Upload 11 files
Browse filesfinetuned this model on the unie_synthetic.json from knowledgator/gliner-multitask-large-v0.5 repo on huggingface.
hyper parameters used:
learning_rate=5e-6,
weight_decay=0.01,
others_lr=7e-6,
others_weight_decay=0.01,
lr_scheduler_type="linear",
warmup_ratio=0.1,
focal_loss_alpha=0.75,
focal_loss_gamma=0,
- added_tokens.json +6 -0
- gliner_config.json +137 -0
- optimizer.pt +3 -0
- pytorch_model.bin +3 -0
- rng_state.pth +3 -0
- scheduler.pt +3 -0
- special_tokens_map.json +15 -0
- spm.model +3 -0
- tokenizer.json +0 -0
- tokenizer_config.json +82 -0
- trainer_state.json +1433 -0
added_tokens.json
ADDED
|
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"<<ENT>>": 128002,
|
| 3 |
+
"<<SEP>>": 128003,
|
| 4 |
+
"[FLERT]": 128001,
|
| 5 |
+
"[MASK]": 128000
|
| 6 |
+
}
|
gliner_config.json
ADDED
|
@@ -0,0 +1,137 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"class_token_index": 128002,
|
| 3 |
+
"dropout": 0.1,
|
| 4 |
+
"embed_ent_token": true,
|
| 5 |
+
"encoder_config": {
|
| 6 |
+
"_name_or_path": "microsoft/deberta-v3-large",
|
| 7 |
+
"add_cross_attention": false,
|
| 8 |
+
"architectures": null,
|
| 9 |
+
"attention_probs_dropout_prob": 0.1,
|
| 10 |
+
"bad_words_ids": null,
|
| 11 |
+
"begin_suppress_tokens": null,
|
| 12 |
+
"bos_token_id": null,
|
| 13 |
+
"chunk_size_feed_forward": 0,
|
| 14 |
+
"cross_attention_hidden_size": null,
|
| 15 |
+
"decoder_start_token_id": null,
|
| 16 |
+
"diversity_penalty": 0.0,
|
| 17 |
+
"do_sample": false,
|
| 18 |
+
"early_stopping": false,
|
| 19 |
+
"encoder_no_repeat_ngram_size": 0,
|
| 20 |
+
"eos_token_id": null,
|
| 21 |
+
"exponential_decay_length_penalty": null,
|
| 22 |
+
"finetuning_task": null,
|
| 23 |
+
"forced_bos_token_id": null,
|
| 24 |
+
"forced_eos_token_id": null,
|
| 25 |
+
"hidden_act": "gelu",
|
| 26 |
+
"hidden_dropout_prob": 0.1,
|
| 27 |
+
"hidden_size": 1024,
|
| 28 |
+
"id2label": {
|
| 29 |
+
"0": "LABEL_0",
|
| 30 |
+
"1": "LABEL_1"
|
| 31 |
+
},
|
| 32 |
+
"initializer_range": 0.02,
|
| 33 |
+
"intermediate_size": 4096,
|
| 34 |
+
"is_decoder": false,
|
| 35 |
+
"is_encoder_decoder": false,
|
| 36 |
+
"label2id": {
|
| 37 |
+
"LABEL_0": 0,
|
| 38 |
+
"LABEL_1": 1
|
| 39 |
+
},
|
| 40 |
+
"layer_norm_eps": 1e-07,
|
| 41 |
+
"length_penalty": 1.0,
|
| 42 |
+
"max_length": 20,
|
| 43 |
+
"max_position_embeddings": 512,
|
| 44 |
+
"max_relative_positions": -1,
|
| 45 |
+
"min_length": 0,
|
| 46 |
+
"model_type": "deberta-v2",
|
| 47 |
+
"no_repeat_ngram_size": 0,
|
| 48 |
+
"norm_rel_ebd": "layer_norm",
|
| 49 |
+
"num_attention_heads": 16,
|
| 50 |
+
"num_beam_groups": 1,
|
| 51 |
+
"num_beams": 1,
|
| 52 |
+
"num_hidden_layers": 24,
|
| 53 |
+
"num_return_sequences": 1,
|
| 54 |
+
"output_attentions": false,
|
| 55 |
+
"output_hidden_states": false,
|
| 56 |
+
"output_scores": false,
|
| 57 |
+
"pad_token_id": 0,
|
| 58 |
+
"pooler_dropout": 0,
|
| 59 |
+
"pooler_hidden_act": "gelu",
|
| 60 |
+
"pooler_hidden_size": 1024,
