Upload folder using huggingface_hub
Browse files- README.md +34 -0
- checkpoint-8938/config.json +70 -0
- checkpoint-8938/generation_config.json +16 -0
- checkpoint-8938/merges.txt +0 -0
- checkpoint-8938/model.safetensors +3 -0
- checkpoint-8938/optimizer.pt +3 -0
- checkpoint-8938/rng_state.pth +3 -0
- checkpoint-8938/scheduler.pt +3 -0
- checkpoint-8938/special_tokens_map.json +15 -0
- checkpoint-8938/tokenizer.json +0 -0
- checkpoint-8938/tokenizer_config.json +57 -0
- checkpoint-8938/trainer_state.json +2567 -0
- checkpoint-8938/training_args.bin +3 -0
- checkpoint-8938/vocab.json +0 -0
- config.json +70 -0
- generation_config.json +16 -0
- merges.txt +0 -0
- model.safetensors +3 -0
- runs/Jun11_09-59-45_r-riken01-t5-base-njeyrhxb-70853-v6p9n/events.out.tfevents.1718099995.r-riken01-t5-base-njeyrhxb-70853-v6p9n.89.0 +2 -2
- runs/Jun11_09-59-45_r-riken01-t5-base-njeyrhxb-70853-v6p9n/events.out.tfevents.1718107440.r-riken01-t5-base-njeyrhxb-70853-v6p9n.89.1 +3 -0
- special_tokens_map.json +15 -0
- tokenizer.json +0 -0
- tokenizer_config.json +57 -0
- training_args.bin +3 -0
- training_params.json +37 -0
- vocab.json +0 -0
README.md
ADDED
|
@@ -0,0 +1,34 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
|
| 2 |
+
---
|
| 3 |
+
tags:
|
| 4 |
+
- autotrain
|
| 5 |
+
- text2text-generation
|
| 6 |
+
base_model: facebook/bart-large-cnn
|
| 7 |
+
widget:
|
| 8 |
+
- text: "I love AutoTrain"
|
| 9 |
+
---
|
| 10 |
+
|
| 11 |
+
# Model Trained Using AutoTrain
|
| 12 |
+
|
| 13 |
+
- Problem type: Seq2Seq
|
| 14 |
+
|
| 15 |
+
## Validation Metrics
|
| 16 |
+
loss: 4.248934268951416
|
| 17 |
+
|
| 18 |
+
rouge1: 25.7685
|
| 19 |
+
|
| 20 |
+
rouge2: 9.8226
|
| 21 |
+
|
| 22 |
+
rougeL: 24.6426
|
| 23 |
+
|
| 24 |
+
rougeLsum: 24.9756
|
| 25 |
+
|
| 26 |
+
gen_len: 61.8854
|
| 27 |
+
|
| 28 |
+
runtime: 700.1847
|
| 29 |
+
|
| 30 |
+
samples_per_second: 1.595
|
| 31 |
+
|
| 32 |
+
steps_per_second: 0.4
|
| 33 |
+
|
| 34 |
+
: 3.0
|
checkpoint-8938/config.json
ADDED
|
@@ -0,0 +1,70 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "facebook/bart-large-cnn",
|
| 3 |
+
"_num_labels": 3,
|
| 4 |
+
"activation_dropout": 0.0,
|
| 5 |
+
"activation_function": "gelu",
|
| 6 |
+
"add_final_layer_norm": false,
|
| 7 |
+
"architectures": [
|
| 8 |
+
"BartForConditionalGeneration"
|
| 9 |
+
],
|
| 10 |
+
"attention_dropout": 0.0,
|
| 11 |
+
"bos_token_id": 0,
|
| 12 |
+
"classif_dropout": 0.0,
|
| 13 |
+
"classifier_dropout": 0.0,
|
| 14 |
+
"d_model": 1024,
|
| 15 |
+
"decoder_attention_heads": 16,
|
| 16 |
+
"decoder_ffn_dim": 4096,
|
| 17 |
+
"decoder_layerdrop": 0.0,
|
| 18 |
+
"decoder_layers": 12,
|
| 19 |
+
"decoder_start_token_id": 2,
|
| 20 |
+
"dropout": 0.1,
|
| 21 |
+
"early_stopping": true,
|
| 22 |
+
"encoder_attention_heads": 16,
|
| 23 |
+
"encoder_ffn_dim": 4096,
|
| 24 |
+
"encoder_layerdrop": 0.0,
|
| 25 |
+
"encoder_layers": 12,
|
| 26 |
+
"eos_token_id": 2,
|
| 27 |
+
"force_bos_token_to_be_generated": true,
|
| 28 |
+
"forced_bos_token_id": 0,
|
| 29 |
+
"forced_eos_token_id": 2,
|
| 30 |
+
"gradient_checkpointing": false,
|
| 31 |
+
"id2label": {
|
| 32 |
+
"0": "LABEL_0",
|
| 33 |
+
"1": "LABEL_1",
|
| 34 |
+
"2": "LABEL_2"
|
| 35 |
+
},
|
| 36 |
+
"init_std": 0.02,
|
| 37 |
+
"is_encoder_decoder": true,
|
| 38 |
+
"label2id": {
|
| 39 |
+
"LABEL_0": 0,
|
| 40 |
+
"LABEL_1": 1,
|
| 41 |
+
"LABEL_2": 2
|
| 42 |
+
},
|
| 43 |
+
"length_penalty": 2.0,
|
| 44 |
+
"max_length": 142,
|
| 45 |
+
"max_position_embeddings": 1024,
|
| 46 |
+
"min_length": 56,
|
| 47 |
+
"model_type": "bart",
|
| 48 |
+
"no_repeat_ngram_size": 3,
|
| 49 |
+
"normalize_before": false,
|
| 50 |
+
"num_beams": 4,
|
| 51 |
+
"num_hidden_layers": 12,
|
| 52 |
+
"output_past": true,
|
| 53 |
+
"pad_token_id": 1,
|
| 54 |
+
"prefix": " ",
|
| 55 |
+
"scale_embedding": false,
|
| 56 |
+
"task_specific_params": {
|
| 57 |
+
"summarization": {
|
| 58 |
+
"early_stopping": true,
|
| 59 |
+
"length_penalty": 2.0,
|
| 60 |
+
"max_length": 142,
|
| 61 |
+
"min_length": 56,
|
| 62 |
+
"no_repeat_ngram_size": 3,
|
| 63 |
+
"num_beams": 4
|
| 64 |
+
}
|
| 65 |
+
},
|
| 66 |
+
"torch_dtype": "float32",
|
| 67 |
+
"transformers_version": "4.41.0",
|
| 68 |
+
"use_cache": false,
|
| 69 |
+
"vocab_size": 50265
|
| 70 |
+
}
|
checkpoint-8938/generation_config.json
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token_id": 0,
|
| 3 |
+
"decoder_start_token_id": 2,
|
| 4 |
+
"early_stopping": true,
|
| 5 |
+
"eos_token_id": 2,
|
| 6 |
+
"forced_bos_token_id": 0,
|
| 7 |
+
"forced_eos_token_id": 2,
|
| 8 |
+
"length_penalty": 2.0,
|
| 9 |
+
"max_length": 142,
|
| 10 |
+
"min_length": 56,
|
| 11 |
+
"no_repeat_ngram_size": 3,
|
| 12 |
+
"num_beams": 4,
|
| 13 |
+
"pad_token_id": 1,
|
| 14 |
+
"transformers_version": "4.41.0",
|
| 15 |
+
"use_cache": false
|
| 16 |
+
}
|
checkpoint-8938/merges.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
checkpoint-8938/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ddd6fefb4de9b6731b7efcc997fdb1cc4db285111506952feaefad76b0cf7029
|
| 3 |
+
size 1625426996
|
checkpoint-8938/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:27c884bb33254e02b91f137180d353c1a0f3e50ff043bb824e4235bf9eb7ffba
|
| 3 |
+
size 3250759951
|
checkpoint-8938/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:38f1c8b673bdcc0876119e83cceb497451dd5a363c4b7203f158e58d54417987
|
| 3 |
+
size 14244
|
checkpoint-8938/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0b2ea4d809759782474d493220a94a73376cbd4e942abfbb7ae792ca7a782093
|
| 3 |
+
size 1064
|
checkpoint-8938/special_tokens_map.json
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token": "<s>",
|
| 3 |
+
"cls_token": "<s>",
|
| 4 |
+
"eos_token": "</s>",
|
| 5 |
+
"mask_token": {
|
| 6 |
+
"content": "<mask>",
|
| 7 |
+
"lstrip": true,
|
| 8 |
+
"normalized": true,
|
| 9 |
+
"rstrip": false,
|
| 10 |
+
"single_word": false
|
| 11 |
+
},
|
| 12 |
+
"pad_token": "<pad>",
|
| 13 |
+
"sep_token": "</s>",
|
| 14 |
+
"unk_token": "<unk>"
|
| 15 |
+
}
|
checkpoint-8938/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
checkpoint-8938/tokenizer_config.json
ADDED
|
@@ -0,0 +1,57 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_prefix_space": false,
|
| 3 |
+
"added_tokens_decoder": {
|
| 4 |
+
"0": {
|
| 5 |
+
"content": "<s>",
|
| 6 |
+
"lstrip": false,
|
| 7 |
+
"normalized": true,
|
| 8 |
+
"rstrip": false,
|
| 9 |
+
"single_word": false,
|
| 10 |
+
"special": true
|
| 11 |
+
},
|
| 12 |
+
"1": {
|
| 13 |
+
"content": "<pad>",
|
| 14 |
+
"lstrip": false,
|
| 15 |
+
"normalized": true,
|
| 16 |
+
"rstrip": false,
|
| 17 |
+
"single_word": false,
|
| 18 |
+
"special": true
|
| 19 |
+
},
|
| 20 |
+
"2": {
|
| 21 |
+
"content": "</s>",
|
| 22 |
+
"lstrip": false,
|
| 23 |
+
"normalized": true,
|
| 24 |
+
"rstrip": false,
|
| 25 |
+
"single_word": false,
|
| 26 |
+
"special": true
|
| 27 |
+
},
|
| 28 |
+
"3": {
|
| 29 |
+
"content": "<unk>",
|
| 30 |
+
"lstrip": false,
|
| 31 |
+
"normalized": true,
|
| 32 |
+
"rstrip": false,
|
| 33 |
+
"single_word": false,
|
| 34 |
+
"special": true
|
| 35 |
+
},
|
| 36 |
+
"50264": {
|
| 37 |
+
"content": "<mask>",
|
| 38 |
+
"lstrip": true,
|
| 39 |
+
"normalized": true,
|
| 40 |
+
"rstrip": false,
|
| 41 |
+
"single_word": false,
|
| 42 |
+
"special": true
|
| 43 |
+
}
|
| 44 |
+
},
|
| 45 |
+
"bos_token": "<s>",
|
| 46 |
+
"clean_up_tokenization_spaces": true,
|
| 47 |
+
"cls_token": "<s>",
|
| 48 |
+
"eos_token": "</s>",
|
| 49 |
+
"errors": "replace",
|
| 50 |
+
"mask_token": "<mask>",
|
| 51 |
+
"model_max_length": 1000000000000000019884624838656,
|
| 52 |
+
"pad_token": "<pad>",
|
| 53 |
+
"sep_token": "</s>",
|
| 54 |
+
"tokenizer_class": "BartTokenizer",
|
| 55 |
+
"trim_offsets": true,
|
| 56 |
+
"unk_token": "<unk>"
|
| 57 |
+
}
|
checkpoint-8938/trainer_state.json
ADDED
|
@@ -0,0 +1,2567 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 4.248934268951416,
|
| 3 |
+
"best_model_checkpoint": "autotrain-l6hey-orl0t/checkpoint-8938",
|
| 4 |
+
"epoch": 2.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 8938,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.005594092638174088,
|
| 13 |
+
"grad_norm": 34.332366943359375,
|
| 14 |
+
"learning_rate": 7.829977628635347e-07,
|
| 15 |
+
"loss": 6.4897,
|
| 16 |
+
"step": 25
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.011188185276348177,
|
| 20 |
+
"grad_norm": 23.168872833251953,
|
| 21 |
+
"learning_rate": 1.7151379567486951e-06,
|
| 22 |
+
"loss": 6.3738,
|
| 23 |
+
"step": 50
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.016782277914522265,
|
| 27 |
+
"grad_norm": 20.714399337768555,
|
| 28 |
+
"learning_rate": 2.6472781506338553e-06,
|
| 29 |
+
"loss": 5.5122,
|
| 30 |
+
"step": 75
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.022376370552696354,
|
| 34 |
+
"grad_norm": 17.31004524230957,
|
| 35 |
+
"learning_rate": 3.542132736763609e-06,
|
| 36 |
+
"loss": 5.5352,
|
| 37 |
+
"step": 100
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.027970463190870442,
|
| 41 |
+
"grad_norm": 20.792905807495117,
|
| 42 |
+
"learning_rate": 4.47427293064877e-06,
|
| 43 |
+
"loss": 5.3902,
|
| 44 |
+
"step": 125
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.03356455582904453,
|
| 48 |
+
"grad_norm": 29.867664337158203,
|
| 49 |
+
"learning_rate": 5.40641312453393e-06,
|
| 50 |
+
"loss": 5.0552,
|
| 51 |
+
"step": 150
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.039158648467218615,
|
| 55 |
+
"grad_norm": 17.143095016479492,
|
| 56 |
+
"learning_rate": 6.338553318419091e-06,
|
| 57 |
+
"loss": 5.0792,
|
| 58 |
+
"step": 175
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.04475274110539271,
|
| 62 |
+
"grad_norm": 16.778640747070312,
|
| 63 |
+
"learning_rate": 7.270693512304251e-06,
|
| 64 |
+
"loss": 5.0412,
|
| 65 |
+
"step": 200
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.05034683374356679,
|
| 69 |
+
"grad_norm": 21.074262619018555,
|
| 70 |
+
"learning_rate": 8.20283370618941e-06,
|
| 71 |
+
"loss": 4.8163,
|
| 72 |
+
"step": 225
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.055940926381740884,
|
| 76 |
+
"grad_norm": 21.221630096435547,
|
| 77 |
+
"learning_rate": 9.134973900074571e-06,
|
| 78 |
+
"loss": 4.9409,
|
| 79 |
+
"step": 250
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.06153501901991497,
|
| 83 |
+
"grad_norm": 21.189231872558594,
|
| 84 |
+
"learning_rate": 1.006711409395973e-05,
|
| 85 |
+
"loss": 5.0649,
|
| 86 |
+
"step": 275
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.06712911165808906,
|
| 90 |
+
"grad_norm": 14.992874145507812,
|
| 91 |
+
"learning_rate": 1.0999254287844893e-05,
|
| 92 |
+
"loss": 4.7542,
|
| 93 |
+
"step": 300
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.07272320429626315,
|
| 97 |
+
"grad_norm": 15.733444213867188,
|
| 98 |
+
"learning_rate": 1.1931394481730052e-05,
|
| 99 |
+
"loss": 4.5938,
|
| 100 |
+
"step": 325
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.07831729693443723,
|
| 104 |
+
"grad_norm": 23.274600982666016,
|
| 105 |
+
"learning_rate": 1.2863534675615213e-05,
|
| 106 |
+
"loss": 4.6138,
|
| 107 |
+
"step": 350
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.08391138957261132,
|
| 111 |
+
"grad_norm": 17.92934799194336,
|
| 112 |
+
"learning_rate": 1.3795674869500374e-05,
|
| 113 |
+
"loss": 4.8809,
|
| 114 |
+
"step": 375
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.08950548221078541,
|
| 118 |
+
"grad_norm": 18.286834716796875,
|
| 119 |
+
"learning_rate": 1.4727815063385533e-05,
|
| 120 |
+
"loss": 4.6919,
|
| 121 |
+
"step": 400
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.09509957484895949,
|
| 125 |
+
"grad_norm": 20.009130477905273,
|
| 126 |
+
"learning_rate": 1.5659955257270695e-05,
|
| 127 |
+
"loss": 4.7557,
|
| 128 |
+
"step": 425
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.10069366748713358,
|
| 132 |
+
"grad_norm": 20.796175003051758,
|
| 133 |
+
"learning_rate": 1.6592095451155853e-05,
|
| 134 |
+
"loss": 4.8059,
|
| 135 |
+
"step": 450
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.10628776012530768,
|
| 139 |
+
"grad_norm": 18.083595275878906,
|
| 140 |
+
"learning_rate": 1.7524235645041014e-05,
|
| 141 |
+
"loss": 4.86,
|
| 142 |
+
