206225681e-05,
|
| 2809 |
+
"loss": 0.0,
|
| 2810 |
+
"step": 3990
|
| 2811 |
+
},
|
| 2812 |
+
{
|
| 2813 |
+
"epoch": 0.9338703087608708,
|
| 2814 |
+
"grad_norm": 0.0012513543479144573,
|
| 2815 |
+
"learning_rate": 2.9779507133592736e-05,
|
| 2816 |
+
"loss": 0.0,
|
| 2817 |
+
"step": 4000
|
| 2818 |
+
},
|
| 2819 |
+
{
|
| 2820 |
+
"epoch": 0.936204984532773,
|
| 2821 |
+
"grad_norm": 0.0014717354206368327,
|
| 2822 |
+
"learning_rate": 2.874189364461738e-05,
|
| 2823 |
+
"loss": 0.0011,
|
| 2824 |
+
"step": 4010
|
| 2825 |
+
},
|
| 2826 |
+
{
|
| 2827 |
+
"epoch": 0.9385396603046752,
|
| 2828 |
+
"grad_norm": 0.0008392130257561803,
|
| 2829 |
+
"learning_rate": 2.7704280155642027e-05,
|
| 2830 |
+
"loss": 0.0,
|
| 2831 |
+
"step": 4020
|
| 2832 |
+
},
|
| 2833 |
+
{
|
| 2834 |
+
"epoch": 0.9408743360765773,
|
| 2835 |
+
"grad_norm": 0.0015690367436036468,
|
| 2836 |
+
"learning_rate": 2.6666666666666667e-05,
|
| 2837 |
+
"loss": 0.0,
|
| 2838 |
+
"step": 4030
|
| 2839 |
+
},
|
| 2840 |
+
{
|
| 2841 |
+
"epoch": 0.9432090118484795,
|
| 2842 |
+
"grad_norm": 0.0011248665396124125,
|
| 2843 |
+
"learning_rate": 2.562905317769131e-05,
|
| 2844 |
+
"loss": 0.002,
|
| 2845 |
+
"step": 4040
|
| 2846 |
+
},
|
| 2847 |
+
{
|
| 2848 |
+
"epoch": 0.9455436876203818,
|
| 2849 |
+
"grad_norm": 0.0010001506889238954,
|
| 2850 |
+
"learning_rate": 2.4591439688715953e-05,
|
| 2851 |
+
"loss": 0.0,
|
| 2852 |
+
"step": 4050
|
| 2853 |
+
},
|
| 2854 |
+
{
|
| 2855 |
+
"epoch": 0.9478783633922839,
|
| 2856 |
+
"grad_norm": 0.001105117262341082,
|
| 2857 |
+
"learning_rate": 2.3553826199740597e-05,
|
| 2858 |
+
"loss": 0.0,
|
| 2859 |
+
"step": 4060
|
| 2860 |
+
},
|
| 2861 |
+
{
|
| 2862 |
+
"epoch": 0.9502130391641861,
|
| 2863 |
+
"grad_norm": 0.010793734341859818,
|
| 2864 |
+
"learning_rate": 2.251621271076524e-05,
|
| 2865 |
+
"loss": 0.0001,
|
| 2866 |
+
"step": 4070
|
| 2867 |
+
},
|
| 2868 |
+
{
|
| 2869 |
+
"epoch": 0.9525477149360883,
|
| 2870 |
+
"grad_norm": 0.0010343483882024884,
|
| 2871 |
+
"learning_rate": 2.1478599221789884e-05,
|
| 2872 |
+
"loss": 0.0001,
|
| 2873 |
+
"step": 4080
|
| 2874 |
+
},
|
| 2875 |
+
{
|
| 2876 |
+
"epoch": 0.9548823907079904,
|
| 2877 |
+
"grad_norm": 0.0006852949154563248,
|
| 2878 |
+
"learning_rate": 2.0440985732814527e-05,
|
| 2879 |
+
"loss": 0.0,
|
| 2880 |
+
"step": 4090
|
| 2881 |
+
},
|
| 2882 |
+
{
|
| 2883 |
+
"epoch": 0.9572170664798926,
|
| 2884 |
+
"grad_norm": 0.001185077242553234,
|
| 2885 |
+
"learning_rate": 1.940337224383917e-05,
|
| 2886 |
+
"loss": 0.0,
|
| 2887 |
+
"step": 4100
|
| 2888 |
+
},
|
| 2889 |
+
{
|
| 2890 |
+
