Upload 14 files
Browse files- qqp/README.md +79 -0
- qqp/all_results.json +17 -0
- qqp/config.json +32 -0
- qqp/eval_results.json +11 -0
- qqp/model.safetensors +3 -0
- qqp/runs/May14_17-08-38_cs-Precision-7960-Tower/events.out.tfevents.1747256924.cs-Precision-7960-Tower.131796.0 +3 -0
- qqp/runs/May14_17-08-38_cs-Precision-7960-Tower/events.out.tfevents.1747277032.cs-Precision-7960-Tower.131796.1 +3 -0
- qqp/special_tokens_map.json +7 -0
- qqp/tokenizer.json +0 -0
- qqp/tokenizer_config.json +56 -0
- qqp/train_results.json +9 -0
- qqp/trainer_state.json +1631 -0
- qqp/training_args.bin +3 -0
- qqp/vocab.txt +0 -0
qqp/README.md
ADDED
|
@@ -0,0 +1,79 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
library_name: transformers
|
| 3 |
+
language:
|
| 4 |
+
- en
|
| 5 |
+
license: apache-2.0
|
| 6 |
+
base_model: google-bert/bert-base-cased
|
| 7 |
+
tags:
|
| 8 |
+
- generated_from_trainer
|
| 9 |
+
datasets:
|
| 10 |
+
- glue
|
| 11 |
+
metrics:
|
| 12 |
+
- accuracy
|
| 13 |
+
- f1
|
| 14 |
+
model-index:
|
| 15 |
+
- name: qqp
|
| 16 |
+
results:
|
| 17 |
+
- task:
|
| 18 |
+
name: Text Classification
|
| 19 |
+
type: text-classification
|
| 20 |
+
dataset:
|
| 21 |
+
name: GLUE QQP
|
| 22 |
+
type: glue
|
| 23 |
+
args: qqp
|
| 24 |
+
metrics:
|
| 25 |
+
- name: Accuracy
|
| 26 |
+
type: accuracy
|
| 27 |
+
value: 0.9088053425674004
|
| 28 |
+
- name: F1
|
| 29 |
+
type: f1
|
| 30 |
+
value: 0.8766435812506274
|
| 31 |
+
---
|
| 32 |
+
|
| 33 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 34 |
+
should probably proofread and complete it, then remove this comment. -->
|
| 35 |
+
|
| 36 |
+
# qqp
|
| 37 |
+
|
| 38 |
+
This model is a fine-tuned version of [google-bert/bert-base-cased](https://huggingface.co/google-bert/bert-base-cased) on the GLUE QQP dataset.
|
| 39 |
+
It achieves the following results on the evaluation set:
|
| 40 |
+
- Loss: 0.5929
|
| 41 |
+
- Accuracy: 0.9088
|
| 42 |
+
- F1: 0.8766
|
| 43 |
+
- Combined Score: 0.8927
|
| 44 |
+
|
| 45 |
+
## Model description
|
| 46 |
+
|
| 47 |
+
More information needed
|
| 48 |
+
|
| 49 |
+
## Intended uses & limitations
|
| 50 |
+
|
| 51 |
+
More information needed
|
| 52 |
+
|
| 53 |
+
## Training and evaluation data
|
| 54 |
+
|
| 55 |
+
More information needed
|
| 56 |
+
|
| 57 |
+
## Training procedure
|
| 58 |
+
|
| 59 |
+
### Training hyperparameters
|
| 60 |
+
|
| 61 |
+
The following hyperparameters were used during training:
|
| 62 |
+
- learning_rate: 1e-05
|
| 63 |
+
- train_batch_size: 32
|
| 64 |
+
- eval_batch_size: 8
|
| 65 |
+
- seed: 42
|
| 66 |
+
- optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
|
| 67 |
+
- lr_scheduler_type: linear
|
| 68 |
+
- num_epochs: 10.0
|
| 69 |
+
|
| 70 |
+
### Training results
|
| 71 |
+
|
| 72 |
+
|
| 73 |
+
|
| 74 |
+
### Framework versions
|
| 75 |
+
|
| 76 |
+
- Transformers 4.49.0
|
| 77 |
+
- Pytorch 2.6.0+cu118
|
| 78 |
+
- Datasets 3.3.1
|
| 79 |
+
- Tokenizers 0.21.0
|
qqp/all_results.json
ADDED
|
@@ -0,0 +1,17 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 10.0,
|
| 3 |
+
"eval_accuracy": 0.9088053425674004,
|
| 4 |
+
"eval_combined_score": 0.8927244619090139,
|
| 5 |
+
"eval_f1": 0.8766435812506274,
|
| 6 |
+
"eval_loss": 0.5928972363471985,
|
| 7 |
+
"eval_runtime": 78.1295,
|
| 8 |
+
"eval_samples": 40430,
|
| 9 |
+
"eval_samples_per_second": 517.474,
|
| 10 |
+
"eval_steps_per_second": 64.687,
|
| 11 |
+
"total_flos": 2.4428070234989568e+17,
|
| 12 |
+
"train_loss": 0.1514742186763295,
|
| 13 |
+
"train_runtime": 20029.6511,
|
| 14 |
+
"train_samples": 363846,
|
| 15 |
+
"train_samples_per_second": 181.654,
|
| 16 |
+
"train_steps_per_second": 5.677
|
| 17 |
+
}
|
qqp/config.json
ADDED
|
@@ -0,0 +1,32 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "google-bert/bert-base-cased",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"BertForSequenceClassification"
|
| 5 |
+
],
|
| 6 |
+
"attention_probs_dropout_prob": 0.1,
|
| 7 |
+
"classifier_dropout": null,
|
| 8 |
+
"finetuning_task": "qqp",
|
| 9 |
+
"gradient_checkpointing": false,
|
| 10 |
+
"hidden_act": "gelu",
|
| 11 |
+
"hidden_dropout_prob": 0.1,
|
| 12 |
+
"hidden_size": 768,
|
| 13 |
+
"initializer_range": 0.02,
|
| 14 |
+
"intermediate_size": 3072,
|
| 15 |
+
"label2id": {
|
| 16 |
+
"duplicate": 1,
|
| 17 |
+
"not_duplicate": 0
|
| 18 |
+
},
|
| 19 |
+
"layer_norm_eps": 1e-12,
|
| 20 |
+
"max_position_embeddings": 512,
|
| 21 |
+
"model_type": "bert",
|
| 22 |
+
"num_attention_heads": 12,
|
| 23 |
+
"num_hidden_layers": 12,
|
| 24 |
+
"pad_token_id": 0,
|
| 25 |
+
"position_embedding_type": "absolute",
|
| 26 |
+
"problem_type": "single_label_classification",
|
| 27 |
+
"torch_dtype": "float32",
|
| 28 |
+
"transformers_version": "4.49.0",
|
| 29 |
+
"type_vocab_size": 2,
|
| 30 |
+
"use_cache": true,
|
| 31 |
+
"vocab_size": 28996
|
| 32 |
+
}
|
qqp/eval_results.json
ADDED
|
@@ -0,0 +1,11 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 10.0,
|
| 3 |
+
"eval_accuracy": 0.9088053425674004,
|
| 4 |
+
"eval_combined_score": 0.8927244619090139,
|
| 5 |
+
"eval_f1": 0.8766435812506274,
|
| 6 |
+
"eval_loss": 0.5928972363471985,
|
| 7 |
+
"eval_runtime": 78.1295,
|
| 8 |
+
"eval_samples": 40430,
|
| 9 |
+
"eval_samples_per_second": 517.474,
|
| 10 |
+
"eval_steps_per_second": 64.687
|
| 11 |
+
}
|
qqp/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:557400d81363c1bd136493dbce5e06be104ed016bc01fb0f0cdd8a04f198562c
|
| 3 |
+
size 440358464
|
qqp/runs/May14_17-08-38_cs-Precision-7960-Tower/events.out.tfevents.1747256924.cs-Precision-7960-Tower.131796.0
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:eb497e14eca1ed4c6b40bcca838fd4ba2675b74b08dbcbe7ec9db485bfeea81e
|
| 3 |
+
size 54169
|
qqp/runs/May14_17-08-38_cs-Precision-7960-Tower/events.out.tfevents.1747277032.cs-Precision-7960-Tower.131796.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b9dbec5238d0673df0a8a084e053d3fa61605bd3ef9f7ca7e348ed28db2805f4
|
| 3 |
+
size 523
|
qqp/special_tokens_map.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"cls_token": "[CLS]",
|
| 3 |
+
"mask_token": "[MASK]",
|
| 4 |
+
"pad_token": "[PAD]",
|
| 5 |
+
"sep_token": "[SEP]",
|
| 6 |
+
"unk_token": "[UNK]"
|
| 7 |
+
}
|
qqp/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
qqp/tokenizer_config.json
ADDED
|
@@ -0,0 +1,56 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"added_tokens_decoder": {
|
| 3 |
+
"0": {
|
| 4 |
+
"content": "[PAD]",
|
| 5 |
+
"lstrip": false,
|
| 6 |
+
"normalized": false,
|
| 7 |
+
"rstrip": false,
|
| 8 |
+
"single_word": false,
|
| 9 |
+
"special": true
|
| 10 |
+
},
|
| 11 |
+
"100": {
|
| 12 |
+
"content": "[UNK]",
|
| 13 |
+
"lstrip": false,
|
| 14 |
+
"normalized": false,
|
| 15 |
+
"rstrip": false,
|
| 16 |
+
"single_word": false,
|
| 17 |
+
"special": true
|
| 18 |
+
},
|
| 19 |
+
"101": {
|
| 20 |
+
"content": "[CLS]",
|
| 21 |
+
"lstrip": false,
|
| 22 |
+
"normalized": false,
|
| 23 |
+
"rstrip": false,
|
| 24 |
+
"single_word": false,
|
| 25 |
+
"special": true
|
| 26 |
+
},
|
| 27 |
+
"102": {
|
| 28 |
+
"content": "[SEP]",
|
| 29 |
+
"lstrip": false,
|
| 30 |
+
"normalized": false,
|
| 31 |
+
"rstrip": false,
|
| 32 |
+
"single_word": false,
|
| 33 |
+
"special": true
|
| 34 |
+
},
|
| 35 |
+
"103": {
|
| 36 |
+
"content": "[MASK]",
|
| 37 |
+
"lstrip": false,
|
| 38 |
+
"normalized": false,
|
| 39 |
+
"rstrip": false,
|
| 40 |
+
"single_word": false,
|
| 41 |
+
"special": true
|
| 42 |
+
}
|
| 43 |
+
},
|
| 44 |
+
"clean_up_tokenization_spaces": false,
|
| 45 |
+
"cls_token": "[CLS]",
|
| 46 |
+
"do_lower_case": false,
|
| 47 |
+