"step": 475
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.11188185276348177,
|
| 146 |
+
"grad_norm": 17.57659339904785,
|
| 147 |
+
"learning_rate": 1.8456375838926178e-05,
|
| 148 |
+
"loss": 4.568,
|
| 149 |
+
"step": 500
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.11747594540165585,
|
| 153 |
+
"grad_norm": 20.379024505615234,
|
| 154 |
+
"learning_rate": 1.9388516032811335e-05,
|
| 155 |
+
"loss": 4.9103,
|
| 156 |
+
"step": 525
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.12307003803982994,
|
| 160 |
+
"grad_norm": 16.093093872070312,
|
| 161 |
+
"learning_rate": 2.0320656226696496e-05,
|
| 162 |
+
"loss": 4.5315,
|
| 163 |
+
"step": 550
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.12866413067800403,
|
| 167 |
+
"grad_norm": 31.44219970703125,
|
| 168 |
+
"learning_rate": 2.1252796420581657e-05,
|
| 169 |
+
"loss": 4.6423,
|
| 170 |
+
"step": 575
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.13425822331617812,
|
| 174 |
+
"grad_norm": 16.757986068725586,
|
| 175 |
+
"learning_rate": 2.2184936614466818e-05,
|
| 176 |
+
"loss": 4.5965,
|
| 177 |
+
"step": 600
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.1398523159543522,
|
| 181 |
+
"grad_norm": 16.31531524658203,
|
| 182 |
+
"learning_rate": 2.311707680835198e-05,
|
| 183 |
+
"loss": 4.6928,
|
| 184 |
+
"step": 625
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.1454464085925263,
|
| 188 |
+
"grad_norm": 13.83728313446045,
|
| 189 |
+
"learning_rate": 2.4049217002237136e-05,
|
| 190 |
+
"loss": 4.5197,
|
| 191 |
+
"step": 650
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.15104050123070037,
|
| 195 |
+
"grad_norm": 17.00248146057129,
|
| 196 |
+
"learning_rate": 2.49813571961223e-05,
|
| 197 |
+
"loss": 4.4602,
|
| 198 |
+
"step": 675
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.15663459386887446,
|
| 202 |
+
"grad_norm": 22.1146183013916,
|
| 203 |
+
"learning_rate": 2.5913497390007457e-05,
|
| 204 |
+
"loss": 4.4573,
|
| 205 |
+
"step": 700
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.16222868650704855,
|
| 209 |
+
"grad_norm": 16.24863624572754,
|
| 210 |
+
"learning_rate": 2.6845637583892618e-05,
|
| 211 |
+
"loss": 4.6094,
|
| 212 |
+
"step": 725
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.16782277914522264,
|
| 216 |
+
"grad_norm": 15.607491493225098,
|
| 217 |
+
"learning_rate": 2.777777777777778e-05,
|
| 218 |
+
"loss": 4.4851,
|
| 219 |
+
"step": 750
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.17341687178339674,
|
| 223 |
+
"grad_norm": 17.399606704711914,
|
| 224 |
+
"learning_rate": 2.8709917971662943e-05,
|
| 225 |
+
"loss": 4.7641,
|
| 226 |
+
"step": 775
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.17901096442157083,
|
| 230 |
+
"grad_norm": 12.437596321105957,
|
| 231 |
+
"learning_rate": 2.9642058165548097e-05,
|
| 232 |
+
"loss": 4.6065,
|
| 233 |
+
"step": 800
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.18460505705974492,
|
| 237 |
+
"grad_norm": 16.83686637878418,
|
| 238 |
+
"learning_rate": 3.057419835943326e-05,
|
| 239 |
+
"loss": 4.471,
|
| 240 |
+
"step": 825
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.19019914969791898,
|
| 244 |
+
"grad_norm": 17.172122955322266,
|
| 245 |
+
"learning_rate": 3.150633855331842e-05,
|
| 246 |
+
"loss": 4.7266,
|
| 247 |
+
"step": 850
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.19579324233609308,
|
| 251 |
+
"grad_norm": 18.03239631652832,
|
| 252 |
+
"learning_rate": 3.243847874720358e-05,
|
| 253 |
+
"loss": 4.46,
|
| 254 |
+
"step": 875
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.20138733497426717,
|
| 258 |
+
"grad_norm": 11.44616985321045,
|
| 259 |
+
"learning_rate": 3.3370618941088744e-05,
|
| 260 |
+
"loss": 4.2646,
|
| 261 |
+
"step": 900
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.20698142761244126,
|
| 265 |
+
"grad_norm": 18.69893455505371,
|
| 266 |
+
"learning_rate": 3.43027591349739e-05,
|
| 267 |
+
"loss": 4.5796,
|
| 268 |
+
"step": 925
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.21257552025061535,
|
| 272 |
+
"grad_norm": 26.265470504760742,
|
| 273 |
+
"learning_rate": 3.523489932885906e-05,
|
| 274 |
+
"loss": 4.476,
|
| 275 |
+
"step": 950
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.21816961288878944,
|
| 279 |
+
"grad_norm": 28.28611946105957,
|
| 280 |
+
"learning_rate": 3.616703952274422e-05,
|
| 281 |
+
"loss": 4.9502,
|
| 282 |
+
"step": 975
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.22376370552696354,
|
| 286 |
+
"grad_norm": 20.53813362121582,
|
| 287 |
+
"learning_rate": 3.709917971662939e-05,
|
| 288 |
+
"loss": 4.5447,
|
| 289 |
+
"step": 1000
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.22935779816513763,
|
| 293 |
+
"grad_norm": 14.012741088867188,
|
| 294 |
+
"learning_rate": 3.8031319910514545e-05,
|
| 295 |
+
"loss": 4.7348,
|
| 296 |
+
"step": 1025
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.2349518908033117,
|
| 300 |
+
"grad_norm": 15.572279930114746,
|
| 301 |
+
"learning_rate": 3.89634601043997e-05,
|
| 302 |
+
"loss": 4.7246,
|
| 303 |
+
"step": 1050
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.24054598344148578,
|
| 307 |
+
"grad_norm": 13.881638526916504,
|
| 308 |
+
"learning_rate": 3.9895600298284866e-05,
|
| 309 |
+
"loss": 4.6181,
|
| 310 |
+
"step": 1075
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.24614007607965988,
|
| 314 |
+
"grad_norm": 12.749186515808105,
|
| 315 |
+
"learning_rate": 4.0827740492170024e-05,
|
| 316 |
+
"loss": 4.3893,
|
| 317 |
+
"step": 1100
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.251734168717834,
|
| 321 |
+
"grad_norm": 17.131837844848633,
|
| 322 |
+
"learning_rate": 4.175988068605519e-05,
|
| 323 |
+
"loss": 4.3859,
|
| 324 |
+
"step": 1125
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.25732826135600806,
|
| 328 |
+
"grad_norm": 13.731075286865234,
|
| 329 |
+
"learning_rate": 4.2692020879940345e-05,
|
| 330 |
+
"loss": 4.4761,
|
| 331 |
+
"step": 1150
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.2629223539941821,
|
| 335 |
+
"grad_norm": 15.064730644226074,
|
| 336 |
+
"learning_rate": 4.36241610738255e-05,
|
| 337 |
+
"loss": 4.4364,
|
| 338 |
+
"step": 1175
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.26851644663235624,
|
| 342 |
+
"grad_norm": 13.714836120605469,
|
| 343 |
+
"learning_rate": 4.455630126771067e-05,
|
| 344 |
+
"loss": 4.5969,
|
| 345 |
+
"step": 1200
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.2741105392705303,
|
| 349 |
+
"grad_norm": 14.1889009475708,
|
| 350 |
+
"learning_rate": 4.5488441461595824e-05,
|
| 351 |
+
"loss": 4.9771,
|
| 352 |
+
"step": 1225
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 0.2797046319087044,
|
| 356 |
+
"grad_norm": 14.543293952941895,
|
| 357 |
+
"learning_rate": 4.642058165548099e-05,
|
| 358 |
+
"loss": 4.6537,
|
| 359 |
+
"step": 1250
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 0.2852987245468785,
|
| 363 |
+
"grad_norm": 18.1214542388916,
|
| 364 |
+
"learning_rate": 4.735272184936615e-05,
|
| 365 |
+
"loss": 4.6208,
|
| 366 |
+
"step": 1275
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 0.2908928171850526,
|
| 370 |
+
"grad_norm": 13.347675323486328,
|
| 371 |
+
"learning_rate": 4.82848620432513e-05,
|
| 372 |
+
"loss": 4.6199,
|
| 373 |
+
"step": 1300
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 0.2964869098232267,
|
| 377 |
+
"grad_norm": 17.083444595336914,
|
| 378 |
+
"learning_rate": 4.921700223713647e-05,
|
| 379 |
+
"loss": 4.8805,
|
| 380 |
+
"step": 1325
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 0.30208100246140074,
|
| 384 |
+
"grad_norm": 24.343929290771484,
|
| 385 |
+
"learning_rate": 4.998342449859108e-05,
|
| 386 |
+
"loss": 4.6357,
|
| 387 |
+
"step": 1350
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 0.30767509509957486,
|
| 391 |
+
"grad_norm": 16.355289459228516,
|
| 392 |
+
"learning_rate": 4.987982761478535e-05,
|
| 393 |
+
"loss": 4.674,
|
| 394 |
+
"step": 1375
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 0.3132691877377489,
|
| 398 |
+
"grad_norm": 20.73784828186035,
|
| 399 |
+
"learning_rate": 4.978037460633184e-05,
|
| 400 |
+
"loss": 4.6211,
|
| 401 |
+
"step": 1400
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.31886328037592304,
|
| 405 |
+
"grad_norm": 13.50486946105957,
|
| 406 |
+
"learning_rate": 4.9676777722526106e-05,
|
| 407 |
+
"loss": 4.7257,
|
| 408 |
+
"step": 1425
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 0.3244573730140971,
|
| 412 |
+
"grad_norm": 14.019612312316895,
|
| 413 |
+
"learning_rate": 4.9573180838720376e-05,
|
| 414 |
+
"loss": 4.6509,
|
| 415 |
+
"step": 1450
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 0.3300514656522712,
|
| 419 |
+
"grad_norm": 18.581026077270508,
|
| 420 |
+
"learning_rate": 4.946958395491464e-05,
|
| 421 |
+
"loss": 4.3779,
|
| 422 |
+
"step": 1475
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 0.3356455582904453,
|
| 426 |
+
"grad_norm": 12.192296981811523,
|
| 427 |
+
"learning_rate": 4.93659870711089e-05,
|
| 428 |
+
"loss": 4.2878,
|
| 429 |
+
"step": 1500
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.34123965092861935,
|
| 433 |
+
"grad_norm": 12.066230773925781,
|
| 434 |
+
"learning_rate": 4.926239018730317e-05,
|
| 435 |
+
"loss": 4.5682,
|
| 436 |
+
"step": 1525
|
| 437 |
+
},
|
| 438 |
+
{
|
| 439 |
+
"epoch": 0.3468337435667935,
|
| 440 |
+
"grad_norm": 10.833414077758789,
|
| 441 |
+
"learning_rate": 4.9158793303497436e-05,
|
| 442 |
+
"loss": 4.1772,
|
| 443 |
+
"step": 1550
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 0.35242783620496754,
|
| 447 |
+
"grad_norm": 10.834297180175781,
|
| 448 |
+
"learning_rate": 4.905519641969169e-05,
|
| 449 |
+
"loss": 4.6879,
|
| 450 |
+
"step": 1575
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 0.35802192884314166,
|
| 454 |
+
"grad_norm": 12.870365142822266,
|
| 455 |
+
"learning_rate": 4.895159953588596e-05,
|
| 456 |
+
"loss": 4.7267,
|
| 457 |
+
"step": 1600
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 0.3636160214813157,
|
| 461 |
+
"grad_norm": 10.991371154785156,
|
| 462 |
+
"learning_rate": 4.8848002652080226e-05,
|
| 463 |
+
"loss": 4.6822,
|
| 464 |
+
"step": 1625
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 0.36921011411948984,
|
| 468 |
+
"grad_norm": 14.67126178741455,
|
| 469 |
+
"learning_rate": 4.874440576827449e-05,
|
| 470 |
+
"loss": 4.3277,
|
| 471 |
+
"step": 1650
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 0.3748042067576639,
|
| 475 |
+
"grad_norm": 13.142556190490723,
|
| 476 |
+
"learning_rate": 4.864080888446876e-05,
|
| 477 |
+
"loss": 4.513,
|
| 478 |
+
"step": 1675
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 0.38039829939583797,
|
| 482 |
+
"grad_norm": 12.647134780883789,
|
| 483 |
+
"learning_rate": 4.853721200066302e-05,
|
| 484 |
+
"loss": 4.6069,
|
| 485 |
+
"step": 1700
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 0.3859923920340121,
|
| 489 |
+
"grad_norm": 12.48798942565918,
|
| 490 |
+
"learning_rate": 4.8433615116857286e-05,
|
| 491 |
+
"loss": 4.846,
|
| 492 |
+
"step": 1725
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 0.39158648467218615,
|
| 496 |
+
"grad_norm": 10.90004825592041,
|
| 497 |
+
"learning_rate": 4.833001823305155e-05,
|
| 498 |
+
"loss": 4.4091,
|
| 499 |
+
"step": 1750
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 0.39718057731036027,
|
| 503 |
+
"grad_norm": 14.13915729522705,
|
| 504 |
+
"learning_rate": 4.822642134924581e-05,
|
| 505 |
+
"loss": 4.4192,
|
| 506 |
+
"step": 1775
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 0.40277466994853434,
|
| 510 |
+
"grad_norm": 13.102397918701172,
|
| 511 |
+
"learning_rate": 4.812282446544008e-05,
|
| 512 |
+
"loss": 4.5444,
|
| 513 |
+
"step": 1800
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 0.40836876258670846,
|
| 517 |
+
"grad_norm": 12.227831840515137,
|
| 518 |
+
"learning_rate": 4.8019227581634346e-05,
|
| 519 |
+
"loss": 4.712,
|
| 520 |
+
"step": 1825
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 0.4139628552248825,
|
| 524 |
+
"grad_norm": 15.215840339660645,
|
| 525 |
+
"learning_rate": 4.791563069782861e-05,
|
| 526 |
+
"loss": 4.6632,