|
| 61 |
+
"pos_att_type": [
|
| 62 |
+
"p2c",
|
| 63 |
+
"c2p"
|
| 64 |
+
],
|
| 65 |
+
"position_biased_input": false,
|
| 66 |
+
"position_buckets": 256,
|
| 67 |
+
"prefix": null,
|
| 68 |
+
"problem_type": null,
|
| 69 |
+
"pruned_heads": {},
|
| 70 |
+
"relative_attention": true,
|
| 71 |
+
"remove_invalid_values": false,
|
| 72 |
+
"repetition_penalty": 1.0,
|
| 73 |
+
"return_dict": true,
|
| 74 |
+
"return_dict_in_generate": false,
|
| 75 |
+
"sep_token_id": null,
|
| 76 |
+
"share_att_key": true,
|
| 77 |
+
"suppress_tokens": null,
|
| 78 |
+
"task_specific_params": null,
|
| 79 |
+
"temperature": 1.0,
|
| 80 |
+
"tf_legacy_loss": false,
|
| 81 |
+
"tie_encoder_decoder": false,
|
| 82 |
+
"tie_word_embeddings": true,
|
| 83 |
+
"tokenizer_class": null,
|
| 84 |
+
"top_k": 50,
|
| 85 |
+
"top_p": 1.0,
|
| 86 |
+
"torch_dtype": null,
|
| 87 |
+
"torchscript": false,
|
| 88 |
+
"type_vocab_size": 0,
|
| 89 |
+
"typical_p": 1.0,
|
| 90 |
+
"use_bfloat16": false,
|
| 91 |
+
"vocab_size": 128004
|
| 92 |
+
},
|
| 93 |
+
"ent_token": "<<ENT>>",
|
| 94 |
+
"eval_every": 1000,
|
| 95 |
+
"fine_tune": true,
|
| 96 |
+
"freeze_token_rep": false,
|
| 97 |
+
"fuse_layers": false,
|
| 98 |
+
"has_rnn": true,
|
| 99 |
+
"hidden_size": 512,
|
| 100 |
+
"labels_encoder": null,
|
| 101 |
+
"labels_encoder_config": null,
|
| 102 |
+
"log_dir": "logs_final",
|
| 103 |
+
"loss_alpha": 0.75,
|
| 104 |
+
"loss_gamma": 0,
|
| 105 |
+
"loss_reduction": "sum",
|
| 106 |
+
"lr_encoder": "5e-6",
|
| 107 |
+
"lr_others": "7e-6",
|
| 108 |
+
"max_len": 768,
|
| 109 |
+
"max_neg_type_ratio": 1,
|
| 110 |
+
"max_types": 30,
|
| 111 |
+
"max_width": 100,
|
| 112 |
+
"model_name": "microsoft/deberta-v3-large",
|
| 113 |
+
"model_type": "gliner",
|
| 114 |
+
"name": "token level gliner large",
|
| 115 |
+
"num_post_fusion_layers": 1,
|
| 116 |
+
"num_steps": 6000,
|
| 117 |
+
"post_fusion_schema": "",
|
| 118 |
+
"prev_path": "logs_large/model_120000",
|
| 119 |
+
"random_drop": true,
|
| 120 |
+
"root_dir": "gliner_logs",
|
| 121 |
+
"save_total_limit": 10,
|
| 122 |
+
"scheduler_type": "linear",
|
| 123 |
+
"sep_token": "<<SEP>>",
|
| 124 |
+
"shuffle_types": true,
|
| 125 |
+
"size_sup": -1,
|
| 126 |
+
"span_mode": "token_level",
|
| 127 |
+
"subtoken_pooling": "first",
|
| 128 |
+
"train_batch_size": 8,
|
| 129 |
+
"train_data": "../../data/unie_ner1.json",
|
| 130 |
+
"transformers_version": "4.45.2",
|
| 131 |
+
"val_data_dir": "none",
|
| 132 |
+
"vocab_size": 128004,
|
| 133 |
+
"warmup_ratio": 0.1,
|
| 134 |
+
"weight_decay_encoder": 0.01,
|
| 135 |
+
"weight_decay_other": 0.01,
|
| 136 |
+
"words_splitter_type": "whitespace"
|
| 137 |
+
}
|
optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e974730d91bc7e38d8a5d00c690d53f68d30adc65b6f6ee11fed87c9a3d1ca50
|
| 3 |
+
size 3522103130
|
pytorch_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1c3fdfb09eee136948ee20fc27696b8f4d879407924ecd6b534b41759c5ef86d
|
| 3 |
+
size 1761030542