"extra_special_tokens": {},
|
| 48 |
+
"mask_token": "[MASK]",
|
| 49 |
+
"model_max_length": 512,
|
| 50 |
+
"pad_token": "[PAD]",
|
| 51 |
+
"sep_token": "[SEP]",
|
| 52 |
+
"strip_accents": null,
|
| 53 |
+
"tokenize_chinese_chars": true,
|
| 54 |
+
"tokenizer_class": "BertTokenizer",
|
| 55 |
+
"unk_token": "[UNK]"
|
| 56 |
+
}
|
qqp/train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 10.0,
|
| 3 |
+
"total_flos": 2.4428070234989568e+17,
|
| 4 |
+
"train_loss": 0.1514742186763295,
|
| 5 |
+
"train_runtime": 20029.6511,
|
| 6 |
+
"train_samples": 363846,
|
| 7 |
+
"train_samples_per_second": 181.654,
|
| 8 |
+
"train_steps_per_second": 5.677
|
| 9 |
+
}
|
qqp/trainer_state.json
ADDED
|
@@ -0,0 +1,1631 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 10.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 113710,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.04397150646381145,
|
| 13 |
+
"grad_norm": 7.127589225769043,
|
| 14 |
+
"learning_rate": 9.956028493536189e-06,
|
| 15 |
+
"loss": 0.6405,
|
| 16 |
+
"step": 500
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.0879430129276229,
|
| 20 |
+
"grad_norm": 10.408858299255371,
|
| 21 |
+
"learning_rate": 9.912056987072378e-06,
|
| 22 |
+
"loss": 0.4966,
|
| 23 |
+
"step": 1000
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.13191451939143434,
|
| 27 |
+
"grad_norm": 8.54762077331543,
|
| 28 |
+
"learning_rate": 9.868085480608566e-06,
|
| 29 |
+
"loss": 0.4625,
|
| 30 |
+
"step": 1500
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.1758860258552458,
|
| 34 |
+
"grad_norm": 8.460881233215332,
|
| 35 |
+
"learning_rate": 9.824113974144756e-06,
|
| 36 |
+
"loss": 0.4435,
|
| 37 |
+
"step": 2000
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.21985753231905725,
|
| 41 |
+
"grad_norm": 6.150853633880615,
|
| 42 |
+
"learning_rate": 9.780142467680944e-06,
|
| 43 |
+
"loss": 0.4288,
|
| 44 |
+
"step": 2500
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.2638290387828687,
|
| 48 |
+
"grad_norm": 7.34383487701416,
|
| 49 |
+
"learning_rate": 9.736170961217132e-06,
|
| 50 |
+
"loss": 0.4045,
|
| 51 |
+
"step": 3000
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.30780054524668016,
|
| 55 |
+
"grad_norm": 10.842817306518555,
|
| 56 |
+
"learning_rate": 9.69219945475332e-06,
|
| 57 |
+
"loss": 0.4014,
|
| 58 |
+
"step": 3500
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.3517720517104916,
|
| 62 |
+
"grad_norm": 11.085785865783691,
|
| 63 |
+
"learning_rate": 9.64822794828951e-06,
|
| 64 |
+
"loss": 0.4087,
|
| 65 |
+
"step": 4000
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.39574355817430307,
|
| 69 |
+
"grad_norm": 10.106760025024414,
|
| 70 |
+
"learning_rate": 9.604256441825697e-06,
|
| 71 |
+
"loss": 0.391,
|
| 72 |
+
"step": 4500
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.4397150646381145,
|
| 76 |
+
"grad_norm": 12.384110450744629,
|
| 77 |
+
"learning_rate": 9.560284935361887e-06,
|
| 78 |
+
"loss": 0.3846,
|
| 79 |
+
"step": 5000
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.483686571101926,
|
| 83 |
+
"grad_norm": 6.796555519104004,
|
| 84 |
+
"learning_rate": 9.516313428898075e-06,
|
| 85 |
+
"loss": 0.3803,
|
| 86 |
+
"step": 5500
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.5276580775657373,
|
| 90 |
+
"grad_norm": 10.099310874938965,
|
| 91 |
+
"learning_rate": 9.472341922434265e-06,
|
| 92 |
+
"loss": 0.3739,
|
| 93 |
+
"step": 6000
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.5716295840295489,
|
| 97 |
+
"grad_norm": 8.258896827697754,
|
| 98 |
+
"learning_rate": 9.428370415970451e-06,
|
| 99 |
+
"loss": 0.3694,
|
| 100 |
+
"step": 6500
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.6156010904933603,
|
| 104 |
+
"grad_norm": 6.4571356773376465,
|
| 105 |
+
"learning_rate": 9.38439890950664e-06,
|
| 106 |
+
"loss": 0.3582,
|
| 107 |
+
"step": 7000
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.6595725969571717,
|
| 111 |
+
"grad_norm": 10.41157341003418,
|
| 112 |
+
"learning_rate": 9.340427403042828e-06,
|
| 113 |
+
"loss": 0.3611,
|
| 114 |
+
"step": 7500
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.7035441034209832,
|
| 118 |
+
"grad_norm": 8.555697441101074,
|
| 119 |
+
"learning_rate": 9.296455896579018e-06,
|
| 120 |
+
"loss": 0.3491,
|
| 121 |
+
"step": 8000
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.7475156098847947,
|
| 125 |
+
"grad_norm": 10.460009574890137,
|
| 126 |
+
"learning_rate": 9.252484390115206e-06,
|
| 127 |
+
"loss": 0.3507,
|
| 128 |
+
"step": 8500
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.7914871163486061,
|
| 132 |
+
"grad_norm": 9.326970100402832,
|
| 133 |
+
"learning_rate": 9.208512883651394e-06,
|
| 134 |
+
"loss": 0.3465,
|
| 135 |
+
"step": 9000
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.8354586228124176,
|
| 139 |
+
"grad_norm": 12.12388801574707,
|
| 140 |
+
"learning_rate": 9.164541377187584e-06,
|
| 141 |
+
"loss": 0.3419,
|
| 142 |
+
"step": 9500
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.879430129276229,
|
| 146 |
+
"grad_norm": 9.742505073547363,
|
| 147 |
+
"learning_rate": 9.120569870723772e-06,
|
| 148 |
+
"loss": 0.3492,
|
| 149 |
+
"step": 10000
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.9234016357400404,
|
| 153 |
+
"grad_norm": 8.42115306854248,
|
| 154 |
+
"learning_rate": 9.07659836425996e-06,
|
| 155 |
+
"loss": 0.3405,
|
| 156 |
+
"step": 10500
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.967373142203852,
|
| 160 |
+
"grad_norm": 11.557007789611816,
|
| 161 |
+
"learning_rate": 9.03262685779615e-06,
|
| 162 |
+
"loss": 0.3353,
|
| 163 |
+
"step": 11000
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 1.0113446486676634,
|
| 167 |
+
"grad_norm": 6.1636786460876465,
|
| 168 |
+
"learning_rate": 8.988655351332337e-06,
|
| 169 |
+
"loss": 0.3218,
|
| 170 |
+
"step": 11500
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 1.0553161551314747,
|
| 174 |
+
"grad_norm": 8.901851654052734,
|
| 175 |
+
"learning_rate": 8.944683844868525e-06,
|
| 176 |
+
"loss": 0.2812,
|
| 177 |
+
"step": 12000
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 1.0992876615952862,
|
| 181 |
+
"grad_norm": 8.753206253051758,
|
| 182 |
+
"learning_rate": 8.900712338404715e-06,
|
| 183 |
+
"loss": 0.2744,
|
| 184 |
+
"step": 12500
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 1.1432591680590978,
|
| 188 |
+
"grad_norm": 19.687353134155273,
|
| 189 |
+
"learning_rate": 8.856740831940903e-06,
|
| 190 |
+
"loss": 0.2716,
|
| 191 |
+
"step": 13000
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 1.187230674522909,
|
| 195 |
+
"grad_norm": 8.538763999938965,
|
| 196 |
+
"learning_rate": 8.812769325477092e-06,
|
| 197 |
+
"loss": 0.2731,
|
| 198 |
+
"step": 13500
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 1.2312021809867206,
|
| 202 |
+
"grad_norm": 14.86081314086914,
|
| 203 |
+
"learning_rate": 8.76879781901328e-06,
|
| 204 |
+
"loss": 0.2686,
|
| 205 |
+
"step": 14000
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 1.2751736874505322,
|
| 209 |
+
"grad_norm": 8.427505493164062,
|
| 210 |
+
"learning_rate": 8.72482631254947e-06,
|
| 211 |
+
"loss": 0.2694,
|
| 212 |
+
"step": 14500
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 1.3191451939143435,
|
| 216 |
+
"grad_norm": 11.6419095993042,
|
| 217 |
+
"learning_rate": 8.680854806085656e-06,
|
| 218 |
+
"loss": 0.2666,
|
| 219 |
+
"step": 15000
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 1.363116700378155,
|
| 223 |
+
"grad_norm": 11.019262313842773,
|
| 224 |
+