|
| 527 |
+
"step": 1850
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 0.41955694786305664,
|
| 531 |
+
"grad_norm": 18.023183822631836,
|
| 532 |
+
"learning_rate": 4.781203381402288e-05,
|
| 533 |
+
"loss": 4.1778,
|
| 534 |
+
"step": 1875
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 0.4251510405012307,
|
| 538 |
+
"grad_norm": 13.230670928955078,
|
| 539 |
+
"learning_rate": 4.770843693021714e-05,
|
| 540 |
+
"loss": 4.4197,
|
| 541 |
+
"step": 1900
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 0.43074513313940477,
|
| 545 |
+
"grad_norm": 14.070335388183594,
|
| 546 |
+
"learning_rate": 4.7604840046411407e-05,
|
| 547 |
+
"loss": 4.6564,
|
| 548 |
+
"step": 1925
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 0.4363392257775789,
|
| 552 |
+
"grad_norm": 20.313472747802734,
|
| 553 |
+
"learning_rate": 4.750124316260567e-05,
|
| 554 |
+
"loss": 4.5667,
|
| 555 |
+
"step": 1950
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 0.44193331841575295,
|
| 559 |
+
"grad_norm": 15.953713417053223,
|
| 560 |
+
"learning_rate": 4.739764627879993e-05,
|
| 561 |
+
"loss": 4.5235,
|
| 562 |
+
"step": 1975
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 0.44752741105392707,
|
| 566 |
+
"grad_norm": 10.95453929901123,
|
| 567 |
+
"learning_rate": 4.7294049394994197e-05,
|
| 568 |
+
"loss": 4.5047,
|
| 569 |
+
"step": 2000
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 0.45312150369210114,
|
| 573 |
+
"grad_norm": 15.660309791564941,
|
| 574 |
+
"learning_rate": 4.719045251118847e-05,
|
| 575 |
+
"loss": 4.1537,
|
| 576 |
+
"step": 2025
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 0.45871559633027525,
|
| 580 |
+
"grad_norm": 11.03080940246582,
|
| 581 |
+
"learning_rate": 4.708685562738273e-05,
|
| 582 |
+
"loss": 4.3448,
|
| 583 |
+
"step": 2050
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 0.4643096889684493,
|
| 587 |
+
"grad_norm": 10.359949111938477,
|
| 588 |
+
"learning_rate": 4.698325874357699e-05,
|
| 589 |
+
"loss": 4.3634,
|
| 590 |
+
"step": 2075
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 0.4699037816066234,
|
| 594 |
+
"grad_norm": 25.401718139648438,
|
| 595 |
+
"learning_rate": 4.6879661859771263e-05,
|
| 596 |
+
"loss": 4.2735,
|
| 597 |
+
"step": 2100
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 0.4754978742447975,
|
| 601 |
+
"grad_norm": 14.679646492004395,
|
| 602 |
+
"learning_rate": 4.677606497596553e-05,
|
| 603 |
+
"loss": 4.4465,
|
| 604 |
+
"step": 2125
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 0.48109196688297157,
|
| 608 |
+
"grad_norm": 11.817214965820312,
|
| 609 |
+
"learning_rate": 4.667246809215979e-05,
|
| 610 |
+
"loss": 4.629,
|
| 611 |
+
"step": 2150
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 0.4866860595211457,
|
| 615 |
+
"grad_norm": 10.622258186340332,
|
| 616 |
+
"learning_rate": 4.656887120835405e-05,
|
| 617 |
+
"loss": 4.4189,
|
| 618 |
+
"step": 2175
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 0.49228015215931975,
|
| 622 |
+
"grad_norm": 15.188981056213379,
|
| 623 |
+
"learning_rate": 4.646527432454832e-05,
|
| 624 |
+
"loss": 4.2514,
|
| 625 |
+
"step": 2200
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 0.49787424479749387,
|
| 629 |
+
"grad_norm": 14.326010704040527,
|
| 630 |
+
"learning_rate": 4.636167744074259e-05,
|
| 631 |
+
"loss": 4.3532,
|
| 632 |
+
"step": 2225
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 0.503468337435668,
|
| 636 |
+
"grad_norm": 16.31020164489746,
|
| 637 |
+
"learning_rate": 4.625808055693685e-05,
|
| 638 |
+
"loss": 4.2333,
|
| 639 |
+
"step": 2250
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.509062430073842,
|
| 643 |
+
"grad_norm": 18.346088409423828,
|
| 644 |
+
"learning_rate": 4.6154483673131113e-05,
|
| 645 |
+
"loss": 4.4483,
|
| 646 |
+
"step": 2275
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"epoch": 0.5146565227120161,
|
| 650 |
+
"grad_norm": 15.169132232666016,
|
| 651 |
+
"learning_rate": 4.6050886789325384e-05,
|
| 652 |
+
"loss": 4.3814,
|
| 653 |
+
"step": 2300
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 0.5202506153501902,
|
| 657 |
+
"grad_norm": 10.011805534362793,
|
| 658 |
+
"learning_rate": 4.594728990551964e-05,
|
| 659 |
+
"loss": 4.6235,
|
| 660 |
+
"step": 2325
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 0.5258447079883642,
|
| 664 |
+
"grad_norm": 11.23599624633789,
|
| 665 |
+
"learning_rate": 4.584369302171391e-05,
|
| 666 |
+
"loss": 4.5861,
|
| 667 |
+
"step": 2350
|
| 668 |
+
},
|
| 669 |
+
{
|
| 670 |
+
"epoch": 0.5314388006265384,
|
| 671 |
+
"grad_norm": 10.484898567199707,
|
| 672 |
+
"learning_rate": 4.5740096137908174e-05,
|
| 673 |
+
"loss": 4.3894,
|
| 674 |
+
"step": 2375
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 0.5370328932647125,
|
| 678 |
+
"grad_norm": 12.192333221435547,
|
| 679 |
+
"learning_rate": 4.563649925410244e-05,
|
| 680 |
+
"loss": 4.448,
|
| 681 |
+
"step": 2400
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 0.5426269859028866,
|
| 685 |
+
"grad_norm": 15.25631332397461,
|
| 686 |
+
"learning_rate": 4.553290237029671e-05,
|
| 687 |
+
"loss": 4.3787,
|
| 688 |
+
"step": 2425
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"epoch": 0.5482210785410606,
|
| 692 |
+
"grad_norm": 10.27658748626709,
|
| 693 |
+
"learning_rate": 4.542930548649097e-05,
|
| 694 |
+
"loss": 4.4333,
|
| 695 |
+
"step": 2450
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 0.5538151711792347,
|
| 699 |
+
"grad_norm": 16.127513885498047,
|
| 700 |
+
"learning_rate": 4.5325708602685234e-05,
|
| 701 |
+
"loss": 4.1792,
|
| 702 |
+
"step": 2475
|
| 703 |
+
},
|
| 704 |
+
{
|
| 705 |
+
"epoch": 0.5594092638174089,
|
| 706 |
+
"grad_norm": 10.214879035949707,
|
| 707 |
+
"learning_rate": 4.5222111718879504e-05,
|
| 708 |
+
"loss": 4.6276,
|
| 709 |
+
"step": 2500
|
| 710 |
+
},
|
| 711 |
+
{
|
| 712 |
+
"epoch": 0.5650033564555829,
|
| 713 |
+
"grad_norm": 10.60392951965332,
|
| 714 |
+
"learning_rate": 4.511851483507376e-05,
|
| 715 |
+
"loss": 4.5925,
|
| 716 |
+
"step": 2525
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 0.570597449093757,
|
| 720 |
+
"grad_norm": 11.073285102844238,
|
| 721 |
+
"learning_rate": 4.5014917951268024e-05,
|
| 722 |
+
"loss": 4.4202,
|
| 723 |
+
"step": 2550
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 0.5761915417319311,
|
| 727 |
+
"grad_norm": 14.709641456604004,
|
| 728 |
+
"learning_rate": 4.4911321067462294e-05,
|
| 729 |
+
"loss": 4.2758,
|
| 730 |
+
"step": 2575
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 0.5817856343701052,
|
| 734 |
+
"grad_norm": 12.901021957397461,
|
| 735 |
+
"learning_rate": 4.480772418365656e-05,
|
| 736 |
+
"loss": 4.3067,
|
| 737 |
+
"step": 2600
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 0.5873797270082792,
|
| 741 |
+
"grad_norm": 11.27915096282959,
|
| 742 |
+
"learning_rate": 4.470412729985082e-05,
|
| 743 |
+
"loss": 4.0111,
|
| 744 |
+
"step": 2625
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 0.5929738196464533,
|
| 748 |
+
"grad_norm": 15.488706588745117,
|
| 749 |
+
"learning_rate": 4.460053041604509e-05,
|
| 750 |
+
"loss": 4.2671,
|
| 751 |
+
"step": 2650
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 0.5985679122846275,
|
| 755 |
+
"grad_norm": 13.603848457336426,
|
| 756 |
+
"learning_rate": 4.4496933532239354e-05,
|
| 757 |
+
"loss": 4.2589,
|
| 758 |
+
"step": 2675
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 0.6041620049228015,
|
| 762 |
+
"grad_norm": 12.827610969543457,
|
| 763 |
+
"learning_rate": 4.439333664843362e-05,
|
| 764 |
+
"loss": 4.6476,
|
| 765 |
+
"step": 2700
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 0.6097560975609756,
|
| 769 |
+
"grad_norm": 16.806106567382812,
|
| 770 |
+
"learning_rate": 4.428973976462788e-05,
|
| 771 |
+
"loss": 4.6188,
|
| 772 |
+
"step": 2725
|
| 773 |
+
},
|
| 774 |
+
{
|
| 775 |
+
"epoch": 0.6153501901991497,
|
| 776 |
+
"grad_norm": 14.598794937133789,
|
| 777 |
+
"learning_rate": 4.4186142880822144e-05,
|
| 778 |
+
"loss": 4.4195,
|
| 779 |
+
"step": 2750
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 0.6209442828373238,
|
| 783 |
+
"grad_norm": 10.380790710449219,
|
| 784 |
+
"learning_rate": 4.4082545997016414e-05,
|
| 785 |
+
"loss": 4.4634,
|
| 786 |
+
"step": 2775
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 0.6265383754754978,
|
| 790 |
+
"grad_norm": 12.62149715423584,
|
| 791 |
+
"learning_rate": 4.397894911321068e-05,
|
| 792 |
+
"loss": 4.4751,
|
| 793 |
+
"step": 2800
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 0.632132468113672,
|
| 797 |
+
"grad_norm": 10.467231750488281,
|
| 798 |
+
"learning_rate": 4.387535222940494e-05,
|
| 799 |
+
"loss": 4.2898,
|
| 800 |
+
"step": 2825
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 0.6377265607518461,
|
| 804 |
+
"grad_norm": 12.244780540466309,
|
| 805 |
+
"learning_rate": 4.377175534559921e-05,
|
| 806 |
+
"loss": 4.3242,
|
| 807 |
+
"step": 2850
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 0.6433206533900201,
|
| 811 |
+
"grad_norm": 12.39667797088623,
|
| 812 |
+
"learning_rate": 4.3668158461793474e-05,
|
| 813 |
+
"loss": 4.2754,
|
| 814 |
+
"step": 2875
|
| 815 |
+
},
|
| 816 |
+
{
|
| 817 |
+
"epoch": 0.6489147460281942,
|
| 818 |
+
"grad_norm": 12.747861862182617,
|
| 819 |
+
"learning_rate": 4.356456157798773e-05,
|
| 820 |
+
"loss": 4.286,
|
| 821 |
+
"step": 2900
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 0.6545088386663683,
|
| 825 |
+
"grad_norm": 22.809650421142578,
|
| 826 |
+
"learning_rate": 4.3460964694182e-05,
|
| 827 |
+
"loss": 4.3482,
|
| 828 |
+
"step": 2925
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"epoch": 0.6601029313045425,
|
| 832 |
+
"grad_norm": 10.659783363342285,
|
| 833 |
+
"learning_rate": 4.3357367810376264e-05,
|
| 834 |
+
"loss": 4.2066,
|
| 835 |
+
"step": 2950
|
| 836 |
+
},
|
| 837 |
+
{
|
| 838 |
+
"epoch": 0.6656970239427165,
|
| 839 |
+
"grad_norm": 14.72028636932373,
|
| 840 |
+
"learning_rate": 4.325377092657053e-05,
|
| 841 |
+
"loss": 4.4523,
|
| 842 |
+
"step": 2975
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 0.6712911165808906,
|
| 846 |
+
"grad_norm": 10.166138648986816,
|
| 847 |
+
"learning_rate": 4.31501740427648e-05,
|
| 848 |
+
"loss": 4.3615,
|
| 849 |
+
"step": 3000
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 0.6768852092190647,
|
| 853 |
+
"grad_norm": 12.992719650268555,
|
| 854 |
+
"learning_rate": 4.304657715895906e-05,
|
| 855 |
+
"loss": 4.5061,
|
| 856 |
+
"step": 3025
|
| 857 |
+
},
|
| 858 |
+
{
|
| 859 |
+
"epoch": 0.6824793018572387,
|
| 860 |
+
"grad_norm": 11.16627311706543,
|
| 861 |
+
"learning_rate": 4.2942980275153324e-05,
|
| 862 |
+
"loss": 4.5588,
|
| 863 |
+
"step": 3050
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 0.6880733944954128,
|
| 867 |
+
"grad_norm": 10.317902565002441,
|
| 868 |
+
"learning_rate": 4.2839383391347594e-05,
|
| 869 |
+
"loss": 4.3462,
|
| 870 |
+
"step": 3075
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"epoch": 0.693667487133587,
|
| 874 |
+
"grad_norm": 12.662385940551758,
|
| 875 |
+
"learning_rate": 4.273578650754185e-05,
|
| 876 |
+
"loss": 4.5361,
|
| 877 |
+
"step": 3100
|
| 878 |
+
},
|
| 879 |
+
{
|
| 880 |
+
"epoch": 0.6992615797717611,
|
| 881 |
+
"grad_norm": 16.921144485473633,
|
| 882 |
+
"learning_rate": 4.263218962373612e-05,
|
| 883 |
+
"loss": 4.2344,
|
| 884 |
+
"step": 3125
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 0.7048556724099351,
|
| 888 |
+
"grad_norm": 17.665006637573242,
|
| 889 |
+
"learning_rate": 4.2528592739930384e-05,
|
| 890 |
+
"loss": 4.2197,
|
| 891 |
+
"step": 3150
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 0.7104497650481092,
|
| 895 |
+
"grad_norm": 9.232120513916016,
|
| 896 |
+
"learning_rate": 4.242499585612465e-05,
|
| 897 |
+
"loss": 4.511,
|
| 898 |
+
"step": 3175
|
| 899 |
+
},
|
| 900 |
+
{
|
| 901 |
+
"epoch": 0.7160438576862833,
|
| 902 |
+
"grad_norm": 12.514689445495605,
|
| 903 |
+
"learning_rate": 4.232139897231892e-05,
|
| 904 |
+
"loss": 4.5251,
|
| 905 |
+
"step": 3200
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 0.7216379503244573,
|
| 909 |
+
"grad_norm": 11.398234367370605,