"epoch": 0.9595517422517947,
|
| 2891 |
+
"grad_norm": 0.055520687252283096,
|
| 2892 |
+
"learning_rate": 1.8365758754863814e-05,
|
| 2893 |
+
"loss": 0.0001,
|
| 2894 |
+
"step": 4110
|
| 2895 |
+
},
|
| 2896 |
+
{
|
| 2897 |
+
"epoch": 0.961886418023697,
|
| 2898 |
+
"grad_norm": 0.0012979560997337103,
|
| 2899 |
+
"learning_rate": 1.7328145265888457e-05,
|
| 2900 |
+
"loss": 0.0,
|
| 2901 |
+
"step": 4120
|
| 2902 |
+
},
|
| 2903 |
+
{
|
| 2904 |
+
"epoch": 0.9642210937955992,
|
| 2905 |
+
"grad_norm": 0.0013245136942714453,
|
| 2906 |
+
"learning_rate": 1.62905317769131e-05,
|
| 2907 |
+
"loss": 0.0,
|
| 2908 |
+
"step": 4130
|
| 2909 |
+
},
|
| 2910 |
+
{
|
| 2911 |
+
"epoch": 0.9665557695675013,
|
| 2912 |
+
"grad_norm": 0.0013994915643706918,
|
| 2913 |
+
"learning_rate": 1.5252918287937746e-05,
|
| 2914 |
+
"loss": 0.002,
|
| 2915 |
+
"step": 4140
|
| 2916 |
+
},
|
| 2917 |
+
{
|
| 2918 |
+
"epoch": 0.9688904453394035,
|
| 2919 |
+
"grad_norm": 0.6272192001342773,
|
| 2920 |
+
"learning_rate": 1.4215304798962386e-05,
|
| 2921 |
+
"loss": 0.0013,
|
| 2922 |
+
"step": 4150
|
| 2923 |
+
},
|
| 2924 |
+
{
|
| 2925 |
+
"epoch": 0.9712251211113057,
|
| 2926 |
+
"grad_norm": 0.0012450398644432425,
|
| 2927 |
+
"learning_rate": 1.3177691309987031e-05,
|
| 2928 |
+
"loss": 0.0002,
|
| 2929 |
+
"step": 4160
|
| 2930 |
+
},
|
| 2931 |
+
{
|
| 2932 |
+
"epoch": 0.9735597968832078,
|
| 2933 |
+
"grad_norm": 0.0009830017806962132,
|
| 2934 |
+
"learning_rate": 1.2140077821011673e-05,
|
| 2935 |
+
"loss": 0.0001,
|
| 2936 |
+
"step": 4170
|
| 2937 |
+
},
|
| 2938 |
+
{
|
| 2939 |
+
"epoch": 0.97589447265511,
|
| 2940 |
+
"grad_norm": 0.0007283110171556473,
|
| 2941 |
+
"learning_rate": 1.1102464332036316e-05,
|
| 2942 |
+
"loss": 0.0,
|
| 2943 |
+
"step": 4180
|
| 2944 |
+
},
|
| 2945 |
+
{
|
| 2946 |
+
"epoch": 0.9782291484270123,
|
| 2947 |
+
"grad_norm": 0.0008772446890361607,
|
| 2948 |
+
"learning_rate": 1.006485084306096e-05,
|
| 2949 |
+
"loss": 0.0,
|
| 2950 |
+
"step": 4190
|
| 2951 |
+
},
|
| 2952 |
+
{
|
| 2953 |
+
"epoch": 0.9805638241989144,
|
| 2954 |
+
"grad_norm": 0.0007983844261616468,
|
| 2955 |
+
"learning_rate": 9.027237354085603e-06,
|
| 2956 |
+
"loss": 0.0002,
|
| 2957 |
+
"step": 4200
|
| 2958 |
+
},
|
| 2959 |
+
{
|
| 2960 |
+
"epoch": 0.9828984999708166,
|
| 2961 |
+
"grad_norm": 0.0045978049747645855,
|
| 2962 |
+
"learning_rate": 7.989623865110247e-06,
|
| 2963 |
+
"loss": 0.0001,
|
| 2964 |
+
"step": 4210
|
| 2965 |
+
},
|
| 2966 |
+
{
|
| 2967 |
+
"epoch": 0.9852331757427187,
|
| 2968 |
+
"grad_norm": 0.0006691565504297614,
|
| 2969 |
+