"learning_rate": 8.636883299621846e-06,
|
| 225 |
+
"loss": 0.2738,
|
| 226 |
+
"step": 15500
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 1.4070882068419663,
|
| 230 |
+
"grad_norm": 12.447158813476562,
|
| 231 |
+
"learning_rate": 8.592911793158034e-06,
|
| 232 |
+
"loss": 0.2712,
|
| 233 |
+
"step": 16000
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 1.4510597133057779,
|
| 237 |
+
"grad_norm": 13.097723007202148,
|
| 238 |
+
"learning_rate": 8.548940286694223e-06,
|
| 239 |
+
"loss": 0.2635,
|
| 240 |
+
"step": 16500
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 1.4950312197695892,
|
| 244 |
+
"grad_norm": 15.15329360961914,
|
| 245 |
+
"learning_rate": 8.504968780230411e-06,
|
| 246 |
+
"loss": 0.2662,
|
| 247 |
+
"step": 17000
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 1.5390027262334007,
|
| 251 |
+
"grad_norm": 6.069318771362305,
|
| 252 |
+
"learning_rate": 8.460997273766601e-06,
|
| 253 |
+
"loss": 0.2604,
|
| 254 |
+
"step": 17500
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 1.5829742326972123,
|
| 258 |
+
"grad_norm": 13.966374397277832,
|
| 259 |
+
"learning_rate": 8.417025767302789e-06,
|
| 260 |
+
"loss": 0.2657,
|
| 261 |
+
"step": 18000
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 1.6269457391610236,
|
| 265 |
+
"grad_norm": 21.772634506225586,
|
| 266 |
+
"learning_rate": 8.373054260838977e-06,
|
| 267 |
+
"loss": 0.262,
|
| 268 |
+
"step": 18500
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 1.6709172456248351,
|
| 272 |
+
"grad_norm": 11.764993667602539,
|
| 273 |
+
"learning_rate": 8.329082754375165e-06,
|
| 274 |
+
"loss": 0.2721,
|
| 275 |
+
"step": 19000
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 1.7148887520886467,
|
| 279 |
+
"grad_norm": 11.220842361450195,
|
| 280 |
+
"learning_rate": 8.285111247911354e-06,
|
| 281 |
+
"loss": 0.2611,
|
| 282 |
+
"step": 19500
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 1.758860258552458,
|
| 286 |
+
"grad_norm": 12.129228591918945,
|
| 287 |
+
"learning_rate": 8.241139741447542e-06,
|
| 288 |
+
"loss": 0.2684,
|
| 289 |
+
"step": 20000
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 1.8028317650162693,
|
| 293 |
+
"grad_norm": 13.62337875366211,
|
| 294 |
+
"learning_rate": 8.19716823498373e-06,
|
| 295 |
+
"loss": 0.2593,
|
| 296 |
+
"step": 20500
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 1.846803271480081,
|
| 300 |
+
"grad_norm": 9.742755889892578,
|
| 301 |
+
"learning_rate": 8.15319672851992e-06,
|
| 302 |
+
"loss": 0.2633,
|
| 303 |
+
"step": 21000
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 1.8907747779438924,
|
| 307 |
+
"grad_norm": 11.295736312866211,
|
| 308 |
+
"learning_rate": 8.109225222056108e-06,
|
| 309 |
+
"loss": 0.2652,
|
| 310 |
+
"step": 21500
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 1.9347462844077037,
|
| 314 |
+
"grad_norm": 8.879314422607422,
|
| 315 |
+
"learning_rate": 8.065253715592298e-06,
|
| 316 |
+
"loss": 0.2651,
|
| 317 |
+
"step": 22000
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 1.9787177908715152,
|
| 321 |
+
"grad_norm": 9.949023246765137,
|
| 322 |
+
"learning_rate": 8.021282209128485e-06,
|
| 323 |
+
"loss": 0.27,
|
| 324 |
+
"step": 22500
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 2.0226892973353268,
|
| 328 |
+
"grad_norm": 4.5543622970581055,
|
| 329 |
+
"learning_rate": 7.977310702664673e-06,
|
| 330 |
+
"loss": 0.2283,
|
| 331 |
+
"step": 23000
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 2.066660803799138,
|
| 335 |
+
"grad_norm": 13.172735214233398,
|
| 336 |
+
"learning_rate": 7.933339196200861e-06,
|
| 337 |
+
"loss": 0.2061,
|
| 338 |
+
"step": 23500
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 2.1106323102629494,
|
| 342 |
+
"grad_norm": 12.874068260192871,
|
| 343 |
+
"learning_rate": 7.889367689737051e-06,
|
| 344 |
+
"loss": 0.2019,
|
| 345 |
+
"step": 24000
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 2.154603816726761,
|
| 349 |
+
"grad_norm": 17.02819061279297,
|
| 350 |
+
"learning_rate": 7.845396183273239e-06,
|
| 351 |
+
"loss": 0.2046,
|
| 352 |
+
"step": 24500
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 2.1985753231905725,
|
| 356 |
+
"grad_norm": 15.347699165344238,
|
| 357 |
+
"learning_rate": 7.801424676809429e-06,
|
| 358 |
+
"loss": 0.2035,
|
| 359 |
+
"step": 25000
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 2.242546829654384,
|
| 363 |
+
"grad_norm": 15.82715892791748,
|
| 364 |
+
"learning_rate": 7.757453170345617e-06,
|
| 365 |
+
"loss": 0.2088,
|
| 366 |
+
"step": 25500
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 2.2865183361181955,
|
| 370 |
+
"grad_norm": 7.533407211303711,
|
| 371 |
+
"learning_rate": 7.713481663881806e-06,
|
| 372 |
+
"loss": 0.2063,
|
| 373 |
+
"step": 26000
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 2.330489842582007,
|
| 377 |
+
"grad_norm": 21.629718780517578,
|
| 378 |
+
"learning_rate": 7.669510157417994e-06,
|
| 379 |
+
"loss": 0.2115,
|
| 380 |
+
"step": 26500
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 2.374461349045818,
|
| 384 |
+
"grad_norm": 15.344660758972168,
|
| 385 |
+
"learning_rate": 7.625538650954182e-06,
|
| 386 |
+
"loss": 0.2008,
|
| 387 |
+
"step": 27000
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 2.41843285550963,
|
| 391 |
+
"grad_norm": 12.559733390808105,
|
| 392 |
+
"learning_rate": 7.58156714449037e-06,
|
| 393 |
+
"loss": 0.1939,
|
| 394 |
+
"step": 27500
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 2.4624043619734413,
|
| 398 |
+
"grad_norm": 13.165969848632812,
|
| 399 |
+
"learning_rate": 7.537595638026559e-06,
|
| 400 |
+
"loss": 0.2105,
|
| 401 |
+
"step": 28000
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 2.5063758684372526,
|
| 405 |
+
"grad_norm": 12.558066368103027,
|
| 406 |
+
"learning_rate": 7.493624131562748e-06,
|
| 407 |
+
"loss": 0.1951,
|
| 408 |
+
"step": 28500
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 2.5503473749010643,
|
| 412 |
+
"grad_norm": 13.841544151306152,
|
| 413 |
+
"learning_rate": 7.449652625098936e-06,
|
| 414 |
+
"loss": 0.1997,
|
| 415 |
+
"step": 29000
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 2.5943188813648757,
|
| 419 |
+
"grad_norm": 6.036389350891113,
|
| 420 |
+
"learning_rate": 7.405681118635125e-06,
|
| 421 |
+
"loss": 0.2056,
|
| 422 |
+
"step": 29500
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 2.638290387828687,
|
| 426 |
+
"grad_norm": 24.980592727661133,
|
| 427 |
+
"learning_rate": 7.361709612171314e-06,
|
| 428 |
+
"loss": 0.2021,
|
| 429 |
+
"step": 30000
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 2.6822618942924983,
|
| 433 |
+
"grad_norm": 3.522369861602783,
|
| 434 |
+
"learning_rate": 7.317738105707503e-06,
|
| 435 |
+
"loss": 0.2109,
|
| 436 |
+
"step": 30500
|
| 437 |
+
},
|
| 438 |
+
{
|
| 439 |
+
"epoch": 2.72623340075631,
|
| 440 |
+
"grad_norm": 1.9453462362289429,
|
| 441 |
+
"learning_rate": 7.273766599243692e-06,
|
| 442 |
+
"loss": 0.1934,
|
| 443 |
+
"step": 31000
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 2.7702049072201214,
|
| 447 |
+
"grad_norm": 14.508398056030273,
|
| 448 |
+
"learning_rate": 7.229795092779879e-06,
|
| 449 |
+
"loss": 0.2064,
|
| 450 |
+
"step": 31500
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 2.8141764136839327,
|
| 454 |
+
"grad_norm": 11.040177345275879,
|
| 455 |
+
"learning_rate": 7.1858235863160675e-06,
|
| 456 |
+
"loss": 0.2016,
|
| 457 |
+
"step": 32000
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 2.8581479201477444,
|
| 461 |
+
"grad_norm": 7.400022506713867,
|