|
| 910 |
+
"learning_rate": 4.221780208851318e-05,
|
| 911 |
+
"loss": 4.1084,
|
| 912 |
+
"step": 3225
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 0.7272320429626314,
|
| 916 |
+
"grad_norm": 8.08095932006836,
|
| 917 |
+
"learning_rate": 4.2114205204707444e-05,
|
| 918 |
+
"loss": 4.1601,
|
| 919 |
+
"step": 3250
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 0.7328261356008056,
|
| 923 |
+
"grad_norm": 12.109641075134277,
|
| 924 |
+
"learning_rate": 4.2010608320901714e-05,
|
| 925 |
+
"loss": 4.274,
|
| 926 |
+
"step": 3275
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 0.7384202282389797,
|
| 930 |
+
"grad_norm": 10.819121360778809,
|
| 931 |
+
"learning_rate": 4.190701143709597e-05,
|
| 932 |
+
"loss": 4.5686,
|
| 933 |
+
"step": 3300
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 0.7440143208771537,
|
| 937 |
+
"grad_norm": 11.090829849243164,
|
| 938 |
+
"learning_rate": 4.180341455329024e-05,
|
| 939 |
+
"loss": 4.5401,
|
| 940 |
+
"step": 3325
|
| 941 |
+
},
|
| 942 |
+
{
|
| 943 |
+
"epoch": 0.7496084135153278,
|
| 944 |
+
"grad_norm": 11.24759578704834,
|
| 945 |
+
"learning_rate": 4.1699817669484504e-05,
|
| 946 |
+
"loss": 4.4797,
|
| 947 |
+
"step": 3350
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 0.7552025061535019,
|
| 951 |
+
"grad_norm": 10.916013717651367,
|
| 952 |
+
"learning_rate": 4.159622078567877e-05,
|
| 953 |
+
"loss": 4.1258,
|
| 954 |
+
"step": 3375
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 0.7607965987916759,
|
| 958 |
+
"grad_norm": 11.953822135925293,
|
| 959 |
+
"learning_rate": 4.149262390187304e-05,
|
| 960 |
+
"loss": 4.3313,
|
| 961 |
+
"step": 3400
|
| 962 |
+
},
|
| 963 |
+
{
|
| 964 |
+
"epoch": 0.7663906914298501,
|
| 965 |
+
"grad_norm": 16.665861129760742,
|
| 966 |
+
"learning_rate": 4.13890270180673e-05,
|
| 967 |
+
"loss": 4.2789,
|
| 968 |
+
"step": 3425
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 0.7719847840680242,
|
| 972 |
+
"grad_norm": 11.539497375488281,
|
| 973 |
+
"learning_rate": 4.1285430134261564e-05,
|
| 974 |
+
"loss": 4.5436,
|
| 975 |
+
"step": 3450
|
| 976 |
+
},
|
| 977 |
+
{
|
| 978 |
+
"epoch": 0.7775788767061983,
|
| 979 |
+
"grad_norm": 11.955995559692383,
|
| 980 |
+
"learning_rate": 4.118183325045583e-05,
|
| 981 |
+
"loss": 4.1149,
|
| 982 |
+
"step": 3475
|
| 983 |
+
},
|
| 984 |
+
{
|
| 985 |
+
"epoch": 0.7831729693443723,
|
| 986 |
+
"grad_norm": 15.087596893310547,
|
| 987 |
+
"learning_rate": 4.107823636665009e-05,
|
| 988 |
+
"loss": 4.3523,
|
| 989 |
+
"step": 3500
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 0.7887670619825464,
|
| 993 |
+
"grad_norm": 14.733497619628906,
|
| 994 |
+
"learning_rate": 4.0974639482844354e-05,
|
| 995 |
+
"loss": 4.2913,
|
| 996 |
+
"step": 3525
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 0.7943611546207205,
|
| 1000 |
+
"grad_norm": 10.125676155090332,
|
| 1001 |
+
"learning_rate": 4.0871042599038624e-05,
|
| 1002 |
+
"loss": 4.3078,
|
| 1003 |
+
"step": 3550
|
| 1004 |
+
},
|
| 1005 |
+
{
|
| 1006 |
+
"epoch": 0.7999552472588946,
|
| 1007 |
+
"grad_norm": 11.222993850708008,
|
| 1008 |
+
"learning_rate": 4.076744571523289e-05,
|
| 1009 |
+
"loss": 4.4803,
|
| 1010 |
+
"step": 3575
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 0.8055493398970687,
|
| 1014 |
+
"grad_norm": 10.871453285217285,
|
| 1015 |
+
"learning_rate": 4.066384883142715e-05,
|
| 1016 |
+
"loss": 4.5347,
|
| 1017 |
+
"step": 3600
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 0.8111434325352428,
|
| 1021 |
+
"grad_norm": 10.571527481079102,
|
| 1022 |
+
"learning_rate": 4.056025194762142e-05,
|
| 1023 |
+
"loss": 4.1658,
|
| 1024 |
+
"step": 3625
|
| 1025 |
+
},
|
| 1026 |
+
{
|
| 1027 |
+
"epoch": 0.8167375251734169,
|
| 1028 |
+
"grad_norm": 12.036689758300781,
|
| 1029 |
+
"learning_rate": 4.0456655063815685e-05,
|
| 1030 |
+
"loss": 4.3284,
|
| 1031 |
+
"step": 3650
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 0.8223316178115909,
|
| 1035 |
+
"grad_norm": 13.614919662475586,
|
| 1036 |
+
"learning_rate": 4.035305818000995e-05,
|
| 1037 |
+
"loss": 4.1228,
|
| 1038 |
+
"step": 3675
|
| 1039 |
+
},
|
| 1040 |
+
{
|
| 1041 |
+
"epoch": 0.827925710449765,
|
| 1042 |
+
"grad_norm": 12.302602767944336,
|
| 1043 |
+
"learning_rate": 4.024946129620421e-05,
|
| 1044 |
+
"loss": 4.0459,
|
| 1045 |
+
"step": 3700
|
| 1046 |
+
},
|
| 1047 |
+
{
|
| 1048 |
+
"epoch": 0.8335198030879392,
|
| 1049 |
+
"grad_norm": 13.102405548095703,
|
| 1050 |
+
"learning_rate": 4.0145864412398474e-05,
|
| 1051 |
+
"loss": 4.4641,
|
| 1052 |
+
"step": 3725
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 0.8391138957261133,
|
| 1056 |
+
"grad_norm": 12.71800422668457,
|
| 1057 |
+
"learning_rate": 4.0042267528592745e-05,
|
| 1058 |
+
"loss": 4.206,
|
| 1059 |
+
"step": 3750
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 0.8447079883642873,
|
| 1063 |
+
"grad_norm": 13.687782287597656,
|
| 1064 |
+
"learning_rate": 3.993867064478701e-05,
|
| 1065 |
+
"loss": 4.388,
|
| 1066 |
+
"step": 3775
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 0.8503020810024614,
|
| 1070 |
+
"grad_norm": 16.609664916992188,
|
| 1071 |
+
"learning_rate": 3.983507376098127e-05,
|
| 1072 |
+
"loss": 4.1664,
|
| 1073 |
+
"step": 3800
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 0.8558961736406355,
|
| 1077 |
+
"grad_norm": 16.731786727905273,
|
| 1078 |
+
"learning_rate": 3.973147687717554e-05,
|
| 1079 |
+
"loss": 4.3956,
|
| 1080 |
+
"step": 3825
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 0.8614902662788095,
|
| 1084 |
+
"grad_norm": 17.152843475341797,
|
| 1085 |
+
"learning_rate": 3.96278799933698e-05,
|
| 1086 |
+
"loss": 4.3338,
|
| 1087 |
+
"step": 3850
|
| 1088 |
+
},
|
| 1089 |
+
{
|
| 1090 |
+
"epoch": 0.8670843589169837,
|
| 1091 |
+
"grad_norm": 14.199774742126465,
|
| 1092 |
+
"learning_rate": 3.952428310956406e-05,
|
| 1093 |
+
"loss": 4.4609,
|
| 1094 |
+
"step": 3875
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 0.8726784515551578,
|
| 1098 |
+
"grad_norm": 11.605820655822754,
|
| 1099 |
+
"learning_rate": 3.942068622575833e-05,
|
| 1100 |
+
"loss": 4.1688,
|
| 1101 |
+
"step": 3900
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 0.8782725441933319,
|
| 1105 |
+
"grad_norm": 11.608319282531738,
|
| 1106 |
+
"learning_rate": 3.9317089341952595e-05,
|
| 1107 |
+
"loss": 4.2122,
|
| 1108 |
+
"step": 3925
|
| 1109 |
+
},
|
| 1110 |
+
{
|
| 1111 |
+
"epoch": 0.8838666368315059,
|
| 1112 |
+
"grad_norm": 12.212512016296387,
|
| 1113 |
+
"learning_rate": 3.921349245814686e-05,
|
| 1114 |
+
"loss": 4.2482,
|
| 1115 |
+
"step": 3950
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 0.88946072946968,
|
| 1119 |
+
"grad_norm": 12.425273895263672,
|
| 1120 |
+
"learning_rate": 3.910989557434113e-05,
|
| 1121 |
+
"loss": 4.5128,
|
| 1122 |
+
"step": 3975
|
| 1123 |
+
},
|
| 1124 |
+
{
|
| 1125 |
+
"epoch": 0.8950548221078541,
|
| 1126 |
+
"grad_norm": 14.292542457580566,
|
| 1127 |
+
"learning_rate": 3.900629869053539e-05,
|
| 1128 |
+
"loss": 4.5421,
|
| 1129 |
+
"step": 4000
|
| 1130 |
+
},
|
| 1131 |
+
{
|
| 1132 |
+
"epoch": 0.9006489147460282,
|
| 1133 |
+
"grad_norm": 9.911199569702148,
|
| 1134 |
+
"learning_rate": 3.8902701806729655e-05,
|
| 1135 |
+
"loss": 4.3237,
|
| 1136 |
+
"step": 4025
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 0.9062430073842023,
|
| 1140 |
+
"grad_norm": 6.303875923156738,
|
| 1141 |
+
"learning_rate": 3.879910492292392e-05,
|
| 1142 |
+
"loss": 4.195,
|
| 1143 |
+
"step": 4050
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 0.9118371000223764,
|
| 1147 |
+
"grad_norm": 8.433326721191406,
|
| 1148 |
+
"learning_rate": 3.869550803911818e-05,
|
| 1149 |
+
"loss": 4.1758,
|
| 1150 |
+
"step": 4075
|
| 1151 |
+
},
|
| 1152 |
+
{
|
| 1153 |
+
"epoch": 0.9174311926605505,
|
| 1154 |
+
"grad_norm": 12.792257308959961,
|
| 1155 |
+
"learning_rate": 3.859191115531245e-05,
|
| 1156 |
+
"loss": 4.1538,
|
| 1157 |
+
"step": 4100
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 0.9230252852987245,
|
| 1161 |
+
"grad_norm": 10.63804817199707,
|
| 1162 |
+
"learning_rate": 3.8488314271506715e-05,
|
| 1163 |
+
"loss": 4.2415,
|
| 1164 |
+
"step": 4125
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"epoch": 0.9286193779368986,
|
| 1168 |
+
"grad_norm": 10.244176864624023,
|
| 1169 |
+
"learning_rate": 3.838471738770098e-05,
|
| 1170 |
+
"loss": 4.624,
|
| 1171 |
+
"step": 4150
|
| 1172 |
+
},
|
| 1173 |
+
{
|
| 1174 |
+
"epoch": 0.9342134705750728,
|
| 1175 |
+
"grad_norm": 14.590502738952637,
|
| 1176 |
+
"learning_rate": 3.828112050389525e-05,
|
| 1177 |
+
"loss": 4.261,
|
| 1178 |
+
"step": 4175
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 0.9398075632132468,
|
| 1182 |
+
"grad_norm": 17.149826049804688,
|
| 1183 |
+
"learning_rate": 3.817752362008951e-05,
|
| 1184 |
+
"loss": 4.0573,
|
| 1185 |
+
"step": 4200
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 0.9454016558514209,
|
| 1189 |
+
"grad_norm": 10.837606430053711,
|
| 1190 |
+
"learning_rate": 3.8073926736283775e-05,
|
| 1191 |
+
"loss": 4.2502,
|
| 1192 |
+
"step": 4225
|
| 1193 |
+
},
|
| 1194 |
+
{
|
| 1195 |
+
"epoch": 0.950995748489595,
|
| 1196 |
+
"grad_norm": 13.960970878601074,
|
| 1197 |
+
"learning_rate": 3.797032985247804e-05,
|
| 1198 |
+
"loss": 4.1431,
|
| 1199 |
+
"step": 4250
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 0.9565898411277691,
|
| 1203 |
+
"grad_norm": 10.603372573852539,
|
| 1204 |
+
"learning_rate": 3.78667329686723e-05,
|
| 1205 |
+
"loss": 4.2805,
|
| 1206 |
+
"step": 4275
|
| 1207 |
+
},
|
| 1208 |
+
{
|
| 1209 |
+
"epoch": 0.9621839337659431,
|
| 1210 |
+
"grad_norm": 14.068360328674316,
|
| 1211 |
+
"learning_rate": 3.776313608486657e-05,
|
| 1212 |
+
"loss": 4.2771,
|
| 1213 |
+
"step": 4300
|
| 1214 |
+
},
|
| 1215 |
+
{
|
| 1216 |
+
"epoch": 0.9677780264041173,
|
| 1217 |
+
"grad_norm": 12.487285614013672,
|
| 1218 |
+
"learning_rate": 3.7659539201060835e-05,
|
| 1219 |
+
"loss": 4.4749,
|
| 1220 |
+
"step": 4325
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 0.9733721190422914,
|
| 1224 |
+
"grad_norm": 11.214025497436523,
|
| 1225 |
+
"learning_rate": 3.75559423172551e-05,
|
| 1226 |
+
"loss": 4.1715,
|
| 1227 |
+
"step": 4350
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 0.9789662116804654,
|
| 1231 |
+
"grad_norm": 12.117270469665527,
|
| 1232 |
+
"learning_rate": 3.745234543344937e-05,
|
| 1233 |
+
"loss": 4.1669,
|
| 1234 |
+
"step": 4375
|
| 1235 |
+
},
|
| 1236 |
+
{
|
| 1237 |
+
"epoch": 0.9845603043186395,
|
| 1238 |
+
"grad_norm": 7.657718181610107,
|
| 1239 |
+
"learning_rate": 3.734874854964363e-05,
|
| 1240 |
+
"loss": 4.1935,
|
| 1241 |
+
"step": 4400
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 0.9901543969568136,
|
| 1245 |
+
"grad_norm": 12.381430625915527,
|
| 1246 |
+
"learning_rate": 3.724929554119012e-05,
|
| 1247 |
+
"loss": 4.3682,
|
| 1248 |
+
"step": 4425
|
| 1249 |
+
},
|
| 1250 |
+
{
|
| 1251 |
+
"epoch": 0.9957484895949877,
|
| 1252 |
+
"grad_norm": 12.749862670898438,
|
| 1253 |
+
"learning_rate": 3.7145698657384385e-05,
|
| 1254 |
+
"loss": 4.2554,
|
| 1255 |
+
"step": 4450
|
| 1256 |
+
},
|
| 1257 |
+
{
|
| 1258 |
+
"epoch": 1.0,
|
| 1259 |
+
"eval_gen_len": 67.1128,
|
| 1260 |
+
"eval_loss": 4.290805816650391,
|
| 1261 |
+
"eval_rouge1": 26.1327,
|
| 1262 |
+
"eval_rouge2": 10.0836,
|
| 1263 |
+
"eval_rougeL": 24.9862,
|
| 1264 |
+
"eval_rougeLsum": 25.321,
|
| 1265 |
+
"eval_runtime": 1004.8822,
|
| 1266 |
+
"eval_samples_per_second": 1.112,
|
| 1267 |
+
"eval_steps_per_second": 0.279,
|
| 1268 |
+
"step": 4469
|
| 1269 |
+
},
|
| 1270 |
+
{
|
| 1271 |
+
"epoch": 1.0013425822331619,
|
| 1272 |
+
"grad_norm": 15.127477645874023,
|
| 1273 |
+
"learning_rate": 3.7042101773578655e-05,
|
| 1274 |
+
"loss": 3.9898,
|
| 1275 |
+
"step": 4475
|
| 1276 |
+
},
|
| 1277 |
+
{
|
| 1278 |
+
"epoch": 1.006936674871336,
|
| 1279 |
+
"grad_norm": 10.354872703552246,
|
| 1280 |
+
"learning_rate": 3.693850488977292e-05,
|
| 1281 |
+
"loss": 3.623,
|
| 1282 |
+
"step": 4500
|
| 1283 |
+
},
|
| 1284 |
+
{
|
| 1285 |
+
"epoch": 1.0125307675095099,
|
| 1286 |
+
"grad_norm": 13.131593704223633,