|
rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d12b1a07ac630f3575f64c55a848ff8e11891b860f86114ef52ec4e09c5140d3
|
| 3 |
+
size 14244
|
scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:21ce244ebdb50b2e7b09459dcc189c7fdcc2ede64246d825a6a36f2a0d13dc53
|
| 3 |
+
size 1064
|
special_tokens_map.json
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token": "[CLS]",
|
| 3 |
+
"cls_token": "[CLS]",
|
| 4 |
+
"eos_token": "[SEP]",
|
| 5 |
+
"mask_token": "[MASK]",
|
| 6 |
+
"pad_token": "[PAD]",
|
| 7 |
+
"sep_token": "[SEP]",
|
| 8 |
+
"unk_token": {
|
| 9 |
+
"content": "[UNK]",
|
| 10 |
+
"lstrip": false,
|
| 11 |
+
"normalized": true,
|
| 12 |
+
"rstrip": false,
|
| 13 |
+
"single_word": false
|
| 14 |
+
}
|
| 15 |
+
}
|
spm.model
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c679fbf93643d19aab7ee10c0b99e460bdbc02fedf34b92b05af343b4af586fd
|
| 3 |
+
size 2464616
|
tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1,82 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"added_tokens_decoder": {
|
| 3 |
+
"0": {
|
| 4 |
+
"content": "[PAD]",
|
| 5 |
+
"lstrip": false,
|
| 6 |
+
"normalized": false,
|
| 7 |
+
"rstrip": false,
|
| 8 |
+
"single_word": false,
|
| 9 |
+
"special": true
|
| 10 |
+
},
|
| 11 |
+
"1": {
|
| 12 |
+
"content": "[CLS]",
|
| 13 |
+
"lstrip": false,
|
| 14 |
+
"normalized": false,
|
| 15 |
+
"rstrip": false,
|
| 16 |
+
"single_word": false,
|
| 17 |
+
"special": true
|
| 18 |
+
},
|
| 19 |
+
"2": {
|
| 20 |
+
"content": "[SEP]",
|
| 21 |
+
"lstrip": false,
|
| 22 |
+
"normalized": false,
|
| 23 |
+
"rstrip": false,
|
| 24 |
+
"single_word": false,
|
| 25 |
+
"special": true
|
| 26 |
+
},
|
| 27 |
+
"3": {
|
| 28 |
+
"content": "[UNK]",
|
| 29 |
+
"lstrip": false,
|
| 30 |
+
"normalized": true,
|
| 31 |
+
"rstrip": false,
|
| 32 |
+
"single_word": false,
|
| 33 |
+
"special": true
|
| 34 |
+
},
|
| 35 |
+
"128000": {
|
| 36 |
+
"content": "[MASK]",
|
| 37 |
+
"lstrip": false,
|
| 38 |
+
"normalized": false,
|
| 39 |
+
"rstrip": false,
|
| 40 |
+
"single_word": false,
|
| 41 |
+
"special": true
|
| 42 |
+
},
|
| 43 |
+
"128001": {
|
| 44 |
+
"content": "[FLERT]",
|
| 45 |
+
"lstrip": false,
|
| 46 |
+
"normalized": true,
|
| 47 |
+
"rstrip": false,
|
| 48 |
+
"single_word": false,
|
| 49 |
+
"special": false
|
| 50 |
+
},
|
| 51 |
+
"128002": {
|
| 52 |
+
"content": "<<ENT>>",
|
| 53 |
+
"lstrip": false,
|
| 54 |
+
"normalized": true,
|
| 55 |
+
"rstrip": false,
|
| 56 |
+
"single_word": false,
|
| 57 |
+
"special": false
|
| 58 |
+
},
|
| 59 |
+
"128003": {
|
| 60 |
+
"content": "<<SEP>>",
|
| 61 |
+
"lstrip": false,
|
| 62 |
+
"normalized": true,
|
| 63 |
+
"rstrip": false,
|
| 64 |
+
"single_word": false,
|
| 65 |
+
"special": false
|
| 66 |
+
}
|
| 67 |
+
},
|
| 68 |
+
"bos_token": "[CLS]",
|
| 69 |
+
"clean_up_tokenization_spaces": false,
|
| 70 |
+
"cls_token": "[CLS]",
|
| 71 |
+
"do_lower_case": false,
|
| 72 |
+
"eos_token": "[SEP]",
|
| 73 |
+
"mask_token": "[MASK]",
|