| 462 |
+
"learning_rate": 7.141852079852256e-06,
|
| 463 |
+
"loss": 0.2043,
|
| 464 |
+
"step": 32500
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 2.9021194266115558,
|
| 468 |
+
"grad_norm": 5.055144309997559,
|
| 469 |
+
"learning_rate": 7.097880573388445e-06,
|
| 470 |
+
"loss": 0.1952,
|
| 471 |
+
"step": 33000
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 2.946090933075367,
|
| 475 |
+
"grad_norm": 10.668570518493652,
|
| 476 |
+
"learning_rate": 7.053909066924634e-06,
|
| 477 |
+
"loss": 0.1949,
|
| 478 |
+
"step": 33500
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 2.9900624395391784,
|
| 482 |
+
"grad_norm": 2.9759585857391357,
|
| 483 |
+
"learning_rate": 7.009937560460822e-06,
|
| 484 |
+
"loss": 0.1987,
|
| 485 |
+
"step": 34000
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 3.03403394600299,
|
| 489 |
+
"grad_norm": 17.75304412841797,
|
| 490 |
+
"learning_rate": 6.965966053997011e-06,
|
| 491 |
+
"loss": 0.1568,
|
| 492 |
+
"step": 34500
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 3.0780054524668015,
|
| 496 |
+
"grad_norm": 30.556795120239258,
|
| 497 |
+
"learning_rate": 6.921994547533199e-06,
|
| 498 |
+
"loss": 0.1476,
|
| 499 |
+
"step": 35000
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 3.1219769589306128,
|
| 503 |
+
"grad_norm": 5.082170486450195,
|
| 504 |
+
"learning_rate": 6.878023041069387e-06,
|
| 505 |
+
"loss": 0.1513,
|
| 506 |
+
"step": 35500
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 3.1659484653944245,
|
| 510 |
+
"grad_norm": 1.4760559797286987,
|
| 511 |
+
"learning_rate": 6.834051534605576e-06,
|
| 512 |
+
"loss": 0.1551,
|
| 513 |
+
"step": 36000
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 3.209919971858236,
|
| 517 |
+
"grad_norm": 2.336458683013916,
|
| 518 |
+
"learning_rate": 6.790080028141764e-06,
|
| 519 |
+
"loss": 0.1554,
|
| 520 |
+
"step": 36500
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 3.253891478322047,
|
| 524 |
+
"grad_norm": 23.16608428955078,
|
| 525 |
+
"learning_rate": 6.746108521677953e-06,
|
| 526 |
+
"loss": 0.1529,
|
| 527 |
+
"step": 37000
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 3.297862984785859,
|
| 531 |
+
"grad_norm": 10.948084831237793,
|
| 532 |
+
"learning_rate": 6.702137015214142e-06,
|
| 533 |
+
"loss": 0.1583,
|
| 534 |
+
"step": 37500
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 3.3418344912496702,
|
| 538 |
+
"grad_norm": 15.917346954345703,
|
| 539 |
+
"learning_rate": 6.6581655087503305e-06,
|
| 540 |
+
"loss": 0.167,
|
| 541 |
+
"step": 38000
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 3.3858059977134816,
|
| 545 |
+
"grad_norm": 1.641739845275879,
|
| 546 |
+
"learning_rate": 6.614194002286519e-06,
|
| 547 |
+
"loss": 0.1539,
|
| 548 |
+
"step": 38500
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 3.4297775041772933,
|
| 552 |
+
"grad_norm": 0.5845322608947754,
|
| 553 |
+
"learning_rate": 6.570222495822708e-06,
|
| 554 |
+
"loss": 0.1576,
|
| 555 |
+
"step": 39000
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 3.4737490106411046,
|
| 559 |
+
"grad_norm": 16.46019744873047,
|
| 560 |
+
"learning_rate": 6.526250989358895e-06,
|
| 561 |
+
"loss": 0.1561,
|
| 562 |
+
"step": 39500
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 3.517720517104916,
|
| 566 |
+
"grad_norm": 31.895326614379883,
|
| 567 |
+
"learning_rate": 6.482279482895084e-06,
|
| 568 |
+
"loss": 0.1459,
|
| 569 |
+
"step": 40000
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 3.5616920235687273,
|
| 573 |
+
"grad_norm": 3.9819185733795166,
|
| 574 |
+
"learning_rate": 6.438307976431273e-06,
|
| 575 |
+
"loss": 0.1623,
|
| 576 |
+
"step": 40500
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 3.605663530032539,
|
| 580 |
+
"grad_norm": 13.1094970703125,
|
| 581 |
+
"learning_rate": 6.3943364699674615e-06,
|
| 582 |
+
"loss": 0.1566,
|
| 583 |
+
"step": 41000
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 3.6496350364963503,
|
| 587 |
+
"grad_norm": 6.929986953735352,
|
| 588 |
+
"learning_rate": 6.35036496350365e-06,
|
| 589 |
+
"loss": 0.1656,
|
| 590 |
+
"step": 41500
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 3.6936065429601617,
|
| 594 |
+
"grad_norm": 7.520695686340332,
|
| 595 |
+
"learning_rate": 6.306393457039839e-06,
|
| 596 |
+
"loss": 0.1599,
|
| 597 |
+
"step": 42000
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 3.7375780494239734,
|
| 601 |
+
"grad_norm": 11.514050483703613,
|
| 602 |
+
"learning_rate": 6.262421950576028e-06,
|
| 603 |
+
"loss": 0.1664,
|
| 604 |
+
"step": 42500
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 3.7815495558877847,
|
| 608 |
+
"grad_norm": 1.5625334978103638,
|
| 609 |
+
"learning_rate": 6.218450444112216e-06,
|
| 610 |
+
"loss": 0.1603,
|
| 611 |
+
"step": 43000
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 3.825521062351596,
|
| 615 |
+
"grad_norm": 22.2111873626709,
|
| 616 |
+
"learning_rate": 6.174478937648405e-06,
|
| 617 |
+
"loss": 0.1645,
|
| 618 |
+
"step": 43500
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 3.8694925688154074,
|
| 622 |
+
"grad_norm": 9.370536804199219,
|
| 623 |
+
"learning_rate": 6.130507431184593e-06,
|
| 624 |
+
"loss": 0.153,
|
| 625 |
+
"step": 44000
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 3.913464075279219,
|
| 629 |
+
"grad_norm": 1.3509089946746826,
|
| 630 |
+
"learning_rate": 6.086535924720781e-06,
|
| 631 |
+
"loss": 0.1605,
|
| 632 |
+
"step": 44500
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 3.9574355817430305,
|
| 636 |
+
"grad_norm": 1.670966386795044,
|
| 637 |
+
"learning_rate": 6.04256441825697e-06,
|
| 638 |
+
"loss": 0.1602,
|
| 639 |
+
"step": 45000
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 4.001407088206842,
|
| 643 |
+
"grad_norm": 22.55781364440918,
|
| 644 |
+
"learning_rate": 5.998592911793158e-06,
|
| 645 |
+
"loss": 0.1582,
|
| 646 |
+
"step": 45500
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"epoch": 4.0453785946706535,
|
| 650 |
+
"grad_norm": 18.12227439880371,
|
| 651 |
+
"learning_rate": 5.954621405329347e-06,
|
| 652 |
+
"loss": 0.1256,
|
| 653 |
+
"step": 46000
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 4.089350101134465,
|
| 657 |
+
"grad_norm": 11.872714042663574,
|
| 658 |
+
"learning_rate": 5.910649898865536e-06,
|
| 659 |
+
"loss": 0.1241,
|
| 660 |
+
"step": 46500
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 4.133321607598276,
|
| 664 |
+
"grad_norm": 26.38855743408203,
|
| 665 |
+
"learning_rate": 5.8666783924017245e-06,
|
| 666 |
+
"loss": 0.1301,
|
| 667 |
+
"step": 47000
|
| 668 |
+
},
|
| 669 |
+
{
|
| 670 |
+
"epoch": 4.1772931140620875,
|
| 671 |
+
"grad_norm": 25.3267879486084,
|
| 672 |
+
"learning_rate": 5.822706885937913e-06,
|
| 673 |
+
"loss": 0.1236,
|
| 674 |
+
"step": 47500
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 4.221264620525899,
|
| 678 |
+
"grad_norm": 27.222675323486328,
|
| 679 |
+
"learning_rate": 5.7787353794741e-06,
|
| 680 |
+
"loss": 0.1217,
|
| 681 |
+
"step": 48000
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 4.265236126989711,
|
| 685 |
+
"grad_norm": 53.458518981933594,
|
| 686 |
+
"learning_rate": 5.734763873010289e-06,
|
| 687 |
+
"loss": 0.1315,
|
| 688 |
+
"step": 48500
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"epoch": 4.309207633453522,
|
| 692 |
+
"grad_norm": 27.357641220092773,
|
| 693 |
+
"learning_rate": 5.690792366546478e-06,
|
| 694 |
+
"loss": 0.1253,
|
| 695 |
+
"step": 49000
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 4.353179139917334,
|
| 699 |
+