|
| 1287 |
+
"learning_rate": 3.683490800596718e-05,
|
| 1288 |
+
"loss": 3.7952,
|
| 1289 |
+
"step": 4525
|
| 1290 |
+
},
|
| 1291 |
+
{
|
| 1292 |
+
"epoch": 1.018124860147684,
|
| 1293 |
+
"grad_norm": 10.636340141296387,
|
| 1294 |
+
"learning_rate": 3.673131112216145e-05,
|
| 1295 |
+
"loss": 3.4462,
|
| 1296 |
+
"step": 4550
|
| 1297 |
+
},
|
| 1298 |
+
{
|
| 1299 |
+
"epoch": 1.0237189527858581,
|
| 1300 |
+
"grad_norm": 14.03700065612793,
|
| 1301 |
+
"learning_rate": 3.662771423835571e-05,
|
| 1302 |
+
"loss": 3.4108,
|
| 1303 |
+
"step": 4575
|
| 1304 |
+
},
|
| 1305 |
+
{
|
| 1306 |
+
"epoch": 1.0293130454240322,
|
| 1307 |
+
"grad_norm": 21.94048309326172,
|
| 1308 |
+
"learning_rate": 3.652411735454997e-05,
|
| 1309 |
+
"loss": 3.7664,
|
| 1310 |
+
"step": 4600
|
| 1311 |
+
},
|
| 1312 |
+
{
|
| 1313 |
+
"epoch": 1.0349071380622064,
|
| 1314 |
+
"grad_norm": 11.382323265075684,
|
| 1315 |
+
"learning_rate": 3.642052047074424e-05,
|
| 1316 |
+
"loss": 3.7189,
|
| 1317 |
+
"step": 4625
|
| 1318 |
+
},
|
| 1319 |
+
{
|
| 1320 |
+
"epoch": 1.0405012307003805,
|
| 1321 |
+
"grad_norm": 11.167036056518555,
|
| 1322 |
+
"learning_rate": 3.6316923586938505e-05,
|
| 1323 |
+
"loss": 3.7947,
|
| 1324 |
+
"step": 4650
|
| 1325 |
+
},
|
| 1326 |
+
{
|
| 1327 |
+
"epoch": 1.0460953233385544,
|
| 1328 |
+
"grad_norm": 13.024956703186035,
|
| 1329 |
+
"learning_rate": 3.621332670313277e-05,
|
| 1330 |
+
"loss": 3.6083,
|
| 1331 |
+
"step": 4675
|
| 1332 |
+
},
|
| 1333 |
+
{
|
| 1334 |
+
"epoch": 1.0516894159767285,
|
| 1335 |
+
"grad_norm": 11.757680892944336,
|
| 1336 |
+
"learning_rate": 3.610972981932704e-05,
|
| 1337 |
+
"loss": 3.5998,
|
| 1338 |
+
"step": 4700
|
| 1339 |
+
},
|
| 1340 |
+
{
|
| 1341 |
+
"epoch": 1.0572835086149026,
|
| 1342 |
+
"grad_norm": 14.893111228942871,
|
| 1343 |
+
"learning_rate": 3.60061329355213e-05,
|
| 1344 |
+
"loss": 3.6173,
|
| 1345 |
+
"step": 4725
|
| 1346 |
+
},
|
| 1347 |
+
{
|
| 1348 |
+
"epoch": 1.0628776012530767,
|
| 1349 |
+
"grad_norm": 9.222747802734375,
|
| 1350 |
+
"learning_rate": 3.5902536051715565e-05,
|
| 1351 |
+
"loss": 3.6453,
|
| 1352 |
+
"step": 4750
|
| 1353 |
+
},
|
| 1354 |
+
{
|
| 1355 |
+
"epoch": 1.0684716938912509,
|
| 1356 |
+
"grad_norm": 25.488880157470703,
|
| 1357 |
+
"learning_rate": 3.579893916790983e-05,
|
| 1358 |
+
"loss": 3.3486,
|
| 1359 |
+
"step": 4775
|
| 1360 |
+
},
|
| 1361 |
+
{
|
| 1362 |
+
"epoch": 1.074065786529425,
|
| 1363 |
+
"grad_norm": 10.05694580078125,
|
| 1364 |
+
"learning_rate": 3.569534228410409e-05,
|
| 1365 |
+
"loss": 3.7705,
|
| 1366 |
+
"step": 4800
|
| 1367 |
+
},
|
| 1368 |
+
{
|
| 1369 |
+
"epoch": 1.079659879167599,
|
| 1370 |
+
"grad_norm": 12.402889251708984,
|
| 1371 |
+
"learning_rate": 3.559174540029836e-05,
|
| 1372 |
+
"loss": 3.3739,
|
| 1373 |
+
"step": 4825
|
| 1374 |
+
},
|
| 1375 |
+
{
|
| 1376 |
+
"epoch": 1.085253971805773,
|
| 1377 |
+
"grad_norm": 10.890093803405762,
|
| 1378 |
+
"learning_rate": 3.5488148516492625e-05,
|
| 1379 |
+
"loss": 3.7975,
|
| 1380 |
+
"step": 4850
|
| 1381 |
+
},
|
| 1382 |
+
{
|
| 1383 |
+
"epoch": 1.090848064443947,
|
| 1384 |
+
"grad_norm": 12.410653114318848,
|
| 1385 |
+
"learning_rate": 3.538455163268689e-05,
|
| 1386 |
+
"loss": 3.5483,
|
| 1387 |
+
"step": 4875
|
| 1388 |
+
},
|
| 1389 |
+
{
|
| 1390 |
+
"epoch": 1.0964421570821212,
|
| 1391 |
+
"grad_norm": 11.636606216430664,
|
| 1392 |
+
"learning_rate": 3.528095474888116e-05,
|
| 1393 |
+
"loss": 3.5182,
|
| 1394 |
+
"step": 4900
|
| 1395 |
+
},
|
| 1396 |
+
{
|
| 1397 |
+
"epoch": 1.1020362497202953,
|
| 1398 |
+
"grad_norm": 14.367986679077148,
|
| 1399 |
+
"learning_rate": 3.517735786507542e-05,
|
| 1400 |
+
"loss": 3.9954,
|
| 1401 |
+
"step": 4925
|
| 1402 |
+
},
|
| 1403 |
+
{
|
| 1404 |
+
"epoch": 1.1076303423584695,
|
| 1405 |
+
"grad_norm": 10.753607749938965,
|
| 1406 |
+
"learning_rate": 3.5073760981269685e-05,
|
| 1407 |
+
"loss": 3.8582,
|
| 1408 |
+
"step": 4950
|
| 1409 |
+
},
|
| 1410 |
+
{
|
| 1411 |
+
"epoch": 1.1132244349966436,
|
| 1412 |
+
"grad_norm": 9.407801628112793,
|
| 1413 |
+
"learning_rate": 3.497016409746395e-05,
|
| 1414 |
+
"loss": 3.6181,
|
| 1415 |
+
"step": 4975
|
| 1416 |
+
},
|
| 1417 |
+
{
|
| 1418 |
+
"epoch": 1.1188185276348177,
|
| 1419 |
+
"grad_norm": 9.98642349243164,
|
| 1420 |
+
"learning_rate": 3.486656721365821e-05,
|
| 1421 |
+
"loss": 3.5989,
|
| 1422 |
+
"step": 5000
|
| 1423 |
+
},
|
| 1424 |
+
{
|
| 1425 |
+
"epoch": 1.1244126202729916,
|
| 1426 |
+
"grad_norm": 9.880094528198242,
|
| 1427 |
+
"learning_rate": 3.476297032985248e-05,
|
| 1428 |
+
"loss": 3.658,
|
| 1429 |
+
"step": 5025
|
| 1430 |
+
},
|
| 1431 |
+
{
|
| 1432 |
+
"epoch": 1.1300067129111657,
|
| 1433 |
+
"grad_norm": 14.001792907714844,
|
| 1434 |
+
"learning_rate": 3.4659373446046745e-05,
|
| 1435 |
+
"loss": 3.7073,
|
| 1436 |
+
"step": 5050
|
| 1437 |
+
},
|
| 1438 |
+
{
|
| 1439 |
+
"epoch": 1.1356008055493398,
|
| 1440 |
+
"grad_norm": 18.54832649230957,
|
| 1441 |
+
"learning_rate": 3.455577656224101e-05,
|
| 1442 |
+
"loss": 3.7806,
|
| 1443 |
+
"step": 5075
|
| 1444 |
+
},
|
| 1445 |
+
{
|
| 1446 |
+
"epoch": 1.141194898187514,
|
| 1447 |
+
"grad_norm": 9.804744720458984,
|
| 1448 |
+
"learning_rate": 3.445217967843528e-05,
|
| 1449 |
+
"loss": 3.6671,
|
| 1450 |
+
"step": 5100
|
| 1451 |
+
},
|
| 1452 |
+
{
|
| 1453 |
+
"epoch": 1.146788990825688,
|
| 1454 |
+
"grad_norm": 8.401939392089844,
|
| 1455 |
+
"learning_rate": 3.434858279462954e-05,
|
| 1456 |
+
"loss": 3.8566,
|
| 1457 |
+
"step": 5125
|
| 1458 |
+
},
|
| 1459 |
+
{
|
| 1460 |
+
"epoch": 1.1523830834638622,
|
| 1461 |
+
"grad_norm": 10.120752334594727,
|
| 1462 |
+
"learning_rate": 3.42449859108238e-05,
|
| 1463 |
+
"loss": 3.8143,
|
| 1464 |
+
"step": 5150
|
| 1465 |
+
},
|
| 1466 |
+
{
|
| 1467 |
+
"epoch": 1.1579771761020363,
|
| 1468 |
+
"grad_norm": 16.10240364074707,
|
| 1469 |
+
"learning_rate": 3.414138902701807e-05,
|
| 1470 |
+
"loss": 3.7451,
|
| 1471 |
+
"step": 5175
|
| 1472 |
+
},
|
| 1473 |
+
{
|
| 1474 |
+
"epoch": 1.1635712687402102,
|
| 1475 |
+
"grad_norm": 10.377949714660645,
|
| 1476 |
+
"learning_rate": 3.403779214321233e-05,
|
| 1477 |
+
"loss": 3.6082,
|
| 1478 |
+
"step": 5200
|
| 1479 |
+
},
|
| 1480 |
+
{
|
| 1481 |
+
"epoch": 1.1691653613783843,
|
| 1482 |
+
"grad_norm": 10.826866149902344,
|
| 1483 |
+
"learning_rate": 3.3934195259406595e-05,
|
| 1484 |
+
"loss": 3.8001,
|
| 1485 |
+
"step": 5225
|
| 1486 |
+
},
|
| 1487 |
+
{
|
| 1488 |
+
"epoch": 1.1747594540165585,
|
| 1489 |
+
"grad_norm": 10.02441120147705,
|
| 1490 |
+
"learning_rate": 3.3830598375600866e-05,
|
| 1491 |
+
"loss": 3.6624,
|
| 1492 |
+
"step": 5250
|
| 1493 |
+
},
|
| 1494 |
+
{
|
| 1495 |
+
"epoch": 1.1803535466547326,
|
| 1496 |
+
"grad_norm": 15.683877944946289,
|
| 1497 |
+
"learning_rate": 3.372700149179513e-05,
|
| 1498 |
+
"loss": 3.5834,
|
| 1499 |
+
"step": 5275
|
| 1500 |
+
},
|
| 1501 |
+
{
|
| 1502 |
+
"epoch": 1.1859476392929067,
|
| 1503 |
+
"grad_norm": 11.696283340454102,
|
| 1504 |
+
"learning_rate": 3.362340460798939e-05,
|
| 1505 |
+
"loss": 3.6069,
|
| 1506 |
+
"step": 5300
|
| 1507 |
+
},
|
| 1508 |
+
{
|
| 1509 |
+
"epoch": 1.1915417319310808,
|
| 1510 |
+
"grad_norm": 13.27725601196289,
|
| 1511 |
+
"learning_rate": 3.351980772418366e-05,
|
| 1512 |
+
"loss": 3.6614,
|
| 1513 |
+
"step": 5325
|
| 1514 |
+
},
|
| 1515 |
+
{
|
| 1516 |
+
"epoch": 1.197135824569255,
|
| 1517 |
+
"grad_norm": 11.811793327331543,
|
| 1518 |
+
"learning_rate": 3.341621084037792e-05,
|
| 1519 |
+
"loss": 3.6186,
|
| 1520 |
+
"step": 5350
|
| 1521 |
+
},
|
| 1522 |
+
{
|
| 1523 |
+
"epoch": 1.2027299172074288,
|
| 1524 |
+
"grad_norm": 30.400972366333008,
|
| 1525 |
+
"learning_rate": 3.331261395657219e-05,
|
| 1526 |
+
"loss": 3.6704,
|
| 1527 |
+
"step": 5375
|
| 1528 |
+
},
|
| 1529 |
+
{
|
| 1530 |
+
"epoch": 1.208324009845603,
|
| 1531 |
+
"grad_norm": 11.845870018005371,
|
| 1532 |
+
"learning_rate": 3.320901707276645e-05,
|
| 1533 |
+
"loss": 3.6688,
|
| 1534 |
+
"step": 5400
|
| 1535 |
+
},
|
| 1536 |
+
{
|
| 1537 |
+
"epoch": 1.213918102483777,
|
| 1538 |
+
"grad_norm": 14.447372436523438,
|
| 1539 |
+
"learning_rate": 3.3105420188960716e-05,
|
| 1540 |
+
"loss": 3.8172,
|
| 1541 |
+
"step": 5425
|
| 1542 |
+
},
|
| 1543 |
+
{
|
| 1544 |
+
"epoch": 1.2195121951219512,
|
| 1545 |
+
"grad_norm": 9.492889404296875,
|
| 1546 |
+
"learning_rate": 3.3001823305154986e-05,
|
| 1547 |
+
"loss": 3.5741,
|
| 1548 |
+
"step": 5450
|
| 1549 |
+
},
|
| 1550 |
+
{
|
| 1551 |
+
"epoch": 1.2251062877601253,
|
| 1552 |
+
"grad_norm": 12.105642318725586,
|
| 1553 |
+
"learning_rate": 3.289822642134925e-05,
|
| 1554 |
+
"loss": 3.5402,
|
| 1555 |
+
"step": 5475
|
| 1556 |
+
},
|
| 1557 |
+
{
|
| 1558 |
+
"epoch": 1.2307003803982994,
|
| 1559 |
+
"grad_norm": 7.518635272979736,
|
| 1560 |
+
"learning_rate": 3.279462953754351e-05,
|
| 1561 |
+
"loss": 3.7333,
|
| 1562 |
+
"step": 5500
|
| 1563 |
+
},
|
| 1564 |
+
{
|
| 1565 |
+
"epoch": 1.2362944730364736,
|
| 1566 |
+
"grad_norm": 11.485749244689941,
|
| 1567 |
+
"learning_rate": 3.269103265373778e-05,
|
| 1568 |
+
"loss": 3.8156,
|
| 1569 |
+
"step": 5525
|
| 1570 |
+
},
|
| 1571 |
+
{
|
| 1572 |
+
"epoch": 1.2418885656746474,
|
| 1573 |
+
"grad_norm": 11.726677894592285,
|
| 1574 |
+
"learning_rate": 3.258743576993204e-05,
|
| 1575 |
+
"loss": 3.937,
|
| 1576 |
+
"step": 5550
|
| 1577 |
+
},
|
| 1578 |
+
{
|
| 1579 |
+
"epoch": 1.2474826583128216,
|
| 1580 |
+
"grad_norm": 13.454861640930176,
|
| 1581 |
+
"learning_rate": 3.24838388861263e-05,
|
| 1582 |
+
"loss": 3.5792,
|
| 1583 |
+
"step": 5575
|
| 1584 |
+
},
|
| 1585 |
+
{
|
| 1586 |
+
"epoch": 1.2530767509509957,
|
| 1587 |
+
"grad_norm": 17.696428298950195,
|
| 1588 |
+
"learning_rate": 3.238024200232057e-05,
|
| 1589 |
+
"loss": 3.5828,
|
| 1590 |
+
"step": 5600
|
| 1591 |
+
},
|
| 1592 |
+
{
|
| 1593 |
+
"epoch": 1.2586708435891698,
|
| 1594 |
+
"grad_norm": 12.128670692443848,
|
| 1595 |
+
"learning_rate": 3.2276645118514836e-05,
|
| 1596 |
+
"loss": 3.6379,
|
| 1597 |
+
"step": 5625
|
| 1598 |
+
},
|
| 1599 |
+
{
|
| 1600 |
+
"epoch": 1.264264936227344,
|
| 1601 |
+
"grad_norm": 11.507698059082031,
|
| 1602 |
+
"learning_rate": 3.21730482347091e-05,
|
| 1603 |
+
"loss": 3.5192,
|
| 1604 |
+
"step": 5650
|
| 1605 |
+
},
|
| 1606 |
+
{
|
| 1607 |
+
"epoch": 1.269859028865518,
|
| 1608 |
+
"grad_norm": 11.207321166992188,
|
| 1609 |
+
"learning_rate": 3.206945135090337e-05,
|
| 1610 |
+
"loss": 3.885,
|
| 1611 |
+
"step": 5675
|
| 1612 |
+
},
|
| 1613 |
+
{
|
| 1614 |
+
"epoch": 1.2754531215036922,
|
| 1615 |
+
"grad_norm": 9.954567909240723,
|
| 1616 |
+
"learning_rate": 3.196585446709763e-05,
|
| 1617 |
+
"loss": 3.6138,
|
| 1618 |
+
"step": 5700
|
| 1619 |
+
},
|
| 1620 |
+
{
|
| 1621 |
+
"epoch": 1.281047214141866,
|
| 1622 |
+
"grad_norm": 9.7274751663208,
|
| 1623 |
+
"learning_rate": 3.1862257583291896e-05,
|
| 1624 |
+
"loss": 3.7395,
|
| 1625 |
+
"step": 5725
|
| 1626 |
+
},
|
| 1627 |
+
{
|
| 1628 |
+
"epoch": 1.2866413067800404,
|
| 1629 |
+
"grad_norm": 10.094833374023438,
|
| 1630 |
+
"learning_rate": 3.175866069948616e-05,
|
| 1631 |
+
"loss": 3.6801,
|
| 1632 |
+
"step": 5750
|
| 1633 |
+
},
|
| 1634 |
+
{
|
| 1635 |
+
"epoch": 1.2922353994182143,
|
| 1636 |
+
"grad_norm": 12.403266906738281,
|
| 1637 |
+
"learning_rate": 3.165506381568042e-05,
|
| 1638 |
+
"loss": 3.6891,
|
| 1639 |
+
"step": 5775
|
| 1640 |
+
},
|
| 1641 |
+
{
|
| 1642 |
+
"epoch": 1.2978294920563884,
|
| 1643 |
+
"grad_norm": 13.569632530212402,
|
| 1644 |
+
"learning_rate": 3.155146693187469e-05,
|
| 1645 |
+
"loss": 3.392,
|
| 1646 |
+
"step": 5800
|
| 1647 |
+
},
|
| 1648 |
+
{
|
| 1649 |
+
"epoch": 1.3034235846945625,
|
| 1650 |
+
"grad_norm": 10.21789836883545,
|
| 1651 |
+
"learning_rate": 3.1447870048068956e-05,
|
| 1652 |
+
"loss": 3.4392,
|
| 1653 |
+
"step": 5825
|
| 1654 |
+
},
|
| 1655 |
+
{
|
| 1656 |
+
"epoch": 1.3090176773327367,
|
| 1657 |
+
"grad_norm": 9.875311851501465,
|
| 1658 |
+
"learning_rate": 3.134427316426322e-05,
|
| 1659 |
+
"loss": 3.6576,
|
| 1660 |
+
"step": 5850
|
| 1661 |
+
},
|
| 1662 |
+
{
|
| 1663 |
+
"epoch": 1.3146117699709108,
|
| 1664 |
+
"grad_norm": 13.931588172912598,