"learning_rate": 6.95201037613489e-06,
|
| 2970 |
+
"loss": 0.0,
|
| 2971 |
+
"step": 4220
|
| 2972 |
+
},
|
| 2973 |
+
{
|
| 2974 |
+
"epoch": 0.9875678515146209,
|
| 2975 |
+
"grad_norm": 0.000847226707264781,
|
| 2976 |
+
"learning_rate": 5.9143968871595335e-06,
|
| 2977 |
+
"loss": 0.0,
|
| 2978 |
+
"step": 4230
|
| 2979 |
+
},
|
| 2980 |
+
{
|
| 2981 |
+
"epoch": 0.9899025272865231,
|
| 2982 |
+
"grad_norm": 0.0008708458044566214,
|
| 2983 |
+
"learning_rate": 4.876783398184177e-06,
|
| 2984 |
+
"loss": 0.0,
|
| 2985 |
+
"step": 4240
|
| 2986 |
+
},
|
| 2987 |
+
{
|
| 2988 |
+
"epoch": 0.9922372030584252,
|
| 2989 |
+
"grad_norm": 0.001017833361402154,
|
| 2990 |
+
"learning_rate": 3.83916990920882e-06,
|
| 2991 |
+
"loss": 0.0,
|
| 2992 |
+
"step": 4250
|
| 2993 |
+
},
|
| 2994 |
+
{
|
| 2995 |
+
"epoch": 0.9945718788303275,
|
| 2996 |
+
"grad_norm": 0.001224992680363357,
|
| 2997 |
+
"learning_rate": 2.8015564202334633e-06,
|
| 2998 |
+
"loss": 0.0,
|
| 2999 |
+
"step": 4260
|
| 3000 |
+
},
|
| 3001 |
+
{
|
| 3002 |
+
"epoch": 0.9969065546022297,
|
| 3003 |
+
"grad_norm": 0.0007001100457273424,
|
| 3004 |
+
"learning_rate": 1.7639429312581063e-06,
|
| 3005 |
+
"loss": 0.0001,
|
| 3006 |
+
"step": 4270
|
| 3007 |
+
},
|
| 3008 |
+
{
|
| 3009 |
+
"epoch": 0.9992412303741318,
|
| 3010 |
+
"grad_norm": 0.0008271584520116448,
|
| 3011 |
+
"learning_rate": 7.263294422827498e-07,
|
| 3012 |
+
"loss": 0.0001,
|
| 3013 |
+
"step": 4280
|
| 3014 |
+
},
|
| 3015 |
+
{
|
| 3016 |
+
"epoch": 1.0,
|
| 3017 |
+
"eval_accuracy": 0.9999124458258547,
|
| 3018 |
+
"eval_loss": 0.00038632494397461414,
|
| 3019 |
+
"eval_runtime": 4629.6003,
|
| 3020 |
+
"eval_samples_per_second": 14.802,
|
| 3021 |
+
"eval_steps_per_second": 0.925,
|
| 3022 |
+
"step": 4284
|
| 3023 |
+
}
|
| 3024 |
+
],
|
| 3025 |
+
"logging_steps": 10,
|
| 3026 |
+
"max_steps": 4284,
|
| 3027 |
+
"num_input_tokens_seen": 0,
|
| 3028 |
+
"num_train_epochs": 1,
|
| 3029 |
+
"save_steps": 500,
|
| 3030 |
+
"stateful_callbacks": {
|
| 3031 |
+
"TrainerControl": {
|
| 3032 |
+
"args": {
|
| 3033 |
+
"should_epoch_stop": false,
|
| 3034 |
+
"should_evaluate": false,
|
| 3035 |
+
"should_log": false,
|
| 3036 |
+
"should_save": true,
|
| 3037 |
+
"should_training_stop": true
|
| 3038 |
+
},
|
| 3039 |
+
"attributes": {}
|
| 3040 |
+
}
|
| 3041 |
+
},
|
| 3042 |
+
"total_flos": 3.44199965772096e+19,
|
| 3043 |
+
"train_batch_size": 16,
|
| 3044 |
+
"trial_name": null,
|
| 3045 |
+
"trial_params": null
|
| 3046 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5d74acea8e8c02da181e27d0af27b74fb2387380a7d2506f15369d44ffd9a9a9
|
| 3 |
+
size 5841
|