"grad_norm": 18.440649032592773,
|
| 700 |
+
"learning_rate": 5.646820860082667e-06,
|
| 701 |
+
"loss": 0.132,
|
| 702 |
+
"step": 49500
|
| 703 |
+
},
|
| 704 |
+
{
|
| 705 |
+
"epoch": 4.397150646381145,
|
| 706 |
+
"grad_norm": 8.659114837646484,
|
| 707 |
+
"learning_rate": 5.6028493536188556e-06,
|
| 708 |
+
"loss": 0.1262,
|
| 709 |
+
"step": 50000
|
| 710 |
+
},
|
| 711 |
+
{
|
| 712 |
+
"epoch": 4.441122152844956,
|
| 713 |
+
"grad_norm": 22.664106369018555,
|
| 714 |
+
"learning_rate": 5.558877847155044e-06,
|
| 715 |
+
"loss": 0.1185,
|
| 716 |
+
"step": 50500
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 4.485093659308768,
|
| 720 |
+
"grad_norm": 2.488041400909424,
|
| 721 |
+
"learning_rate": 5.514906340691233e-06,
|
| 722 |
+
"loss": 0.1264,
|
| 723 |
+
"step": 51000
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 4.529065165772579,
|
| 727 |
+
"grad_norm": 35.29222869873047,
|
| 728 |
+
"learning_rate": 5.470934834227422e-06,
|
| 729 |
+
"loss": 0.134,
|
| 730 |
+
"step": 51500
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 4.573036672236391,
|
| 734 |
+
"grad_norm": 25.144588470458984,
|
| 735 |
+
"learning_rate": 5.426963327763609e-06,
|
| 736 |
+
"loss": 0.1265,
|
| 737 |
+
"step": 52000
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 4.617008178700202,
|
| 741 |
+
"grad_norm": 1.2160580158233643,
|
| 742 |
+
"learning_rate": 5.382991821299798e-06,
|
| 743 |
+
"loss": 0.1306,
|
| 744 |
+
"step": 52500
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 4.660979685164014,
|
| 748 |
+
"grad_norm": 22.59756088256836,
|
| 749 |
+
"learning_rate": 5.339020314835987e-06,
|
| 750 |
+
"loss": 0.127,
|
| 751 |
+
"step": 53000
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 4.704951191627825,
|
| 755 |
+
"grad_norm": 17.720712661743164,
|
| 756 |
+
"learning_rate": 5.295048808372175e-06,
|
| 757 |
+
"loss": 0.1233,
|
| 758 |
+
"step": 53500
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 4.748922698091636,
|
| 762 |
+
"grad_norm": 24.757396697998047,
|
| 763 |
+
"learning_rate": 5.251077301908364e-06,
|
| 764 |
+
"loss": 0.135,
|
| 765 |
+
"step": 54000
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 4.792894204555449,
|
| 769 |
+
"grad_norm": 13.89427375793457,
|
| 770 |
+
"learning_rate": 5.207105795444552e-06,
|
| 771 |
+
"loss": 0.1232,
|
| 772 |
+
"step": 54500
|
| 773 |
+
},
|
| 774 |
+
{
|
| 775 |
+
"epoch": 4.83686571101926,
|
| 776 |
+
"grad_norm": 6.014657974243164,
|
| 777 |
+
"learning_rate": 5.163134288980741e-06,
|
| 778 |
+
"loss": 0.1322,
|
| 779 |
+
"step": 55000
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 4.880837217483071,
|
| 783 |
+
"grad_norm": 16.53226089477539,
|
| 784 |
+
"learning_rate": 5.11916278251693e-06,
|
| 785 |
+
"loss": 0.1254,
|
| 786 |
+
"step": 55500
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 4.9248087239468825,
|
| 790 |
+
"grad_norm": 4.393960952758789,
|
| 791 |
+
"learning_rate": 5.075191276053118e-06,
|
| 792 |
+
"loss": 0.1343,
|
| 793 |
+
"step": 56000
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 4.968780230410694,
|
| 797 |
+
"grad_norm": 6.577863693237305,
|
| 798 |
+
"learning_rate": 5.0312197695893065e-06,
|
| 799 |
+
"loss": 0.1302,
|
| 800 |
+
"step": 56500
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 5.012751736874505,
|
| 804 |
+
"grad_norm": 0.385308176279068,
|
| 805 |
+
"learning_rate": 4.987248263125494e-06,
|
| 806 |
+
"loss": 0.1134,
|
| 807 |
+
"step": 57000
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 5.0567232433383165,
|
| 811 |
+
"grad_norm": 17.352643966674805,
|
| 812 |
+
"learning_rate": 4.943276756661683e-06,
|
| 813 |
+
"loss": 0.1024,
|
| 814 |
+
"step": 57500
|
| 815 |
+
},
|
| 816 |
+
{
|
| 817 |
+
"epoch": 5.100694749802129,
|
| 818 |
+
"grad_norm": 44.88164520263672,
|
| 819 |
+
"learning_rate": 4.899305250197872e-06,
|
| 820 |
+
"loss": 0.1095,
|
| 821 |
+
"step": 58000
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 5.14466625626594,
|
| 825 |
+
"grad_norm": 2.88139009475708,
|
| 826 |
+
"learning_rate": 4.855333743734061e-06,
|
| 827 |
+
"loss": 0.1022,
|
| 828 |
+
"step": 58500
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"epoch": 5.188637762729751,
|
| 832 |
+
"grad_norm": 30.369657516479492,
|
| 833 |
+
"learning_rate": 4.81136223727025e-06,
|
| 834 |
+
"loss": 0.1029,
|
| 835 |
+
"step": 59000
|
| 836 |
+
},
|
| 837 |
+
{
|
| 838 |
+
"epoch": 5.232609269193563,
|
| 839 |
+
"grad_norm": 21.04435920715332,
|
| 840 |
+
"learning_rate": 4.7673907308064375e-06,
|
| 841 |
+
"loss": 0.0973,
|
| 842 |
+
"step": 59500
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 5.276580775657374,
|
| 846 |
+
"grad_norm": 0.3326770067214966,
|
| 847 |
+
"learning_rate": 4.723419224342626e-06,
|
| 848 |
+
"loss": 0.0969,
|
| 849 |
+
"step": 60000
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 5.320552282121185,
|
| 853 |
+
"grad_norm": 26.21080207824707,
|
| 854 |
+
"learning_rate": 4.679447717878815e-06,
|
| 855 |
+
"loss": 0.1072,
|
| 856 |
+
"step": 60500
|
| 857 |
+
},
|
| 858 |
+
{
|
| 859 |
+
"epoch": 5.364523788584997,
|
| 860 |
+
"grad_norm": 66.27179718017578,
|
| 861 |
+
"learning_rate": 4.635476211415004e-06,
|
| 862 |
+
"loss": 0.11,
|
| 863 |
+
"step": 61000
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 5.408495295048809,
|
| 867 |
+
"grad_norm": 0.6907067894935608,
|
| 868 |
+
"learning_rate": 4.591504704951192e-06,
|
| 869 |
+
"loss": 0.1077,
|
| 870 |
+
"step": 61500
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"epoch": 5.45246680151262,
|
| 874 |
+
"grad_norm": 18.443483352661133,
|
| 875 |
+
"learning_rate": 4.547533198487381e-06,
|
| 876 |
+
"loss": 0.0996,
|
| 877 |
+
"step": 62000
|
| 878 |
+
},
|
| 879 |
+
{
|
| 880 |
+
"epoch": 5.496438307976431,
|
| 881 |
+
"grad_norm": 30.685518264770508,
|
| 882 |
+
"learning_rate": 4.5035616920235695e-06,
|
| 883 |
+
"loss": 0.0937,
|
| 884 |
+
"step": 62500
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 5.540409814440243,
|
| 888 |
+
"grad_norm": 8.975428581237793,
|
| 889 |
+
"learning_rate": 4.459590185559758e-06,
|
| 890 |
+
"loss": 0.1074,
|
| 891 |
+
"step": 63000
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 5.584381320904054,
|
| 895 |
+
"grad_norm": 21.28555679321289,
|
| 896 |
+
"learning_rate": 4.415618679095946e-06,
|
| 897 |
+
"loss": 0.0958,
|
| 898 |
+
"step": 63500
|
| 899 |
+
},
|
| 900 |
+
{
|
| 901 |
+
"epoch": 5.628352827367865,
|
| 902 |
+
"grad_norm": 35.58885192871094,
|
| 903 |
+
"learning_rate": 4.371647172632135e-06,
|
| 904 |
+
"loss": 0.1072,
|
| 905 |
+
"step": 64000
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 5.672324333831677,
|
| 909 |
+
"grad_norm": 31.6656436920166,
|
| 910 |
+
"learning_rate": 4.327675666168324e-06,
|
| 911 |
+
"loss": 0.1075,
|
| 912 |
+
"step": 64500
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 5.716295840295489,
|
| 916 |
+
"grad_norm": 1.064644455909729,
|
| 917 |
+
"learning_rate": 4.283704159704512e-06,
|
| 918 |
+
"loss": 0.1069,
|
| 919 |
+
"step": 65000
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 5.7602673467593,
|
| 923 |
+
"grad_norm": 8.807537078857422,
|
| 924 |
+
"learning_rate": 4.2397326532407005e-06,
|
| 925 |
+
"loss": 0.1059,
|
| 926 |
+
"step": 65500
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 5.8042388532231115,
|
| 930 |
+
"grad_norm": 6.445720672607422,
|
| 931 |
+
"learning_rate": 4.1957611467768885e-06,
|
| 932 |
+
"loss": 0.1046,
|
| 933 |
+
"step": 66000
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 5.848210359686923,
|
| 937 |
+
"grad_norm": 10.035616874694824,