|
| 1665 |
+
"learning_rate": 3.124067628045749e-05,
|
| 1666 |
+
"loss": 3.869,
|
| 1667 |
+
"step": 5875
|
| 1668 |
+
},
|
| 1669 |
+
{
|
| 1670 |
+
"epoch": 1.3202058626090847,
|
| 1671 |
+
"grad_norm": 9.532690048217773,
|
| 1672 |
+
"learning_rate": 3.113707939665175e-05,
|
| 1673 |
+
"loss": 3.7711,
|
| 1674 |
+
"step": 5900
|
| 1675 |
+
},
|
| 1676 |
+
{
|
| 1677 |
+
"epoch": 1.325799955247259,
|
| 1678 |
+
"grad_norm": 9.777695655822754,
|
| 1679 |
+
"learning_rate": 3.1033482512846016e-05,
|
| 1680 |
+
"loss": 3.4998,
|
| 1681 |
+
"step": 5925
|
| 1682 |
+
},
|
| 1683 |
+
{
|
| 1684 |
+
"epoch": 1.331394047885433,
|
| 1685 |
+
"grad_norm": 10.633079528808594,
|
| 1686 |
+
"learning_rate": 3.092988562904028e-05,
|
| 1687 |
+
"loss": 3.6,
|
| 1688 |
+
"step": 5950
|
| 1689 |
+
},
|
| 1690 |
+
{
|
| 1691 |
+
"epoch": 1.336988140523607,
|
| 1692 |
+
"grad_norm": 11.918797492980957,
|
| 1693 |
+
"learning_rate": 3.082628874523454e-05,
|
| 1694 |
+
"loss": 3.6084,
|
| 1695 |
+
"step": 5975
|
| 1696 |
+
},
|
| 1697 |
+
{
|
| 1698 |
+
"epoch": 1.3425822331617812,
|
| 1699 |
+
"grad_norm": 11.27762222290039,
|
| 1700 |
+
"learning_rate": 3.072269186142881e-05,
|
| 1701 |
+
"loss": 3.6122,
|
| 1702 |
+
"step": 6000
|
| 1703 |
+
},
|
| 1704 |
+
{
|
| 1705 |
+
"epoch": 1.3481763257999553,
|
| 1706 |
+
"grad_norm": 18.506898880004883,
|
| 1707 |
+
"learning_rate": 3.0619094977623076e-05,
|
| 1708 |
+
"loss": 3.2751,
|
| 1709 |
+
"step": 6025
|
| 1710 |
+
},
|
| 1711 |
+
{
|
| 1712 |
+
"epoch": 1.3537704184381294,
|
| 1713 |
+
"grad_norm": 21.525028228759766,
|
| 1714 |
+
"learning_rate": 3.051549809381734e-05,
|
| 1715 |
+
"loss": 3.4856,
|
| 1716 |
+
"step": 6050
|
| 1717 |
+
},
|
| 1718 |
+
{
|
| 1719 |
+
"epoch": 1.3593645110763033,
|
| 1720 |
+
"grad_norm": 13.029548645019531,
|
| 1721 |
+
"learning_rate": 3.0411901210011606e-05,
|
| 1722 |
+
"loss": 3.5935,
|
| 1723 |
+
"step": 6075
|
| 1724 |
+
},
|
| 1725 |
+
{
|
| 1726 |
+
"epoch": 1.3649586037144776,
|
| 1727 |
+
"grad_norm": 9.536824226379395,
|
| 1728 |
+
"learning_rate": 3.0308304326205873e-05,
|
| 1729 |
+
"loss": 3.569,
|
| 1730 |
+
"step": 6100
|
| 1731 |
+
},
|
| 1732 |
+
{
|
| 1733 |
+
"epoch": 1.3705526963526515,
|
| 1734 |
+
"grad_norm": 11.298256874084473,
|
| 1735 |
+
"learning_rate": 3.0204707442400133e-05,
|
| 1736 |
+
"loss": 3.8319,
|
| 1737 |
+
"step": 6125
|
| 1738 |
+
},
|
| 1739 |
+
{
|
| 1740 |
+
"epoch": 1.3761467889908257,
|
| 1741 |
+
"grad_norm": 8.461627006530762,
|
| 1742 |
+
"learning_rate": 3.0101110558594396e-05,
|
| 1743 |
+
"loss": 3.7151,
|
| 1744 |
+
"step": 6150
|
| 1745 |
+
},
|
| 1746 |
+
{
|
| 1747 |
+
"epoch": 1.3817408816289998,
|
| 1748 |
+
"grad_norm": 8.925524711608887,
|
| 1749 |
+
"learning_rate": 2.9997513674788663e-05,
|
| 1750 |
+
"loss": 3.5296,
|
| 1751 |
+
"step": 6175
|
| 1752 |
+
},
|
| 1753 |
+
{
|
| 1754 |
+
"epoch": 1.387334974267174,
|
| 1755 |
+
"grad_norm": 12.109110832214355,
|
| 1756 |
+
"learning_rate": 2.989391679098293e-05,
|
| 1757 |
+
"loss": 3.4691,
|
| 1758 |
+
"step": 6200
|
| 1759 |
+
},
|
| 1760 |
+
{
|
| 1761 |
+
"epoch": 1.392929066905348,
|
| 1762 |
+
"grad_norm": 9.915245056152344,
|
| 1763 |
+
"learning_rate": 2.9790319907177193e-05,
|
| 1764 |
+
"loss": 3.5228,
|
| 1765 |
+
"step": 6225
|
| 1766 |
+
},
|
| 1767 |
+
{
|
| 1768 |
+
"epoch": 1.398523159543522,
|
| 1769 |
+
"grad_norm": 12.354578018188477,
|
| 1770 |
+
"learning_rate": 2.968672302337146e-05,
|
| 1771 |
+
"loss": 3.5543,
|
| 1772 |
+
"step": 6250
|
| 1773 |
+
},
|
| 1774 |
+
{
|
| 1775 |
+
"epoch": 1.4041172521816963,
|
| 1776 |
+
"grad_norm": 13.05074405670166,
|
| 1777 |
+
"learning_rate": 2.9583126139565726e-05,
|
| 1778 |
+
"loss": 3.5762,
|
| 1779 |
+
"step": 6275
|
| 1780 |
+
},
|
| 1781 |
+
{
|
| 1782 |
+
"epoch": 1.4097113448198701,
|
| 1783 |
+
"grad_norm": 9.809946060180664,
|
| 1784 |
+
"learning_rate": 2.9479529255759986e-05,
|
| 1785 |
+
"loss": 3.4119,
|
| 1786 |
+
"step": 6300
|
| 1787 |
+
},
|
| 1788 |
+
{
|
| 1789 |
+
"epoch": 1.4153054374580443,
|
| 1790 |
+
"grad_norm": 10.61130142211914,
|
| 1791 |
+
"learning_rate": 2.937593237195425e-05,
|
| 1792 |
+
"loss": 3.5493,
|
| 1793 |
+
"step": 6325
|
| 1794 |
+
},
|
| 1795 |
+
{
|
| 1796 |
+
"epoch": 1.4208995300962184,
|
| 1797 |
+
"grad_norm": 8.541769027709961,
|
| 1798 |
+
"learning_rate": 2.9272335488148516e-05,
|
| 1799 |
+
"loss": 3.607,
|
| 1800 |
+
"step": 6350
|
| 1801 |
+
},
|
| 1802 |
+
{
|
| 1803 |
+
"epoch": 1.4264936227343925,
|
| 1804 |
+
"grad_norm": 14.672987937927246,
|
| 1805 |
+
"learning_rate": 2.9168738604342783e-05,
|
| 1806 |
+
"loss": 3.3681,
|
| 1807 |
+
"step": 6375
|
| 1808 |
+
},
|
| 1809 |
+
{
|
| 1810 |
+
"epoch": 1.4320877153725666,
|
| 1811 |
+
"grad_norm": 16.417062759399414,
|
| 1812 |
+
"learning_rate": 2.9065141720537046e-05,
|
| 1813 |
+
"loss": 3.581,
|
| 1814 |
+
"step": 6400
|
| 1815 |
+
},
|
| 1816 |
+
{
|
| 1817 |
+
"epoch": 1.4376818080107405,
|
| 1818 |
+
"grad_norm": 11.234773635864258,
|
| 1819 |
+
"learning_rate": 2.8961544836731313e-05,
|
| 1820 |
+
"loss": 3.8083,
|
| 1821 |
+
"step": 6425
|
| 1822 |
+
},
|
| 1823 |
+
{
|
| 1824 |
+
"epoch": 1.4432759006489149,
|
| 1825 |
+
"grad_norm": 8.443843841552734,
|
| 1826 |
+
"learning_rate": 2.885794795292558e-05,
|
| 1827 |
+
"loss": 3.4413,
|
| 1828 |
+
"step": 6450
|
| 1829 |
+
},
|
| 1830 |
+
{
|
| 1831 |
+
"epoch": 1.4488699932870888,
|
| 1832 |
+
"grad_norm": 11.358126640319824,
|
| 1833 |
+
"learning_rate": 2.8754351069119843e-05,
|
| 1834 |
+
"loss": 3.5337,
|
| 1835 |
+
"step": 6475
|
| 1836 |
+
},
|
| 1837 |
+
{
|
| 1838 |
+
"epoch": 1.4544640859252629,
|
| 1839 |
+
"grad_norm": 15.18321418762207,
|
| 1840 |
+
"learning_rate": 2.8650754185314106e-05,
|
| 1841 |
+
"loss": 3.7784,
|
| 1842 |
+
"step": 6500
|
| 1843 |
+
},
|
| 1844 |
+
{
|
| 1845 |
+
"epoch": 1.460058178563437,
|
| 1846 |
+
"grad_norm": 10.894759178161621,
|
| 1847 |
+
"learning_rate": 2.854715730150837e-05,
|
| 1848 |
+
"loss": 3.736,
|
| 1849 |
+
"step": 6525
|
| 1850 |
+
},
|
| 1851 |
+
{
|
| 1852 |
+
"epoch": 1.4656522712016111,
|
| 1853 |
+
"grad_norm": 8.780854225158691,
|
| 1854 |
+
"learning_rate": 2.8443560417702636e-05,
|
| 1855 |
+
"loss": 3.5959,
|
| 1856 |
+
"step": 6550
|
| 1857 |
+
},
|
| 1858 |
+
{
|
| 1859 |
+
"epoch": 1.4712463638397852,
|
| 1860 |
+
"grad_norm": 16.36128044128418,
|
| 1861 |
+
"learning_rate": 2.8339963533896903e-05,
|
| 1862 |
+
"loss": 3.6979,
|
| 1863 |
+
"step": 6575
|
| 1864 |
+
},
|
| 1865 |
+
{
|
| 1866 |
+
"epoch": 1.4768404564779591,
|
| 1867 |
+
"grad_norm": 10.243796348571777,
|
| 1868 |
+
"learning_rate": 2.8236366650091166e-05,
|
| 1869 |
+
"loss": 3.6813,
|
| 1870 |
+
"step": 6600
|
| 1871 |
+
},
|
| 1872 |
+
{
|
| 1873 |
+
"epoch": 1.4824345491161335,
|
| 1874 |
+
"grad_norm": 13.104743003845215,
|
| 1875 |
+
"learning_rate": 2.8132769766285433e-05,
|
| 1876 |
+
"loss": 3.6853,
|
| 1877 |
+
"step": 6625
|
| 1878 |
+
},
|
| 1879 |
+
{
|
| 1880 |
+
"epoch": 1.4880286417543074,
|
| 1881 |
+
"grad_norm": 11.391397476196289,
|
| 1882 |
+
"learning_rate": 2.80291728824797e-05,
|
| 1883 |
+
"loss": 3.5911,
|
| 1884 |
+
"step": 6650
|
| 1885 |
+
},
|
| 1886 |
+
{
|
| 1887 |
+
"epoch": 1.4936227343924815,
|
| 1888 |
+
"grad_norm": 11.6659574508667,
|
| 1889 |
+
"learning_rate": 2.792557599867396e-05,
|
| 1890 |
+
"loss": 3.5579,
|
| 1891 |
+
"step": 6675
|
| 1892 |
+
},
|
| 1893 |
+
{
|
| 1894 |
+
"epoch": 1.4992168270306556,
|
| 1895 |
+
"grad_norm": 11.695647239685059,
|
| 1896 |
+
"learning_rate": 2.7821979114868223e-05,
|
| 1897 |
+
"loss": 3.5243,
|
| 1898 |
+
"step": 6700
|
| 1899 |
+
},
|
| 1900 |
+
{
|
| 1901 |
+
"epoch": 1.5048109196688297,
|
| 1902 |
+
"grad_norm": 9.701094627380371,
|
| 1903 |
+
"learning_rate": 2.771838223106249e-05,
|
| 1904 |
+
"loss": 3.6699,
|
| 1905 |
+
"step": 6725
|
| 1906 |
+
},
|
| 1907 |
+
{
|
| 1908 |
+
"epoch": 1.5104050123070039,
|
| 1909 |
+
"grad_norm": 15.949247360229492,
|
| 1910 |
+
"learning_rate": 2.7614785347256757e-05,
|
| 1911 |
+
"loss": 3.7613,
|
| 1912 |
+
"step": 6750
|
| 1913 |
+
},
|
| 1914 |
+
{
|
| 1915 |
+
"epoch": 1.5159991049451778,
|
| 1916 |
+
"grad_norm": 13.379142761230469,
|
| 1917 |
+
"learning_rate": 2.751118846345102e-05,
|
| 1918 |
+
"loss": 3.7097,
|
| 1919 |
+
"step": 6775
|
| 1920 |
+
},
|
| 1921 |
+
{
|
| 1922 |
+
"epoch": 1.521593197583352,
|
| 1923 |
+
"grad_norm": 10.693124771118164,
|
| 1924 |
+
"learning_rate": 2.7407591579645287e-05,
|
| 1925 |
+
"loss": 3.7779,
|
| 1926 |
+
"step": 6800
|
| 1927 |
+
},
|
| 1928 |
+
{
|
| 1929 |
+
"epoch": 1.527187290221526,
|
| 1930 |
+
"grad_norm": 7.9651312828063965,
|
| 1931 |
+
"learning_rate": 2.7303994695839553e-05,
|
| 1932 |
+
"loss": 3.6414,
|
| 1933 |
+
"step": 6825
|
| 1934 |
+
},
|
| 1935 |
+
{
|
| 1936 |
+
"epoch": 1.5327813828597001,
|
| 1937 |
+
"grad_norm": 11.157812118530273,
|
| 1938 |
+
"learning_rate": 2.7200397812033817e-05,
|
| 1939 |
+
"loss": 3.6576,
|
| 1940 |
+
"step": 6850
|
| 1941 |
+
},
|
| 1942 |
+
{
|
| 1943 |
+
"epoch": 1.5383754754978742,
|
| 1944 |
+
"grad_norm": 12.323993682861328,
|
| 1945 |
+
"learning_rate": 2.7096800928228077e-05,
|
| 1946 |
+
"loss": 3.4723,
|
| 1947 |
+
"step": 6875
|
| 1948 |
+
},
|
| 1949 |
+
{
|
| 1950 |
+
"epoch": 1.5439695681360484,
|
| 1951 |
+
"grad_norm": 10.302526473999023,
|
| 1952 |
+
"learning_rate": 2.6993204044422343e-05,
|
| 1953 |
+
"loss": 3.7213,
|
| 1954 |
+
"step": 6900
|
| 1955 |
+
},
|
| 1956 |
+
{
|
| 1957 |
+
"epoch": 1.5495636607742225,
|
| 1958 |
+
"grad_norm": 10.622782707214355,
|
| 1959 |
+
"learning_rate": 2.688960716061661e-05,
|
| 1960 |
+
"loss": 3.582,
|
| 1961 |
+
"step": 6925
|
| 1962 |
+
},
|
| 1963 |
+
{
|
| 1964 |
+
"epoch": 1.5551577534123964,
|
| 1965 |
+
"grad_norm": 12.592206001281738,
|
| 1966 |
+
"learning_rate": 2.6786010276810873e-05,
|
| 1967 |
+
"loss": 3.524,
|
| 1968 |
+
"step": 6950
|
| 1969 |
+
},
|
| 1970 |
+
{
|
| 1971 |
+
"epoch": 1.5607518460505707,
|
| 1972 |
+
"grad_norm": 10.893891334533691,
|
| 1973 |
+
"learning_rate": 2.668241339300514e-05,
|
| 1974 |
+
"loss": 3.4272,
|
| 1975 |
+
"step": 6975
|
| 1976 |
+
},
|
| 1977 |
+
{
|
| 1978 |
+
"epoch": 1.5663459386887446,
|
| 1979 |
+
"grad_norm": 11.728677749633789,
|
| 1980 |
+
"learning_rate": 2.6578816509199407e-05,
|
| 1981 |
+
"loss": 3.5618,
|
| 1982 |
+
"step": 7000
|
| 1983 |
+
},
|
| 1984 |
+
{
|
| 1985 |
+
"epoch": 1.5719400313269187,
|
| 1986 |
+
"grad_norm": 14.411259651184082,
|
| 1987 |
+
"learning_rate": 2.647521962539367e-05,
|
| 1988 |
+
"loss": 3.6041,
|
| 1989 |
+
"step": 7025
|
| 1990 |
+
},
|
| 1991 |
+
{
|
| 1992 |
+
"epoch": 1.5775341239650928,
|
| 1993 |
+
"grad_norm": 12.213258743286133,
|
| 1994 |
+
"learning_rate": 2.6371622741587937e-05,
|
| 1995 |
+
"loss": 3.5627,
|
| 1996 |
+
"step": 7050
|
| 1997 |
+
},
|
| 1998 |
+
{
|
| 1999 |
+
"epoch": 1.583128216603267,
|
| 2000 |
+
"grad_norm": 12.341785430908203,
|
| 2001 |
+
"learning_rate": 2.6268025857782197e-05,
|
| 2002 |
+
"loss": 3.631,
|
| 2003 |
+
"step": 7075
|
| 2004 |
+
},
|
| 2005 |
+
{
|
| 2006 |
+
"epoch": 1.588722309241441,
|
| 2007 |
+
"grad_norm": 22.097862243652344,
|
| 2008 |
+
"learning_rate": 2.6164428973976464e-05,
|
| 2009 |
+
"loss": 3.5977,
|
| 2010 |
+
"step": 7100
|
| 2011 |
+
},
|
| 2012 |
+
{
|
| 2013 |
+
"epoch": 1.594316401879615,
|
| 2014 |
+
"grad_norm": 11.355073928833008,
|
| 2015 |
+
"learning_rate": 2.6060832090170727e-05,
|
| 2016 |
+
"loss": 3.6045,
|
| 2017 |
+
"step": 7125
|
| 2018 |
+
},
|
| 2019 |
+
{
|
| 2020 |
+
"epoch": 1.5999104945177893,
|
| 2021 |
+
"grad_norm": 12.087318420410156,
|
| 2022 |
+
"learning_rate": 2.5957235206364994e-05,
|
| 2023 |
+
"loss": 3.5971,
|
| 2024 |
+
"step": 7150
|
| 2025 |
+
},
|
| 2026 |
+
{
|
| 2027 |
+
"epoch": 1.6055045871559632,
|
| 2028 |
+
"grad_norm": 12.888108253479004,
|
| 2029 |
+
"learning_rate": 2.585363832255926e-05,
|
| 2030 |
+
"loss": 3.4293,
|
| 2031 |
+
"step": 7175
|
| 2032 |
+
},
|
| 2033 |
+
{
|
| 2034 |
+
"epoch": 1.6110986797941373,
|
| 2035 |
+
"grad_norm": 11.495614051818848,
|
| 2036 |
+
"learning_rate": 2.5750041438753524e-05,
|
| 2037 |
+
"loss": 3.9471,
|
| 2038 |
+
"step": 7200
|
| 2039 |
+
},
|
| 2040 |
+
{
|
| 2041 |
+
"epoch": 1.6166927724323115,
|
| 2042 |
+