| 74 |
+
"model_max_length": 1000000000000000019884624838656,
|
| 75 |
+
"pad_token": "[PAD]",
|
| 76 |
+
"sep_token": "[SEP]",
|
| 77 |
+
"sp_model_kwargs": {},
|
| 78 |
+
"split_by_punct": false,
|
| 79 |
+
"tokenizer_class": "DebertaV2Tokenizer",
|
| 80 |
+
"unk_token": "[UNK]",
|
| 81 |
+
"vocab_type": "spm"
|
| 82 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1433 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 2.2886961298148445,
|
| 5 |
+
"eval_steps": 500000,
|
| 6 |
+
"global_step": 100000,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.011443480649074222,
|
| 13 |
+
"grad_norm": 603.6091918945312,
|
| 14 |
+
"learning_rate": 1.1666666666666667e-07,
|
| 15 |
+
"loss": 10.731,
|
| 16 |
+
"step": 500
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.022886961298148444,
|
| 20 |
+
"grad_norm": 718.9038696289062,
|
| 21 |
+
"learning_rate": 2.3333333333333333e-07,
|
| 22 |
+
"loss": 9.2432,
|
| 23 |
+
"step": 1000
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.034330441947222665,
|
| 27 |
+
"grad_norm": 38.804962158203125,
|
| 28 |
+
"learning_rate": 3.5000000000000004e-07,
|
| 29 |
+
"loss": 9.1395,
|
| 30 |
+
"step": 1500
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.04577392259629689,
|
| 34 |
+
"grad_norm": 619.4507446289062,
|
| 35 |
+
"learning_rate": 4.6666666666666666e-07,
|
| 36 |
+
"loss": 11.2638,
|
| 37 |
+
"step": 2000
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.05721740324537111,
|
| 41 |
+
"grad_norm": 0.11708039790391922,
|
| 42 |
+
"learning_rate": 5.833333333333333e-07,
|
| 43 |
+
"loss": 9.1618,
|
| 44 |
+
"step": 2500
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.06866088389444533,
|
| 48 |
+
"grad_norm": 206.2231903076172,
|
| 49 |
+
"learning_rate": 7.000000000000001e-07,
|
| 50 |
+
"loss": 12.4357,
|
| 51 |
+
"step": 3000
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.08010436454351956,
|
| 55 |
+
"grad_norm": 26.35240936279297,
|
| 56 |
+
"learning_rate": 8.166666666666666e-07,
|
| 57 |
+
"loss": 9.4019,
|
| 58 |
+
"step": 3500
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.09154784519259378,
|
| 62 |
+
"grad_norm": 390.7612609863281,
|
| 63 |
+
"learning_rate": 9.333333333333333e-07,
|
| 64 |
+
"loss": 10.9883,
|
| 65 |
+
"step": 4000
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.10299132584166801,
|
| 69 |
+
"grad_norm": 560.8733520507812,
|
| 70 |
+
"learning_rate": 1.05e-06,
|
| 71 |
+
"loss": 11.4073,
|
| 72 |
+
"step": 4500
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.11443480649074222,
|
| 76 |
+
"grad_norm": 733.6063232421875,
|
| 77 |
+
"learning_rate": 1.1666666666666666e-06,
|
| 78 |
+
"loss": 9.4747,
|
| 79 |
+
"step": 5000
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.12587828713981644,
|
| 83 |
+
"grad_norm": 176.73806762695312,
|
| 84 |
+
"learning_rate": 1.2833333333333333e-06,
|
| 85 |
+
"loss": 11.2038,
|
| 86 |
+
"step": 5500
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.13732176778889066,