|
| 938 |
+
"learning_rate": 4.151789640313077e-06,
|
| 939 |
+
"loss": 0.1061,
|
| 940 |
+
"step": 66500
|
| 941 |
+
},
|
| 942 |
+
{
|
| 943 |
+
"epoch": 5.892181866150734,
|
| 944 |
+
"grad_norm": 0.3110928535461426,
|
| 945 |
+
"learning_rate": 4.107818133849266e-06,
|
| 946 |
+
"loss": 0.1087,
|
| 947 |
+
"step": 67000
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 5.9361533726145455,
|
| 951 |
+
"grad_norm": 1.1699435710906982,
|
| 952 |
+
"learning_rate": 4.063846627385454e-06,
|
| 953 |
+
"loss": 0.101,
|
| 954 |
+
"step": 67500
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 5.980124879078357,
|
| 958 |
+
"grad_norm": 4.037410736083984,
|
| 959 |
+
"learning_rate": 4.019875120921643e-06,
|
| 960 |
+
"loss": 0.1123,
|
| 961 |
+
"step": 68000
|
| 962 |
+
},
|
| 963 |
+
{
|
| 964 |
+
"epoch": 6.024096385542169,
|
| 965 |
+
"grad_norm": 17.392499923706055,
|
| 966 |
+
"learning_rate": 3.975903614457832e-06,
|
| 967 |
+
"loss": 0.0904,
|
| 968 |
+
"step": 68500
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 6.06806789200598,
|
| 972 |
+
"grad_norm": 2.474785089492798,
|
| 973 |
+
"learning_rate": 3.93193210799402e-06,
|
| 974 |
+
"loss": 0.078,
|
| 975 |
+
"step": 69000
|
| 976 |
+
},
|
| 977 |
+
{
|
| 978 |
+
"epoch": 6.112039398469792,
|
| 979 |
+
"grad_norm": 6.585563659667969,
|
| 980 |
+
"learning_rate": 3.887960601530208e-06,
|
| 981 |
+
"loss": 0.0805,
|
| 982 |
+
"step": 69500
|
| 983 |
+
},
|
| 984 |
+
{
|
| 985 |
+
"epoch": 6.156010904933603,
|
| 986 |
+
"grad_norm": 22.513330459594727,
|
| 987 |
+
"learning_rate": 3.843989095066397e-06,
|
| 988 |
+
"loss": 0.0839,
|
| 989 |
+
"step": 70000
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 6.199982411397414,
|
| 993 |
+
"grad_norm": 13.566338539123535,
|
| 994 |
+
"learning_rate": 3.800017588602586e-06,
|
| 995 |
+
"loss": 0.0812,
|
| 996 |
+
"step": 70500
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 6.2439539178612256,
|
| 1000 |
+
"grad_norm": 0.06740322709083557,
|
| 1001 |
+
"learning_rate": 3.7560460821387747e-06,
|
| 1002 |
+
"loss": 0.085,
|
| 1003 |
+
"step": 71000
|
| 1004 |
+
},
|
| 1005 |
+
{
|
| 1006 |
+
"epoch": 6.287925424325038,
|
| 1007 |
+
"grad_norm": 7.70227575302124,
|
| 1008 |
+
"learning_rate": 3.7120745756749626e-06,
|
| 1009 |
+
"loss": 0.0797,
|
| 1010 |
+
"step": 71500
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 6.331896930788849,
|
| 1014 |
+
"grad_norm": 0.8043323159217834,
|
| 1015 |
+
"learning_rate": 3.6681030692111514e-06,
|
| 1016 |
+
"loss": 0.083,
|
| 1017 |
+
"step": 72000
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 6.37586843725266,
|
| 1021 |
+
"grad_norm": 0.23601041734218597,
|
| 1022 |
+
"learning_rate": 3.62413156274734e-06,
|
| 1023 |
+
"loss": 0.0881,
|
| 1024 |
+
"step": 72500
|
| 1025 |
+
},
|
| 1026 |
+
{
|
| 1027 |
+
"epoch": 6.419839943716472,
|
| 1028 |
+
"grad_norm": 24.715349197387695,
|
| 1029 |
+
"learning_rate": 3.5801600562835286e-06,
|
| 1030 |
+
"loss": 0.0845,
|
| 1031 |
+
"step": 73000
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 6.463811450180283,
|
| 1035 |
+
"grad_norm": 32.790096282958984,
|
| 1036 |
+
"learning_rate": 3.5361885498197174e-06,
|
| 1037 |
+
"loss": 0.0873,
|
| 1038 |
+
"step": 73500
|
| 1039 |
+
},
|
| 1040 |
+
{
|
| 1041 |
+
"epoch": 6.507782956644094,
|
| 1042 |
+
"grad_norm": 0.44961002469062805,
|
| 1043 |
+
"learning_rate": 3.4922170433559053e-06,
|
| 1044 |
+
"loss": 0.0926,
|
| 1045 |
+
"step": 74000
|
| 1046 |
+
},
|
| 1047 |
+
{
|
| 1048 |
+
"epoch": 6.5517544631079065,
|
| 1049 |
+
"grad_norm": 30.2086181640625,
|
| 1050 |
+
"learning_rate": 3.448245536892094e-06,
|
| 1051 |
+
"loss": 0.0869,
|
| 1052 |
+
"step": 74500
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 6.595725969571718,
|
| 1056 |
+
"grad_norm": 21.381420135498047,
|
| 1057 |
+
"learning_rate": 3.404274030428283e-06,
|
| 1058 |
+
"loss": 0.0897,
|
| 1059 |
+
"step": 75000
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 6.639697476035529,
|
| 1063 |
+
"grad_norm": 9.947957038879395,
|
| 1064 |
+
"learning_rate": 3.3603025239644717e-06,
|
| 1065 |
+
"loss": 0.0862,
|
| 1066 |
+
"step": 75500
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 6.6836689824993405,
|
| 1070 |
+
"grad_norm": 0.4237179160118103,
|
| 1071 |
+
"learning_rate": 3.3163310175006597e-06,
|
| 1072 |
+
"loss": 0.0839,
|
| 1073 |
+
"step": 76000
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 6.727640488963152,
|
| 1077 |
+
"grad_norm": 8.52582836151123,
|
| 1078 |
+
"learning_rate": 3.2723595110368485e-06,
|
| 1079 |
+
"loss": 0.0838,
|
| 1080 |
+
"step": 76500
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 6.771611995426963,
|
| 1084 |
+
"grad_norm": 21.18242073059082,
|
| 1085 |
+
"learning_rate": 3.228388004573037e-06,
|
| 1086 |
+
"loss": 0.0858,
|
| 1087 |
+
"step": 77000
|
| 1088 |
+
},
|
| 1089 |
+
{
|
| 1090 |
+
"epoch": 6.815583501890774,
|
| 1091 |
+
"grad_norm": 42.046844482421875,
|
| 1092 |
+
"learning_rate": 3.1844164981092256e-06,
|
| 1093 |
+
"loss": 0.0825,
|
| 1094 |
+
"step": 77500
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 6.859555008354587,
|
| 1098 |
+
"grad_norm": 3.5966274738311768,
|
| 1099 |
+
"learning_rate": 3.140444991645414e-06,
|
| 1100 |
+
"loss": 0.0873,
|
| 1101 |
+
"step": 78000
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 6.903526514818398,
|
| 1105 |
+
"grad_norm": 25.500396728515625,
|
| 1106 |
+
"learning_rate": 3.0964734851816024e-06,
|
| 1107 |
+
"loss": 0.0885,
|
| 1108 |
+
"step": 78500
|
| 1109 |
+
},
|
| 1110 |
+
{
|
| 1111 |
+
"epoch": 6.947498021282209,
|
| 1112 |
+
"grad_norm": 59.15459442138672,
|
| 1113 |
+
"learning_rate": 3.052501978717791e-06,
|
| 1114 |
+
"loss": 0.0855,
|
| 1115 |
+
"step": 79000
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 6.991469527746021,
|
| 1119 |
+
"grad_norm": 0.11808668822050095,
|
| 1120 |
+
"learning_rate": 3.00853047225398e-06,
|
| 1121 |
+
"loss": 0.0837,
|
| 1122 |
+
"step": 79500
|
| 1123 |
+
},
|
| 1124 |
+
{
|
| 1125 |
+
"epoch": 7.035441034209832,
|
| 1126 |
+
"grad_norm": 0.7973224520683289,
|
| 1127 |
+
"learning_rate": 2.964558965790168e-06,
|
| 1128 |
+
"loss": 0.0648,
|
| 1129 |
+
"step": 80000
|
| 1130 |
+
},
|
| 1131 |
+
{
|
| 1132 |
+
"epoch": 7.079412540673643,
|
| 1133 |
+
"grad_norm": 46.84673309326172,
|
| 1134 |
+
"learning_rate": 2.9205874593263567e-06,
|
| 1135 |
+
"loss": 0.0709,
|
| 1136 |
+
"step": 80500
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 7.1233840471374545,
|
| 1140 |
+
"grad_norm": 0.13476693630218506,
|
| 1141 |
+
"learning_rate": 2.8766159528625455e-06,
|
| 1142 |
+
"loss": 0.0751,
|
| 1143 |
+
"step": 81000
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 7.167355553601267,
|
| 1147 |
+
"grad_norm": 30.29783821105957,
|
| 1148 |
+
"learning_rate": 2.832644446398734e-06,
|
| 1149 |
+
"loss": 0.0677,
|
| 1150 |
+
"step": 81500
|
| 1151 |
+
},
|
| 1152 |
+
{
|
| 1153 |
+
"epoch": 7.211327060065078,
|
| 1154 |
+
"grad_norm": 12.256621360778809,
|
| 1155 |
+
"learning_rate": 2.7886729399349222e-06,
|
| 1156 |
+
"loss": 0.0717,
|
| 1157 |
+
"step": 82000
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 7.255298566528889,
|
| 1161 |
+
"grad_norm": 18.782052993774414,
|
| 1162 |
+
"learning_rate": 2.744701433471111e-06,
|
| 1163 |
+
"loss": 0.0652,
|
| 1164 |
+
"step": 82500
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"epoch": 7.299270072992701,
|
| 1168 |
+
"grad_norm": 4.884202480316162,
|
| 1169 |
+
"learning_rate": 2.7007299270072994e-06,
|
| 1170 |
+
"loss": 0.0702,
|
| 1171 |
+
"step": 83000
|