"grad_norm": 12.585895538330078,
|
| 2043 |
+
"learning_rate": 2.564644455494779e-05,
|
| 2044 |
+
"loss": 3.5671,
|
| 2045 |
+
"step": 7225
|
| 2046 |
+
},
|
| 2047 |
+
{
|
| 2048 |
+
"epoch": 1.6222868650704856,
|
| 2049 |
+
"grad_norm": 8.79129409790039,
|
| 2050 |
+
"learning_rate": 2.554284767114205e-05,
|
| 2051 |
+
"loss": 3.6453,
|
| 2052 |
+
"step": 7250
|
| 2053 |
+
},
|
| 2054 |
+
{
|
| 2055 |
+
"epoch": 1.6278809577086597,
|
| 2056 |
+
"grad_norm": 10.552870750427246,
|
| 2057 |
+
"learning_rate": 2.5439250787336317e-05,
|
| 2058 |
+
"loss": 3.5649,
|
| 2059 |
+
"step": 7275
|
| 2060 |
+
},
|
| 2061 |
+
{
|
| 2062 |
+
"epoch": 1.6334750503468336,
|
| 2063 |
+
"grad_norm": 15.649266242980957,
|
| 2064 |
+
"learning_rate": 2.533565390353058e-05,
|
| 2065 |
+
"loss": 3.5291,
|
| 2066 |
+
"step": 7300
|
| 2067 |
+
},
|
| 2068 |
+
{
|
| 2069 |
+
"epoch": 1.639069142985008,
|
| 2070 |
+
"grad_norm": 13.944864273071289,
|
| 2071 |
+
"learning_rate": 2.5232057019724847e-05,
|
| 2072 |
+
"loss": 3.3558,
|
| 2073 |
+
"step": 7325
|
| 2074 |
+
},
|
| 2075 |
+
{
|
| 2076 |
+
"epoch": 1.6446632356231818,
|
| 2077 |
+
"grad_norm": 14.092317581176758,
|
| 2078 |
+
"learning_rate": 2.5128460135919114e-05,
|
| 2079 |
+
"loss": 3.6211,
|
| 2080 |
+
"step": 7350
|
| 2081 |
+
},
|
| 2082 |
+
{
|
| 2083 |
+
"epoch": 1.650257328261356,
|
| 2084 |
+
"grad_norm": 12.18454647064209,
|
| 2085 |
+
"learning_rate": 2.5024863252113377e-05,
|
| 2086 |
+
"loss": 3.7732,
|
| 2087 |
+
"step": 7375
|
| 2088 |
+
},
|
| 2089 |
+
{
|
| 2090 |
+
"epoch": 1.65585142089953,
|
| 2091 |
+
"grad_norm": 10.779006958007812,
|
| 2092 |
+
"learning_rate": 2.492126636830764e-05,
|
| 2093 |
+
"loss": 3.6804,
|
| 2094 |
+
"step": 7400
|
| 2095 |
+
},
|
| 2096 |
+
{
|
| 2097 |
+
"epoch": 1.6614455135377042,
|
| 2098 |
+
"grad_norm": 10.292470932006836,
|
| 2099 |
+
"learning_rate": 2.4817669484501907e-05,
|
| 2100 |
+
"loss": 3.5316,
|
| 2101 |
+
"step": 7425
|
| 2102 |
+
},
|
| 2103 |
+
{
|
| 2104 |
+
"epoch": 1.6670396061758783,
|
| 2105 |
+
"grad_norm": 13.621623039245605,
|
| 2106 |
+
"learning_rate": 2.4714072600696174e-05,
|
| 2107 |
+
"loss": 3.4568,
|
| 2108 |
+
"step": 7450
|
| 2109 |
+
},
|
| 2110 |
+
{
|
| 2111 |
+
"epoch": 1.6726336988140522,
|
| 2112 |
+
"grad_norm": 11.503890991210938,
|
| 2113 |
+
"learning_rate": 2.4610475716890437e-05,
|
| 2114 |
+
"loss": 3.3641,
|
| 2115 |
+
"step": 7475
|
| 2116 |
+
},
|
| 2117 |
+
{
|
| 2118 |
+
"epoch": 1.6782277914522266,
|
| 2119 |
+
"grad_norm": 13.564423561096191,
|
| 2120 |
+
"learning_rate": 2.45068788330847e-05,
|
| 2121 |
+
"loss": 3.5669,
|
| 2122 |
+
"step": 7500
|
| 2123 |
+
},
|
| 2124 |
+
{
|
| 2125 |
+
"epoch": 1.6838218840904005,
|
| 2126 |
+
"grad_norm": 9.360475540161133,
|
| 2127 |
+
"learning_rate": 2.4403281949278967e-05,
|
| 2128 |
+
"loss": 3.6688,
|
| 2129 |
+
"step": 7525
|
| 2130 |
+
},
|
| 2131 |
+
{
|
| 2132 |
+
"epoch": 1.6894159767285746,
|
| 2133 |
+
"grad_norm": 13.366150856018066,
|
| 2134 |
+
"learning_rate": 2.4299685065473234e-05,
|
| 2135 |
+
"loss": 3.3199,
|
| 2136 |
+
"step": 7550
|
| 2137 |
+
},
|
| 2138 |
+
{
|
| 2139 |
+
"epoch": 1.6950100693667487,
|
| 2140 |
+
"grad_norm": 10.094856262207031,
|
| 2141 |
+
"learning_rate": 2.4196088181667497e-05,
|
| 2142 |
+
"loss": 3.4633,
|
| 2143 |
+
"step": 7575
|
| 2144 |
+
},
|
| 2145 |
+
{
|
| 2146 |
+
"epoch": 1.7006041620049228,
|
| 2147 |
+
"grad_norm": 9.134540557861328,
|
| 2148 |
+
"learning_rate": 2.409249129786176e-05,
|
| 2149 |
+
"loss": 3.6078,
|
| 2150 |
+
"step": 7600
|
| 2151 |
+
},
|
| 2152 |
+
{
|
| 2153 |
+
"epoch": 1.706198254643097,
|
| 2154 |
+
"grad_norm": 11.252095222473145,
|
| 2155 |
+
"learning_rate": 2.3988894414056027e-05,
|
| 2156 |
+
"loss": 3.7198,
|
| 2157 |
+
"step": 7625
|
| 2158 |
+
},
|
| 2159 |
+
{
|
| 2160 |
+
"epoch": 1.7117923472812708,
|
| 2161 |
+
"grad_norm": 6.201746940612793,
|
| 2162 |
+
"learning_rate": 2.388529753025029e-05,
|
| 2163 |
+
"loss": 3.5264,
|
| 2164 |
+
"step": 7650
|
| 2165 |
+
},
|
| 2166 |
+
{
|
| 2167 |
+
"epoch": 1.7173864399194452,
|
| 2168 |
+
"grad_norm": 12.331997871398926,
|
| 2169 |
+
"learning_rate": 2.3781700646444557e-05,
|
| 2170 |
+
"loss": 3.4556,
|
| 2171 |
+
"step": 7675
|
| 2172 |
+
},
|
| 2173 |
+
{
|
| 2174 |
+
"epoch": 1.722980532557619,
|
| 2175 |
+
"grad_norm": 12.859444618225098,
|
| 2176 |
+
"learning_rate": 2.367810376263882e-05,
|
| 2177 |
+
"loss": 3.4533,
|
| 2178 |
+
"step": 7700
|
| 2179 |
+
},
|
| 2180 |
+
{
|
| 2181 |
+
"epoch": 1.7285746251957932,
|
| 2182 |
+
"grad_norm": 13.243956565856934,
|
| 2183 |
+
"learning_rate": 2.3574506878833087e-05,
|
| 2184 |
+
"loss": 3.7355,
|
| 2185 |
+
"step": 7725
|
| 2186 |
+
},
|
| 2187 |
+
{
|
| 2188 |
+
"epoch": 1.7341687178339673,
|
| 2189 |
+
"grad_norm": 11.75436019897461,
|
| 2190 |
+
"learning_rate": 2.347090999502735e-05,
|
| 2191 |
+
"loss": 3.6108,
|
| 2192 |
+
"step": 7750
|
| 2193 |
+
},
|
| 2194 |
+
{
|
| 2195 |
+
"epoch": 1.7397628104721414,
|
| 2196 |
+
"grad_norm": 13.429585456848145,
|
| 2197 |
+
"learning_rate": 2.3367313111221614e-05,
|
| 2198 |
+
"loss": 3.7925,
|
| 2199 |
+
"step": 7775
|
| 2200 |
+
},
|
| 2201 |
+
{
|
| 2202 |
+
"epoch": 1.7453569031103155,
|
| 2203 |
+
"grad_norm": 11.077943801879883,
|
| 2204 |
+
"learning_rate": 2.326371622741588e-05,
|
| 2205 |
+
"loss": 3.6788,
|
| 2206 |
+
"step": 7800
|
| 2207 |
+
},
|
| 2208 |
+
{
|
| 2209 |
+
"epoch": 1.7509509957484894,
|
| 2210 |
+
"grad_norm": 7.700258731842041,
|
| 2211 |
+
"learning_rate": 2.3160119343610144e-05,
|
| 2212 |
+
"loss": 3.4562,
|
| 2213 |
+
"step": 7825
|
| 2214 |
+
},
|
| 2215 |
+
{
|
| 2216 |
+
"epoch": 1.7565450883866638,
|
| 2217 |
+
"grad_norm": 11.34974479675293,
|
| 2218 |
+
"learning_rate": 2.305652245980441e-05,
|
| 2219 |
+
"loss": 3.3855,
|
| 2220 |
+
"step": 7850
|
| 2221 |
+
},
|
| 2222 |
+
{
|
| 2223 |
+
"epoch": 1.7621391810248377,
|
| 2224 |
+
"grad_norm": 11.598840713500977,
|
| 2225 |
+
"learning_rate": 2.2952925575998674e-05,
|
| 2226 |
+
"loss": 3.6097,
|
| 2227 |
+
"step": 7875
|
| 2228 |
+
},
|
| 2229 |
+
{
|
| 2230 |
+
"epoch": 1.7677332736630118,
|
| 2231 |
+
"grad_norm": 11.69258975982666,
|
| 2232 |
+
"learning_rate": 2.284932869219294e-05,
|
| 2233 |
+
"loss": 3.6146,
|
| 2234 |
+
"step": 7900
|
| 2235 |
+
},
|
| 2236 |
+
{
|
| 2237 |
+
"epoch": 1.773327366301186,
|
| 2238 |
+
"grad_norm": 10.501328468322754,
|
| 2239 |
+
"learning_rate": 2.2745731808387204e-05,
|
| 2240 |
+
"loss": 3.6597,
|
| 2241 |
+
"step": 7925
|
| 2242 |
+
},
|
| 2243 |
+
{
|
| 2244 |
+
"epoch": 1.77892145893936,
|
| 2245 |
+
"grad_norm": 12.03715705871582,
|
| 2246 |
+
"learning_rate": 2.264213492458147e-05,
|
| 2247 |
+
"loss": 3.6975,
|
| 2248 |
+
"step": 7950
|
| 2249 |
+
},
|
| 2250 |
+
{
|
| 2251 |
+
"epoch": 1.7845155515775342,
|
| 2252 |
+
"grad_norm": 13.386404991149902,
|
| 2253 |
+
"learning_rate": 2.2538538040775734e-05,
|
| 2254 |
+
"loss": 3.6164,
|
| 2255 |
+
"step": 7975
|
| 2256 |
+
},
|
| 2257 |
+
{
|
| 2258 |
+
"epoch": 1.790109644215708,
|
| 2259 |
+
"grad_norm": 9.877335548400879,
|
| 2260 |
+
"learning_rate": 2.2434941156969997e-05,
|
| 2261 |
+
"loss": 3.614,
|
| 2262 |
+
"step": 8000
|
| 2263 |
+
},
|
| 2264 |
+
{
|
| 2265 |
+
"epoch": 1.7957037368538824,
|
| 2266 |
+
"grad_norm": 12.025654792785645,
|
| 2267 |
+
"learning_rate": 2.2335488148516494e-05,
|
| 2268 |
+
"loss": 3.743,
|
| 2269 |
+
"step": 8025
|
| 2270 |
+
},
|
| 2271 |
+
{
|
| 2272 |
+
"epoch": 1.8012978294920563,
|
| 2273 |
+
"grad_norm": 12.508997917175293,
|
| 2274 |
+
"learning_rate": 2.2231891264710757e-05,
|
| 2275 |
+
"loss": 3.7309,
|
| 2276 |
+
"step": 8050
|
| 2277 |
+
},
|
| 2278 |
+
{
|
| 2279 |
+
"epoch": 1.8068919221302304,
|
| 2280 |
+
"grad_norm": 12.994415283203125,
|
| 2281 |
+
"learning_rate": 2.2128294380905024e-05,
|
| 2282 |
+
"loss": 3.5284,
|
| 2283 |
+
"step": 8075
|
| 2284 |
+
},
|
| 2285 |
+
{
|
| 2286 |
+
"epoch": 1.8124860147684045,
|
| 2287 |
+
"grad_norm": 12.859843254089355,
|
| 2288 |
+
"learning_rate": 2.202469749709929e-05,
|
| 2289 |
+
"loss": 3.686,
|
| 2290 |
+
"step": 8100
|
| 2291 |
+
},
|
| 2292 |
+
{
|
| 2293 |
+
"epoch": 1.8180801074065787,
|
| 2294 |
+
"grad_norm": 15.91470718383789,
|
| 2295 |
+
"learning_rate": 2.192110061329355e-05,
|
| 2296 |
+
"loss": 3.3395,
|
| 2297 |
+
"step": 8125
|
| 2298 |
+
},
|
| 2299 |
+
{
|
| 2300 |
+
"epoch": 1.8236742000447528,
|
| 2301 |
+
"grad_norm": 10.755178451538086,
|
| 2302 |
+
"learning_rate": 2.1817503729487817e-05,
|
| 2303 |
+
"loss": 3.6407,
|
| 2304 |
+
"step": 8150
|
| 2305 |
+
},
|
| 2306 |
+
{
|
| 2307 |
+
"epoch": 1.8292682926829267,
|
| 2308 |
+
"grad_norm": 10.679194450378418,
|
| 2309 |
+
"learning_rate": 2.1713906845682084e-05,
|
| 2310 |
+
"loss": 3.5509,
|
| 2311 |
+
"step": 8175
|
| 2312 |
+
},
|
| 2313 |
+
{
|
| 2314 |
+
"epoch": 1.834862385321101,
|
| 2315 |
+
"grad_norm": 18.6633243560791,
|
| 2316 |
+
"learning_rate": 2.1610309961876348e-05,
|
| 2317 |
+
"loss": 3.5166,
|
| 2318 |
+
"step": 8200
|
| 2319 |
+
},
|
| 2320 |
+
{
|
| 2321 |
+
"epoch": 1.840456477959275,
|
| 2322 |
+
"grad_norm": 7.9321112632751465,
|
| 2323 |
+
"learning_rate": 2.150671307807061e-05,
|
| 2324 |
+
"loss": 3.36,
|
| 2325 |
+
"step": 8225
|
| 2326 |
+
},
|
| 2327 |
+
{
|
| 2328 |
+
"epoch": 1.8460505705974493,
|
| 2329 |
+
"grad_norm": 10.757131576538086,
|
| 2330 |
+
"learning_rate": 2.1403116194264878e-05,
|
| 2331 |
+
"loss": 3.4811,
|
| 2332 |
+
"step": 8250
|
| 2333 |
+
},
|
| 2334 |
+
{
|
| 2335 |
+
"epoch": 1.8516446632356232,
|
| 2336 |
+
"grad_norm": 15.632428169250488,
|
| 2337 |
+
"learning_rate": 2.1299519310459144e-05,
|
| 2338 |
+
"loss": 3.5502,
|
| 2339 |
+
"step": 8275
|
| 2340 |
+
},
|
| 2341 |
+
{
|
| 2342 |
+
"epoch": 1.8572387558737973,
|
| 2343 |
+
"grad_norm": 19.17276954650879,
|
| 2344 |
+
"learning_rate": 2.1195922426653408e-05,
|
| 2345 |
+
"loss": 3.5601,
|
| 2346 |
+
"step": 8300
|
| 2347 |
+
},
|
| 2348 |
+
{
|
| 2349 |
+
"epoch": 1.8628328485119714,
|
| 2350 |
+
"grad_norm": 11.047025680541992,
|
| 2351 |
+
"learning_rate": 2.109232554284767e-05,
|
| 2352 |
+
"loss": 3.6373,
|
| 2353 |
+
"step": 8325
|
| 2354 |
+
},
|
| 2355 |
+
{
|
| 2356 |
+
"epoch": 1.8684269411501453,
|
| 2357 |
+
"grad_norm": 15.699575424194336,
|
| 2358 |
+
"learning_rate": 2.0988728659041938e-05,
|
| 2359 |
+
"loss": 3.7378,
|
| 2360 |
+
"step": 8350
|
| 2361 |
+
},
|
| 2362 |
+
{
|
| 2363 |
+
"epoch": 1.8740210337883196,
|
| 2364 |
+
"grad_norm": 13.09723949432373,
|
| 2365 |
+
"learning_rate": 2.08851317752362e-05,
|
| 2366 |
+
"loss": 3.8717,
|
| 2367 |
+
"step": 8375
|
| 2368 |
+
},
|
| 2369 |
+
{
|
| 2370 |
+
"epoch": 1.8796151264264935,
|
| 2371 |
+
"grad_norm": 8.441289901733398,
|
| 2372 |
+
"learning_rate": 2.0781534891430468e-05,
|
| 2373 |
+
"loss": 3.2558,
|
| 2374 |
+
"step": 8400
|
| 2375 |
+
},
|
| 2376 |
+
{
|
| 2377 |
+
"epoch": 1.8852092190646679,
|
| 2378 |
+
"grad_norm": 12.046778678894043,
|
| 2379 |
+
"learning_rate": 2.067793800762473e-05,
|
| 2380 |
+
"loss": 3.3936,
|
| 2381 |
+
"step": 8425
|
| 2382 |
+
},
|
| 2383 |
+
{
|
| 2384 |
+
"epoch": 1.8908033117028418,
|
| 2385 |
+
"grad_norm": 10.983031272888184,
|
| 2386 |
+
"learning_rate": 2.0574341123818998e-05,
|
| 2387 |
+
"loss": 3.4321,
|
| 2388 |
+
"step": 8450
|
| 2389 |
+
},
|
| 2390 |
+
{
|
| 2391 |
+
"epoch": 1.896397404341016,
|
| 2392 |
+
"grad_norm": 12.274590492248535,
|
| 2393 |
+
"learning_rate": 2.047074424001326e-05,
|
| 2394 |
+
"loss": 3.4638,
|
| 2395 |
+
"step": 8475
|
| 2396 |
+
},
|
| 2397 |
+
{
|
| 2398 |
+
"epoch": 1.90199149697919,
|
| 2399 |
+
"grad_norm": 15.135939598083496,
|
| 2400 |
+
"learning_rate": 2.0367147356207524e-05,
|
| 2401 |
+
"loss": 3.5686,
|
| 2402 |
+
"step": 8500
|
| 2403 |
+
},
|
| 2404 |
+
{
|
| 2405 |
+
"epoch": 1.907585589617364,
|
| 2406 |
+
"grad_norm": 11.194721221923828,
|
| 2407 |
+
"learning_rate": 2.026355047240179e-05,
|
| 2408 |
+
"loss": 3.7067,
|
| 2409 |
+
"step": 8525
|
| 2410 |
+
},
|
| 2411 |
+
{
|
| 2412 |
+
"epoch": 1.9131796822555382,
|
| 2413 |
+
"grad_norm": 15.062312126159668,
|
| 2414 |
+
"learning_rate": 2.0159953588596054e-05,
|
| 2415 |
+
"loss": 3.5299,
|
| 2416 |
+
"step": 8550
|
| 2417 |
+
},
|
| 2418 |
+
{
|
| 2419 |
+
"epoch": 1.9187737748937121,