|
| 90 |
+
"grad_norm": 40.61717224121094,
|
| 91 |
+
"learning_rate": 1.4000000000000001e-06,
|
| 92 |
+
"loss": 11.3759,
|
| 93 |
+
"step": 6000
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.1487652484379649,
|
| 97 |
+
"grad_norm": 2191.57470703125,
|
| 98 |
+
"learning_rate": 1.5166666666666666e-06,
|
| 99 |
+
"loss": 10.7893,
|
| 100 |
+
"step": 6500
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.16020872908703912,
|
| 104 |
+
"grad_norm": 10.336786270141602,
|
| 105 |
+
"learning_rate": 1.6333333333333333e-06,
|
| 106 |
+
"loss": 11.6208,
|
| 107 |
+
"step": 7000
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.17165220973611334,
|
| 111 |
+
"grad_norm": 0.11516555398702621,
|
| 112 |
+
"learning_rate": 1.75e-06,
|
| 113 |
+
"loss": 10.8105,
|
| 114 |
+
"step": 7500
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.18309569038518755,
|
| 118 |
+
"grad_norm": 0.2883591651916504,
|
| 119 |
+
"learning_rate": 1.8666666666666667e-06,
|
| 120 |
+
"loss": 11.147,
|
| 121 |
+
"step": 8000
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.19453917103426177,
|
| 125 |
+
"grad_norm": 862.2401123046875,
|
| 126 |
+
"learning_rate": 1.983333333333333e-06,
|
| 127 |
+
"loss": 10.8087,
|
| 128 |
+
"step": 8500
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.20598265168333602,
|
| 132 |
+
"grad_norm": 0.004448544699698687,
|
| 133 |
+
"learning_rate": 2.1e-06,
|
| 134 |
+
"loss": 11.3934,
|
| 135 |
+
"step": 9000
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.21742613233241023,
|
| 139 |
+
"grad_norm": 0.008581358008086681,
|
| 140 |
+
"learning_rate": 2.2166666666666665e-06,
|
| 141 |
+
"loss": 13.5201,
|
| 142 |
+
"step": 9500
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.22886961298148445,
|
| 146 |
+
"grad_norm": 0.170095756649971,
|
| 147 |
+
"learning_rate": 2.333333333333333e-06,
|
| 148 |
+
"loss": 10.7812,
|
| 149 |
+
"step": 10000
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.24031309363055867,
|
| 153 |
+
"grad_norm": 0.8723887801170349,
|
| 154 |
+
"learning_rate": 2.45e-06,
|
| 155 |
+
"loss": 10.8569,
|
| 156 |
+
"step": 10500
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.2517565742796329,
|
| 160 |
+
"grad_norm": 1424.5357666015625,
|
| 161 |
+
"learning_rate": 2.5666666666666665e-06,
|
| 162 |
+
"loss": 13.1824,
|
| 163 |
+
"step": 11000
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.26320005492870713,
|
| 167 |
+
"grad_norm": 2.006991386413574,
|
| 168 |
+
"learning_rate": 2.6833333333333336e-06,
|
| 169 |
+
"loss": 12.9087,
|
| 170 |
+
"step": 11500
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.2746435355777813,
|
| 174 |
+
"grad_norm": 492.7056884765625,
|
| 175 |
+
"learning_rate": 2.8000000000000003e-06,
|
| 176 |
+
"loss": 11.9267,
|
| 177 |
+
"step": 12000
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.28608701622685556,
|
| 181 |
+
"grad_norm": 6.009059906005859,
|
| 182 |
+
"learning_rate": 2.9166666666666666e-06,
|