| 1172 |
+
},
|
| 1173 |
+
{
|
| 1174 |
+
"epoch": 7.343241579456512,
|
| 1175 |
+
"grad_norm": 0.8015443682670593,
|
| 1176 |
+
"learning_rate": 2.656758420543488e-06,
|
| 1177 |
+
"loss": 0.0742,
|
| 1178 |
+
"step": 83500
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 7.387213085920323,
|
| 1182 |
+
"grad_norm": 0.11882288008928299,
|
| 1183 |
+
"learning_rate": 2.6127869140796765e-06,
|
| 1184 |
+
"loss": 0.0677,
|
| 1185 |
+
"step": 84000
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 7.431184592384135,
|
| 1189 |
+
"grad_norm": 24.17609214782715,
|
| 1190 |
+
"learning_rate": 2.568815407615865e-06,
|
| 1191 |
+
"loss": 0.0693,
|
| 1192 |
+
"step": 84500
|
| 1193 |
+
},
|
| 1194 |
+
{
|
| 1195 |
+
"epoch": 7.475156098847947,
|
| 1196 |
+
"grad_norm": 8.958883285522461,
|
| 1197 |
+
"learning_rate": 2.5248439011520537e-06,
|
| 1198 |
+
"loss": 0.0754,
|
| 1199 |
+
"step": 85000
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 7.519127605311758,
|
| 1203 |
+
"grad_norm": 0.03229560703039169,
|
| 1204 |
+
"learning_rate": 2.480872394688242e-06,
|
| 1205 |
+
"loss": 0.0696,
|
| 1206 |
+
"step": 85500
|
| 1207 |
+
},
|
| 1208 |
+
{
|
| 1209 |
+
"epoch": 7.5630991117755695,
|
| 1210 |
+
"grad_norm": 26.437501907348633,
|
| 1211 |
+
"learning_rate": 2.436900888224431e-06,
|
| 1212 |
+
"loss": 0.0704,
|
| 1213 |
+
"step": 86000
|
| 1214 |
+
},
|
| 1215 |
+
{
|
| 1216 |
+
"epoch": 7.607070618239381,
|
| 1217 |
+
"grad_norm": 33.26008605957031,
|
| 1218 |
+
"learning_rate": 2.3929293817606197e-06,
|
| 1219 |
+
"loss": 0.0679,
|
| 1220 |
+
"step": 86500
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 7.651042124703192,
|
| 1224 |
+
"grad_norm": 4.0587334632873535,
|
| 1225 |
+
"learning_rate": 2.348957875296808e-06,
|
| 1226 |
+
"loss": 0.0727,
|
| 1227 |
+
"step": 87000
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 7.695013631167003,
|
| 1231 |
+
"grad_norm": 1.8055332899093628,
|
| 1232 |
+
"learning_rate": 2.3049863688329964e-06,
|
| 1233 |
+
"loss": 0.0714,
|
| 1234 |
+
"step": 87500
|
| 1235 |
+
},
|
| 1236 |
+
{
|
| 1237 |
+
"epoch": 7.738985137630815,
|
| 1238 |
+
"grad_norm": 0.08867678046226501,
|
| 1239 |
+
"learning_rate": 2.2610148623691848e-06,
|
| 1240 |
+
"loss": 0.0714,
|
| 1241 |
+
"step": 88000
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 7.782956644094627,
|
| 1245 |
+
"grad_norm": 51.05652618408203,
|
| 1246 |
+
"learning_rate": 2.2170433559053736e-06,
|
| 1247 |
+
"loss": 0.0662,
|
| 1248 |
+
"step": 88500
|
| 1249 |
+
},
|
| 1250 |
+
{
|
| 1251 |
+
"epoch": 7.826928150558438,
|
| 1252 |
+
"grad_norm": 0.3058519959449768,
|
| 1253 |
+
"learning_rate": 2.173071849441562e-06,
|
| 1254 |
+
"loss": 0.0637,
|
| 1255 |
+
"step": 89000
|
| 1256 |
+
},
|
| 1257 |
+
{
|
| 1258 |
+
"epoch": 7.87089965702225,
|
| 1259 |
+
"grad_norm": 0.05084119364619255,
|
| 1260 |
+
"learning_rate": 2.1291003429777507e-06,
|
| 1261 |
+
"loss": 0.0676,
|
| 1262 |
+
"step": 89500
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 7.914871163486061,
|
| 1266 |
+
"grad_norm": 0.23956036567687988,
|
| 1267 |
+
"learning_rate": 2.085128836513939e-06,
|
| 1268 |
+
"loss": 0.0818,
|
| 1269 |
+
"step": 90000
|
| 1270 |
+
},
|
| 1271 |
+
{
|
| 1272 |
+
"epoch": 7.958842669949872,
|
| 1273 |
+
"grad_norm": 15.964754104614258,
|
| 1274 |
+
"learning_rate": 2.041157330050128e-06,
|
| 1275 |
+
"loss": 0.076,
|
| 1276 |
+
"step": 90500
|
| 1277 |
+
},
|
| 1278 |
+
{
|
| 1279 |
+
"epoch": 8.002814176413684,
|
| 1280 |
+
"grad_norm": 1.4704182147979736,
|
| 1281 |
+
"learning_rate": 1.9971858235863163e-06,
|
| 1282 |
+
"loss": 0.0664,
|
| 1283 |
+
"step": 91000
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 8.046785682877495,
|
| 1287 |
+
"grad_norm": 63.657440185546875,
|
| 1288 |
+
"learning_rate": 1.953214317122505e-06,
|
| 1289 |
+
"loss": 0.0583,
|
| 1290 |
+
"step": 91500
|
| 1291 |
+
},
|
| 1292 |
+
{
|
| 1293 |
+
"epoch": 8.090757189341307,
|
| 1294 |
+
"grad_norm": 8.421917915344238,
|
| 1295 |
+
"learning_rate": 1.9092428106586934e-06,
|
| 1296 |
+
"loss": 0.0613,
|
| 1297 |
+
"step": 92000
|
| 1298 |
+
},
|
| 1299 |
+
{
|
| 1300 |
+
"epoch": 8.134728695805117,
|
| 1301 |
+
"grad_norm": 0.17341230809688568,
|
| 1302 |
+
"learning_rate": 1.865271304194882e-06,
|
| 1303 |
+
"loss": 0.0554,
|
| 1304 |
+
"step": 92500
|
| 1305 |
+
},
|
| 1306 |
+
{
|
| 1307 |
+
"epoch": 8.17870020226893,
|
| 1308 |
+
"grad_norm": 0.09282279014587402,
|
| 1309 |
+
"learning_rate": 1.8212997977310704e-06,
|
| 1310 |
+
"loss": 0.0607,
|
| 1311 |
+
"step": 93000
|
| 1312 |
+
},
|
| 1313 |
+
{
|
| 1314 |
+
"epoch": 8.222671708732742,
|
| 1315 |
+
"grad_norm": 0.0653434619307518,
|
| 1316 |
+
"learning_rate": 1.777328291267259e-06,
|
| 1317 |
+
"loss": 0.058,
|
| 1318 |
+
"step": 93500
|
| 1319 |
+
},
|
| 1320 |
+
{
|
| 1321 |
+
"epoch": 8.266643215196552,
|
| 1322 |
+
"grad_norm": 20.4820556640625,
|
| 1323 |
+
"learning_rate": 1.7333567848034475e-06,
|
| 1324 |
+
"loss": 0.0564,
|
| 1325 |
+
"step": 94000
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 8.310614721660365,
|
| 1329 |
+
"grad_norm": 2.014326333999634,
|
| 1330 |
+
"learning_rate": 1.6893852783396361e-06,
|
| 1331 |
+
"loss": 0.0589,
|
| 1332 |
+
"step": 94500
|
| 1333 |
+
},
|
| 1334 |
+
{
|
| 1335 |
+
"epoch": 8.354586228124175,
|
| 1336 |
+
"grad_norm": 0.2292569875717163,
|
| 1337 |
+
"learning_rate": 1.6454137718758245e-06,
|
| 1338 |
+
"loss": 0.0572,
|
| 1339 |
+
"step": 95000
|
| 1340 |
+
},
|
| 1341 |
+
{
|
| 1342 |
+
"epoch": 8.398557734587987,
|
| 1343 |
+
"grad_norm": 0.014584074728190899,
|
| 1344 |
+
"learning_rate": 1.6014422654120133e-06,
|
| 1345 |
+
"loss": 0.0497,
|
| 1346 |
+
"step": 95500
|
| 1347 |
+
},
|
| 1348 |
+
{
|
| 1349 |
+
"epoch": 8.442529241051798,
|
| 1350 |
+
"grad_norm": 30.090499877929688,
|
| 1351 |
+
"learning_rate": 1.5574707589482016e-06,
|
| 1352 |
+
"loss": 0.0609,
|
| 1353 |
+
"step": 96000
|
| 1354 |
+
},
|
| 1355 |
+
{
|
| 1356 |
+
"epoch": 8.48650074751561,
|
| 1357 |
+
"grad_norm": 0.038359276950359344,
|
| 1358 |
+
"learning_rate": 1.5134992524843902e-06,
|
| 1359 |
+
"loss": 0.0544,
|
| 1360 |
+
"step": 96500
|
| 1361 |
+
},
|
| 1362 |
+
{
|
| 1363 |
+
"epoch": 8.530472253979422,
|
| 1364 |
+
"grad_norm": 0.1624363213777542,
|
| 1365 |
+
"learning_rate": 1.4695277460205786e-06,
|
| 1366 |
+
"loss": 0.0578,
|
| 1367 |
+
"step": 97000
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 8.574443760443232,
|
| 1371 |
+
"grad_norm": 0.23937764763832092,
|
| 1372 |
+
"learning_rate": 1.4255562395567674e-06,
|
| 1373 |
+
"loss": 0.058,
|
| 1374 |
+
"step": 97500
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"epoch": 8.618415266907045,
|
| 1378 |
+
"grad_norm": 0.08367595821619034,
|
| 1379 |
+
"learning_rate": 1.3815847330929558e-06,
|
| 1380 |
+
"loss": 0.0567,
|
| 1381 |
+
"step": 98000
|
| 1382 |
+
},
|
| 1383 |
+
{
|
| 1384 |
+
"epoch": 8.662386773370855,
|
| 1385 |
+
"grad_norm": 0.07520689815282822,
|
| 1386 |
+
"learning_rate": 1.3376132266291445e-06,
|
| 1387 |
+
"loss": 0.0507,
|
| 1388 |
+
"step": 98500
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 8.706358279834667,
|
| 1392 |
+
"grad_norm": 0.20980244874954224,
|
| 1393 |
+
"learning_rate": 1.293641720165333e-06,
|
| 1394 |
+
"loss": 0.0621,
|
| 1395 |
+
"step": 99000
|
| 1396 |
+
},
|
| 1397 |
+
{
|
| 1398 |
+
"epoch": 8.75032978629848,
|
| 1399 |
+
"grad_norm": 2.3352205753326416,
|
| 1400 |
+
"learning_rate": 1.2496702137015215e-06,
|
| 1401 |
+
"loss": 0.053,
|
| 1402 |
+
"step": 99500
|
| 1403 |
+
},
|
| 1404 |
+
{
|
| 1405 |
+
"epoch": 8.79430129276229,