|
| 2420 |
+
"grad_norm": 12.282342910766602,
|
| 2421 |
+
"learning_rate": 2.005635670479032e-05,
|
| 2422 |
+
"loss": 3.5803,
|
| 2423 |
+
"step": 8575
|
| 2424 |
+
},
|
| 2425 |
+
{
|
| 2426 |
+
"epoch": 1.9243678675318865,
|
| 2427 |
+
"grad_norm": 14.33022689819336,
|
| 2428 |
+
"learning_rate": 1.9952759820984584e-05,
|
| 2429 |
+
"loss": 3.5835,
|
| 2430 |
+
"step": 8600
|
| 2431 |
+
},
|
| 2432 |
+
{
|
| 2433 |
+
"epoch": 1.9299619601700604,
|
| 2434 |
+
"grad_norm": 8.249588966369629,
|
| 2435 |
+
"learning_rate": 1.984916293717885e-05,
|
| 2436 |
+
"loss": 3.2694,
|
| 2437 |
+
"step": 8625
|
| 2438 |
+
},
|
| 2439 |
+
{
|
| 2440 |
+
"epoch": 1.9355560528082345,
|
| 2441 |
+
"grad_norm": 9.1649169921875,
|
| 2442 |
+
"learning_rate": 1.9745566053373115e-05,
|
| 2443 |
+
"loss": 3.6105,
|
| 2444 |
+
"step": 8650
|
| 2445 |
+
},
|
| 2446 |
+
{
|
| 2447 |
+
"epoch": 1.9411501454464086,
|
| 2448 |
+
"grad_norm": 8.755537986755371,
|
| 2449 |
+
"learning_rate": 1.964196916956738e-05,
|
| 2450 |
+
"loss": 3.5554,
|
| 2451 |
+
"step": 8675
|
| 2452 |
+
},
|
| 2453 |
+
{
|
| 2454 |
+
"epoch": 1.9467442380845825,
|
| 2455 |
+
"grad_norm": 8.148399353027344,
|
| 2456 |
+
"learning_rate": 1.9538372285761645e-05,
|
| 2457 |
+
"loss": 3.5073,
|
| 2458 |
+
"step": 8700
|
| 2459 |
+
},
|
| 2460 |
+
{
|
| 2461 |
+
"epoch": 1.9523383307227569,
|
| 2462 |
+
"grad_norm": 10.864067077636719,
|
| 2463 |
+
"learning_rate": 1.943477540195591e-05,
|
| 2464 |
+
"loss": 3.6951,
|
| 2465 |
+
"step": 8725
|
| 2466 |
+
},
|
| 2467 |
+
{
|
| 2468 |
+
"epoch": 1.9579324233609308,
|
| 2469 |
+
"grad_norm": 13.049738883972168,
|
| 2470 |
+
"learning_rate": 1.9331178518150175e-05,
|
| 2471 |
+
"loss": 3.5186,
|
| 2472 |
+
"step": 8750
|
| 2473 |
+
},
|
| 2474 |
+
{
|
| 2475 |
+
"epoch": 1.963526515999105,
|
| 2476 |
+
"grad_norm": 5.955536842346191,
|
| 2477 |
+
"learning_rate": 1.922758163434444e-05,
|
| 2478 |
+
"loss": 3.479,
|
| 2479 |
+
"step": 8775
|
| 2480 |
+
},
|
| 2481 |
+
{
|
| 2482 |
+
"epoch": 1.969120608637279,
|
| 2483 |
+
"grad_norm": 12.35295581817627,
|
| 2484 |
+
"learning_rate": 1.9123984750538705e-05,
|
| 2485 |
+
"loss": 3.5103,
|
| 2486 |
+
"step": 8800
|
| 2487 |
+
},
|
| 2488 |
+
{
|
| 2489 |
+
"epoch": 1.9747147012754531,
|
| 2490 |
+
"grad_norm": 9.945602416992188,
|
| 2491 |
+
"learning_rate": 1.9020387866732968e-05,
|
| 2492 |
+
"loss": 3.6919,
|
| 2493 |
+
"step": 8825
|
| 2494 |
+
},
|
| 2495 |
+
{
|
| 2496 |
+
"epoch": 1.9803087939136272,
|
| 2497 |
+
"grad_norm": 9.716672897338867,
|
| 2498 |
+
"learning_rate": 1.8916790982927235e-05,
|
| 2499 |
+
"loss": 3.7097,
|
| 2500 |
+
"step": 8850
|
| 2501 |
+
},
|
| 2502 |
+
{
|
| 2503 |
+
"epoch": 1.9859028865518011,
|
| 2504 |
+
"grad_norm": 7.3651041984558105,
|
| 2505 |
+
"learning_rate": 1.88131940991215e-05,
|
| 2506 |
+
"loss": 3.4111,
|
| 2507 |
+
"step": 8875
|
| 2508 |
+
},
|
| 2509 |
+
{
|
| 2510 |
+
"epoch": 1.9914969791899755,
|
| 2511 |
+
"grad_norm": 11.258004188537598,
|
| 2512 |
+
"learning_rate": 1.8709597215315765e-05,
|
| 2513 |
+
"loss": 3.4201,
|
| 2514 |
+
"step": 8900
|
| 2515 |
+
},
|
| 2516 |
+
{
|
| 2517 |
+
"epoch": 1.9970910718281494,
|
| 2518 |
+
"grad_norm": 14.962812423706055,
|
| 2519 |
+
"learning_rate": 1.8606000331510028e-05,
|
| 2520 |
+
"loss": 3.4142,
|
| 2521 |
+
"step": 8925
|
| 2522 |
+
},
|
| 2523 |
+
{
|
| 2524 |
+
"epoch": 2.0,
|
| 2525 |
+
"eval_gen_len": 61.8854,
|
| 2526 |
+
"eval_loss": 4.248934268951416,
|
| 2527 |
+
"eval_rouge1": 25.7685,
|
| 2528 |
+
"eval_rouge2": 9.8226,
|
| 2529 |
+
"eval_rougeL": 24.6426,
|
| 2530 |
+
"eval_rougeLsum": 24.9756,
|
| 2531 |
+
"eval_runtime": 700.259,
|
| 2532 |
+
"eval_samples_per_second": 1.595,
|
| 2533 |
+
"eval_steps_per_second": 0.4,
|
| 2534 |
+
"step": 8938
|
| 2535 |
+
}
|
| 2536 |
+
],
|
| 2537 |
+
"logging_steps": 25,
|
| 2538 |
+
"max_steps": 13407,
|
| 2539 |
+
"num_input_tokens_seen": 0,
|
| 2540 |
+
"num_train_epochs": 3,
|
| 2541 |
+
"save_steps": 500,
|
| 2542 |
+
"stateful_callbacks": {
|
| 2543 |
+
"EarlyStoppingCallback": {
|
| 2544 |
+
"args": {
|
| 2545 |
+
"early_stopping_patience": 5,
|
| 2546 |
+
"early_stopping_threshold": 0.01
|
| 2547 |
+
},
|
| 2548 |
+
"attributes": {
|
| 2549 |
+
"early_stopping_patience_counter": 0
|
| 2550 |
+
}
|
| 2551 |
+
},
|
| 2552 |
+
"TrainerControl": {
|
| 2553 |
+
"args": {
|
| 2554 |
+
"should_epoch_stop": false,
|
| 2555 |
+
"should_evaluate": false,
|
| 2556 |
+
"should_log": false,
|
| 2557 |
+
"should_save": true,
|
| 2558 |
+
"should_training_stop": false
|
| 2559 |
+
},
|
| 2560 |
+
"attributes": {}
|
| 2561 |
+
}
|
| 2562 |
+
},
|
| 2563 |
+
"total_flos": 1151792816578560.0,
|
| 2564 |
+
"train_batch_size": 2,
|
| 2565 |
+
"trial_name": null,
|
| 2566 |
+
"trial_params": null
|
| 2567 |
+
}
|
checkpoint-8938/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:01e34e737c610c7cc15ed5a164693fecd2333673e9691a688c4593b566063c81
|
| 3 |
+
size 5304
|
checkpoint-8938/vocab.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
config.json
ADDED
|
@@ -0,0 +1,70 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "facebook/bart-large-cnn",
|
| 3 |
+
"_num_labels": 3,
|
| 4 |
+
"activation_dropout": 0.0,
|
| 5 |
+
"activation_function": "gelu",
|
| 6 |
+
"add_final_layer_norm": false,
|
| 7 |
+
"architectures": [
|
| 8 |
+
"BartForConditionalGeneration"
|
| 9 |
+
],
|
| 10 |
+
"attention_dropout": 0.0,
|
| 11 |
+
"bos_token_id": 0,
|
| 12 |
+
"classif_dropout": 0.0,
|
| 13 |
+
"classifier_dropout": 0.0,
|
| 14 |
+
"d_model": 1024,
|
| 15 |
+
"decoder_attention_heads": 16,
|
| 16 |
+
"decoder_ffn_dim": 4096,
|
| 17 |
+
"decoder_layerdrop": 0.0,
|
| 18 |
+
"decoder_layers": 12,
|
| 19 |
+
"decoder_start_token_id": 2,
|
| 20 |
+
"dropout": 0.1,
|
| 21 |
+
"early_stopping": true,
|
| 22 |
+
"encoder_attention_heads": 16,
|
| 23 |
+
"encoder_ffn_dim": 4096,
|
| 24 |
+
"encoder_layerdrop": 0.0,
|
| 25 |
+
"encoder_layers": 12,
|
| 26 |
+
"eos_token_id": 2,
|
| 27 |
+
"force_bos_token_to_be_generated": true,
|
| 28 |
+
"forced_bos_token_id": 0,
|
| 29 |
+
"forced_eos_token_id": 2,
|
| 30 |
+
"gradient_checkpointing": false,
|
| 31 |
+
"id2label": {
|
| 32 |
+
"0": "LABEL_0",
|
| 33 |
+
"1": "LABEL_1",
|
| 34 |
+
"2": "LABEL_2"
|
| 35 |
+
},
|
| 36 |
+
"init_std": 0.02,
|
| 37 |
+
"is_encoder_decoder": true,
|
| 38 |
+
"label2id": {
|
| 39 |
+
"LABEL_0": 0,
|
| 40 |
+
"LABEL_1": 1,
|
| 41 |
+
"LABEL_2": 2
|
| 42 |
+
},
|
| 43 |
+
"length_penalty": 2.0,
|
| 44 |
+
"max_length": 142,
|
| 45 |
+
"max_position_embeddings": 1024,
|
| 46 |
+
"min_length": 56,
|
| 47 |
+
"model_type": "bart",
|
| 48 |
+
"no_repeat_ngram_size": 3,
|
| 49 |
+
"normalize_before": false,
|
| 50 |
+
"num_beams": 4,
|
| 51 |
+
"num_hidden_layers": 12,
|
| 52 |
+
"output_past": true,
|
| 53 |
+
"pad_token_id": 1,
|
| 54 |
+
"prefix": " ",
|
| 55 |
+
"scale_embedding": false,
|
| 56 |
+
"task_specific_params": {
|
| 57 |
+
"summarization": {
|
| 58 |
+
"early_stopping": true,
|
| 59 |
+
"length_penalty": 2.0,
|
| 60 |
+
"max_length": 142,
|
| 61 |
+
"min_length": 56,
|
| 62 |
+
"no_repeat_ngram_size": 3,
|
| 63 |
+
"num_beams": 4
|
| 64 |
+
}
|
| 65 |
+
},
|
| 66 |
+
"torch_dtype": "float32",
|
| 67 |
+
"transformers_version": "4.41.0",
|
| 68 |
+
"use_cache": true,
|
| 69 |
+
"vocab_size": 50265
|
| 70 |
+
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token_id": 0,
|
| 3 |
+
"decoder_start_token_id": 2,
|
| 4 |
+
"early_stopping": true,
|
| 5 |
+
"eos_token_id": 2,
|
| 6 |
+
"forced_bos_token_id": 0,
|
| 7 |
+
"forced_eos_token_id": 2,
|
| 8 |
+
"length_penalty": 2.0,
|
| 9 |
+
"max_length": 142,
|
| 10 |
+
"min_length": 56,
|
| 11 |
+
"no_repeat_ngram_size": 3,
|
| 12 |
+
"num_beams": 4,
|
| 13 |
+
"pad_token_id": 1,
|
| 14 |
+
"transformers_version": "4.41.0",
|
| 15 |
+
"use_cache": false
|
| 16 |
+
}
|
merges.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ddd6fefb4de9b6731b7efcc997fdb1cc4db285111506952feaefad76b0cf7029
|
| 3 |
+
size 1625426996
|
runs/Jun11_09-59-45_r-riken01-t5-base-njeyrhxb-70853-v6p9n/events.out.tfevents.1718099995.r-riken01-t5-base-njeyrhxb-70853-v6p9n.89.0
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1975b3f11670ffece3daafd9757177cb3a1e42fd643692d800cb91443b8c7d44
|
| 3 |
+
size 120990
|
runs/Jun11_09-59-45_r-riken01-t5-base-njeyrhxb-70853-v6p9n/events.out.tfevents.1718107440.r-riken01-t5-base-njeyrhxb-70853-v6p9n.89.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:16a4b7dc3ee69997a329da242f480dc47666bf51ef5b91042c0df44fd10f6fdc
|
| 3 |
+
size 613
|
special_tokens_map.json
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token": "<s>",
|
| 3 |
+
"cls_token": "<s>",
|
| 4 |
+
"eos_token": "</s>",
|
| 5 |
+
"mask_token": {
|
| 6 |
+
"content": "<mask>",
|
| 7 |
+
"lstrip": true,
|
| 8 |
+
"normalized": true,
|
| 9 |
+
"rstrip": false,
|
| 10 |
+
"single_word": false
|
| 11 |
+
},
|
| 12 |
+
"pad_token": "<pad>",
|
| 13 |
+
"sep_token": "</s>",
|
| 14 |
+
"unk_token": "<unk>"
|
| 15 |
+
}
|
tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1,57 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_prefix_space": false,
|
| 3 |
+
"added_tokens_decoder": {
|
| 4 |
+
"0": {
|
| 5 |
+
"content": "<s>",
|
| 6 |
+
"lstrip": false,
|
| 7 |
+
"normalized": true,
|
| 8 |
+
"rstrip": false,
|
| 9 |
+
"single_word": false,
|
| 10 |
+
"special": true
|
| 11 |
+
},
|
| 12 |
+
"1": {
|
| 13 |
+
"content": "<pad>",
|
| 14 |
+
"lstrip": false,
|
| 15 |
+
"normalized": true,
|
| 16 |
+
"rstrip": false,
|
| 17 |
+
"single_word": false,
|
| 18 |
+
"special": true
|
| 19 |
+
},
|
| 20 |
+
"2": {
|
| 21 |
+
"content": "</s>",
|
| 22 |
+
"lstrip": false,
|
| 23 |
+
"normalized": true,
|
| 24 |
+
"rstrip": false,
|
| 25 |
+
"single_word": false,
|
| 26 |
+
"special": true
|
| 27 |
+
},
|
| 28 |
+
"3": {
|
| 29 |
+
"content": "<unk>",
|
| 30 |
+
"lstrip": false,
|
| 31 |
+
"normalized": true,
|
| 32 |
+
"rstrip": false,
|
| 33 |
+
"single_word": false,
|
| 34 |
+
"special": true
|
| 35 |
+
},
|
| 36 |
+
"50264": {
|
| 37 |
+
"content": "<mask>",
|
| 38 |
+
"lstrip": true,
|
| 39 |
+
"normalized": true,
|
| 40 |
+
"rstrip": false,
|
| 41 |
+
"single_word": false,
|
| 42 |
+
"special": true
|
| 43 |
+
}
|
| 44 |
+
},
|
| 45 |
+
"bos_token": "<s>",
|
| 46 |
+
"clean_up_tokenization_spaces": true,
|
| 47 |
+
"cls_token": "<s>",
|
| 48 |
+
"eos_token": "</s>",
|
| 49 |
+
"errors": "replace",
|
| 50 |
+
"mask_token": "<mask>",
|
| 51 |
+
"model_max_length": 1000000000000000019884624838656,
|
| 52 |
+
"pad_token": "<pad>",
|
| 53 |
+
"sep_token": "</s>",
|
| 54 |
+
"tokenizer_class": "BartTokenizer",
|
| 55 |
+
"trim_offsets": true,
|
| 56 |
+
"unk_token": "<unk>"
|
| 57 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:01e34e737c610c7cc15ed5a164693fecd2333673e9691a688c4593b566063c81
|
| 3 |
+
size 5304
|
training_params.json
ADDED
|
@@ -0,0 +1,37 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"data_path": "autotrain-l6hey-orl0t/autotrain-data",
|
| 3 |
+
"model": "facebook/bart-large-cnn",
|
| 4 |
+
"username": "riken01",
|
| 5 |
+
"seed": 42,
|
| 6 |
+
"train_split": "train",
|
| 7 |
+
"valid_split": "validation",
|
| 8 |
+
"project_name": "autotrain-l6hey-orl0t",
|
| 9 |
+
"push_to_hub": true,
|
| 10 |
+
"text_column": "autotrain_text",
|
| 11 |
+
"target_column": "autotrain_label",
|
| 12 |
+
"lr": 5e-05,
|
| 13 |
+
"epochs": 3,
|
| 14 |
+
"max_seq_length": 128,
|
| 15 |
+
"max_target_length": 128,
|
| 16 |
+
"batch_size": 2,
|
| 17 |
+
"warmup_ratio": 0.1,
|
| 18 |
+
"gradient_accumulation": 1,
|
| 19 |
+
"optimizer": "adamw_torch",
|
| 20 |
+
"scheduler": "linear",
|
| 21 |
+
"weight_decay": 0.0,
|
| 22 |
+
"max_grad_norm": 1.0,
|
| 23 |
+
"logging_steps": -1,
|
| 24 |
+
"evaluation_strategy": "epoch",
|
| 25 |
+
"auto_find_batch_size": false,
|
| 26 |
+
"mixed_precision": "fp16",
|
| 27 |
+
"save_total_limit": 1,
|
| 28 |
+
"peft": false,
|
| 29 |
+
"quantization": "int4",
|
| 30 |
+
"lora_r": 16,
|
| 31 |
+
"lora_alpha": 32,
|
| 32 |
+
"lora_dropout": 0.05,
|
| 33 |
+
"target_modules": "all-linear",
|
| 34 |
+
"log": "tensorboard",
|
| 35 |
+
"early_stopping_patience": 5,
|
| 36 |
+
"early_stopping_threshold": 0.01
|
| 37 |
+
}
|
vocab.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|