|
| 1406 |
+
"grad_norm": 0.1555749624967575,
|
| 1407 |
+
"learning_rate": 1.20569870723771e-06,
|
| 1408 |
+
"loss": 0.0572,
|
| 1409 |
+
"step": 100000
|
| 1410 |
+
},
|
| 1411 |
+
{
|
| 1412 |
+
"epoch": 8.838272799226102,
|
| 1413 |
+
"grad_norm": 17.412338256835938,
|
| 1414 |
+
"learning_rate": 1.1617272007738987e-06,
|
| 1415 |
+
"loss": 0.0562,
|
| 1416 |
+
"step": 100500
|
| 1417 |
+
},
|
| 1418 |
+
{
|
| 1419 |
+
"epoch": 8.882244305689913,
|
| 1420 |
+
"grad_norm": 0.07252756506204605,
|
| 1421 |
+
"learning_rate": 1.1177556943100872e-06,
|
| 1422 |
+
"loss": 0.0627,
|
| 1423 |
+
"step": 101000
|
| 1424 |
+
},
|
| 1425 |
+
{
|
| 1426 |
+
"epoch": 8.926215812153725,
|
| 1427 |
+
"grad_norm": 2.5696682929992676,
|
| 1428 |
+
"learning_rate": 1.0737841878462756e-06,
|
| 1429 |
+
"loss": 0.0593,
|
| 1430 |
+
"step": 101500
|
| 1431 |
+
},
|
| 1432 |
+
{
|
| 1433 |
+
"epoch": 8.970187318617535,
|
| 1434 |
+
"grad_norm": 38.372249603271484,
|
| 1435 |
+
"learning_rate": 1.0298126813824642e-06,
|
| 1436 |
+
"loss": 0.0533,
|
| 1437 |
+
"step": 102000
|
| 1438 |
+
},
|
| 1439 |
+
{
|
| 1440 |
+
"epoch": 9.014158825081347,
|
| 1441 |
+
"grad_norm": 0.0533902570605278,
|
| 1442 |
+
"learning_rate": 9.858411749186528e-07,
|
| 1443 |
+
"loss": 0.0542,
|
| 1444 |
+
"step": 102500
|
| 1445 |
+
},
|
| 1446 |
+
{
|
| 1447 |
+
"epoch": 9.05813033154516,
|
| 1448 |
+
"grad_norm": 1.9466789960861206,
|
| 1449 |
+
"learning_rate": 9.418696684548414e-07,
|
| 1450 |
+
"loss": 0.0524,
|
| 1451 |
+
"step": 103000
|
| 1452 |
+
},
|
| 1453 |
+
{
|
| 1454 |
+
"epoch": 9.10210183800897,
|
| 1455 |
+
"grad_norm": 38.15435791015625,
|
| 1456 |
+
"learning_rate": 8.978981619910298e-07,
|
| 1457 |
+
"loss": 0.0459,
|
| 1458 |
+
"step": 103500
|
| 1459 |
+
},
|
| 1460 |
+
{
|
| 1461 |
+
"epoch": 9.146073344472782,
|
| 1462 |
+
"grad_norm": 0.12077120691537857,
|
| 1463 |
+
"learning_rate": 8.539266555272184e-07,
|
| 1464 |
+
"loss": 0.0486,
|
| 1465 |
+
"step": 104000
|
| 1466 |
+
},
|
| 1467 |
+
{
|
| 1468 |
+
"epoch": 9.190044850936593,
|
| 1469 |
+
"grad_norm": 5.501252174377441,
|
| 1470 |
+
"learning_rate": 8.09955149063407e-07,
|
| 1471 |
+
"loss": 0.0491,
|
| 1472 |
+
"step": 104500
|
| 1473 |
+
},
|
| 1474 |
+
{
|
| 1475 |
+
"epoch": 9.234016357400405,
|
| 1476 |
+
"grad_norm": 0.05747182294726372,
|
| 1477 |
+
"learning_rate": 7.659836425995955e-07,
|
| 1478 |
+
"loss": 0.053,
|
| 1479 |
+
"step": 105000
|
| 1480 |
+
},
|
| 1481 |
+
{
|
| 1482 |
+
"epoch": 9.277987863864215,
|
| 1483 |
+
"grad_norm": 0.712147057056427,
|
| 1484 |
+
"learning_rate": 7.22012136135784e-07,
|
| 1485 |
+
"loss": 0.0612,
|
| 1486 |
+
"step": 105500
|
| 1487 |
+
},
|
| 1488 |
+
{
|
| 1489 |
+
"epoch": 9.321959370328027,
|
| 1490 |
+
"grad_norm": 79.47268676757812,
|
| 1491 |
+
"learning_rate": 6.780406296719725e-07,
|
| 1492 |
+
"loss": 0.0445,
|
| 1493 |
+
"step": 106000
|
| 1494 |
+
},
|
| 1495 |
+
{
|
| 1496 |
+
"epoch": 9.36593087679184,
|
| 1497 |
+
"grad_norm": 0.05525458604097366,
|
| 1498 |
+
"learning_rate": 6.340691232081611e-07,
|
| 1499 |
+
"loss": 0.0484,
|
| 1500 |
+
"step": 106500
|
| 1501 |
+
},
|
| 1502 |
+
{
|
| 1503 |
+
"epoch": 9.40990238325565,
|
| 1504 |
+
"grad_norm": 80.07630157470703,
|
| 1505 |
+
"learning_rate": 5.900976167443497e-07,
|
| 1506 |
+
"loss": 0.0486,
|
| 1507 |
+
"step": 107000
|
| 1508 |
+
},
|
| 1509 |
+
{
|
| 1510 |
+
"epoch": 9.453873889719462,
|
| 1511 |
+
"grad_norm": 31.79228973388672,
|
| 1512 |
+
"learning_rate": 5.461261102805383e-07,
|
| 1513 |
+
"loss": 0.0488,
|
| 1514 |
+
"step": 107500
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 9.497845396183273,
|
| 1518 |
+
"grad_norm": 27.02912139892578,
|
| 1519 |
+
"learning_rate": 5.021546038167269e-07,
|
| 1520 |
+
"loss": 0.0533,
|
| 1521 |
+
"step": 108000
|
| 1522 |
+
},
|
| 1523 |
+
{
|
| 1524 |
+
"epoch": 9.541816902647085,
|
| 1525 |
+
"grad_norm": 0.05077936500310898,
|
| 1526 |
+
"learning_rate": 4.581830973529153e-07,
|
| 1527 |
+
"loss": 0.0485,
|
| 1528 |
+
"step": 108500
|
| 1529 |
+
},
|
| 1530 |
+
{
|
| 1531 |
+
"epoch": 9.585788409110895,
|
| 1532 |
+
"grad_norm": 0.13254009187221527,
|
| 1533 |
+
"learning_rate": 4.142115908891039e-07,
|
| 1534 |
+
"loss": 0.0475,
|
| 1535 |
+
"step": 109000
|
| 1536 |
+
},
|
| 1537 |
+
{
|
| 1538 |
+
"epoch": 9.629759915574708,
|
| 1539 |
+
"grad_norm": 4.606433391571045,
|
| 1540 |
+
"learning_rate": 3.7024008442529244e-07,
|
| 1541 |
+
"loss": 0.0465,
|
| 1542 |
+
"step": 109500
|
| 1543 |
+
},
|
| 1544 |
+
{
|
| 1545 |
+
"epoch": 9.67373142203852,
|
| 1546 |
+
"grad_norm": 12.37204647064209,
|
| 1547 |
+
"learning_rate": 3.2626857796148096e-07,
|
| 1548 |
+
"loss": 0.0431,
|
| 1549 |
+
"step": 110000
|
| 1550 |
+
},
|
| 1551 |
+
{
|
| 1552 |
+
"epoch": 9.71770292850233,
|
| 1553 |
+
"grad_norm": 68.52936553955078,
|
| 1554 |
+
"learning_rate": 2.8229707149766955e-07,
|
| 1555 |
+
"loss": 0.0498,
|
| 1556 |
+
"step": 110500
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 9.761674434966142,
|
| 1560 |
+
"grad_norm": 11.432622909545898,
|
| 1561 |
+
"learning_rate": 2.3832556503385807e-07,
|
| 1562 |
+
"loss": 0.0512,
|
| 1563 |
+
"step": 111000
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 9.805645941429953,
|
| 1567 |
+
"grad_norm": 0.3197621703147888,
|
| 1568 |
+
"learning_rate": 1.9435405857004663e-07,
|
| 1569 |
+
"loss": 0.0494,
|
| 1570 |
+
"step": 111500
|
| 1571 |
+
},
|
| 1572 |
+
{
|
| 1573 |
+
"epoch": 9.849617447893765,
|
| 1574 |
+
"grad_norm": 0.03288858011364937,
|
| 1575 |
+
"learning_rate": 1.5038255210623516e-07,
|
| 1576 |
+
"loss": 0.0504,
|
| 1577 |
+
"step": 112000
|
| 1578 |
+
},
|
| 1579 |
+
{
|
| 1580 |
+
"epoch": 9.893588954357575,
|
| 1581 |
+
"grad_norm": 0.5140511989593506,
|
| 1582 |
+
"learning_rate": 1.0641104564242372e-07,
|
| 1583 |
+
"loss": 0.0476,
|
| 1584 |
+
"step": 112500
|
| 1585 |
+
},
|
| 1586 |
+
{
|
| 1587 |
+
"epoch": 9.937560460821388,
|
| 1588 |
+
"grad_norm": 0.2467552274465561,
|
| 1589 |
+
"learning_rate": 6.243953917861227e-08,
|
| 1590 |
+
"loss": 0.0484,
|
| 1591 |
+
"step": 113000
|
| 1592 |
+
},
|
| 1593 |
+
{
|
| 1594 |
+
"epoch": 9.9815319672852,
|
| 1595 |
+
"grad_norm": 9.55329418182373,
|
| 1596 |
+
"learning_rate": 1.846803271480081e-08,
|
| 1597 |
+
"loss": 0.0474,
|
| 1598 |
+
"step": 113500
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 10.0,
|
| 1602 |
+
"step": 113710,
|
| 1603 |
+
"total_flos": 2.4428070234989568e+17,
|
| 1604 |
+
"train_loss": 0.1514742186763295,
|
| 1605 |
+
"train_runtime": 20029.6511,
|
| 1606 |
+
"train_samples_per_second": 181.654,
|
| 1607 |
+
"train_steps_per_second": 5.677
|
| 1608 |
+
}
|
| 1609 |
+
],
|
| 1610 |
+
"logging_steps": 500,
|
| 1611 |
+
"max_steps": 113710,
|
| 1612 |
+
"num_input_tokens_seen": 0,
|
| 1613 |
+
"num_train_epochs": 10,
|
| 1614 |
+
"save_steps": 500,
|
| 1615 |
+
"stateful_callbacks": {
|
| 1616 |
+
"TrainerControl": {
|
| 1617 |
+
"args": {
|
| 1618 |
+
"should_epoch_stop": false,
|
| 1619 |
+
"should_evaluate": false,
|
| 1620 |
+
"should_log": false,
|
| 1621 |
+
"should_save": true,
|
| 1622 |
+
"should_training_stop": true
|
| 1623 |
+
},
|
| 1624 |
+
"attributes": {}
|
| 1625 |
+
}
|
| 1626 |
+
},
|
| 1627 |
+
"total_flos": 2.4428070234989568e+17,
|
| 1628 |
+
"train_batch_size": 32,
|
| 1629 |
+
"trial_name": null,
|
| 1630 |
+
"trial_params": null
|
| 1631 |
+
}
|
qqp/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a261d42e7036391510c8a43a896afac531bf5570083d0797565a66d0a5033ea2
|
| 3 |
+
size 5304
|
qqp/vocab.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|