| 183 |
+
"loss": 12.8387,
|
| 184 |
+
"step": 12500
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.2975304968759298,
|
| 188 |
+
"grad_norm": 1738.22900390625,
|
| 189 |
+
"learning_rate": 3.0333333333333332e-06,
|
| 190 |
+
"loss": 13.1134,
|
| 191 |
+
"step": 13000
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.308973977525004,
|
| 195 |
+
"grad_norm": 1644.960205078125,
|
| 196 |
+
"learning_rate": 3.15e-06,
|
| 197 |
+
"loss": 10.5768,
|
| 198 |
+
"step": 13500
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.32041745817407824,
|
| 202 |
+
"grad_norm": 2049.489501953125,
|
| 203 |
+
"learning_rate": 3.2666666666666666e-06,
|
| 204 |
+
"loss": 12.9949,
|
| 205 |
+
"step": 14000
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.33186093882315243,
|
| 209 |
+
"grad_norm": 2.0113472938537598,
|
| 210 |
+
"learning_rate": 3.3833333333333333e-06,
|
| 211 |
+
"loss": 12.5091,
|
| 212 |
+
"step": 14500
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.3433044194722267,
|
| 216 |
+
"grad_norm": 767.9404296875,
|
| 217 |
+
"learning_rate": 3.5e-06,
|
| 218 |
+
"loss": 12.5176,
|
| 219 |
+
"step": 15000
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.3547479001213009,
|
| 223 |
+
"grad_norm": 342.45513916015625,
|
| 224 |
+
"learning_rate": 3.616666666666667e-06,
|
| 225 |
+
"loss": 12.6625,
|
| 226 |
+
"step": 15500
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.3661913807703751,
|
| 230 |
+
"grad_norm": 771.4798583984375,
|
| 231 |
+
"learning_rate": 3.7333333333333333e-06,
|
| 232 |
+
"loss": 14.2435,
|
| 233 |
+
"step": 16000
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.37763486141944935,
|
| 237 |
+
"grad_norm": 4853.9267578125,
|
| 238 |
+
"learning_rate": 3.85e-06,
|
| 239 |
+
"loss": 15.1053,
|
| 240 |
+
"step": 16500
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.38907834206852354,
|
| 244 |
+
"grad_norm": 0.0017908277222886682,
|
| 245 |
+
"learning_rate": 3.966666666666666e-06,
|
| 246 |
+
"loss": 12.6706,
|
| 247 |
+
"step": 17000
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.4005218227175978,
|
| 251 |
+
"grad_norm": 0.2733666002750397,
|
| 252 |
+
"learning_rate": 4.083333333333334e-06,
|
| 253 |
+
"loss": 12.4815,
|
| 254 |
+
"step": 17500
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.41196530336667203,
|
| 258 |
+
"grad_norm": 0.0,
|
| 259 |
+
"learning_rate": 4.2e-06,
|
| 260 |
+
"loss": 12.5318,
|
| 261 |
+
"step": 18000
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.4234087840157462,
|
| 265 |
+
"grad_norm": 1851.682861328125,
|
| 266 |
+
"learning_rate": 4.316666666666667e-06,
|
| 267 |
+
"loss": 13.7826,
|
| 268 |
+
"step": 18500
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.43485226466482046,
|
| 272 |
+
"grad_norm": 157.1131591796875,
|
| 273 |
+
"learning_rate": 4.433333333333333e-06,
|
| 274 |
+
"loss": 12.2737,
|
| 275 |
+
"step": 19000
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.44629574531389465,
|
| 279 |
+
"grad_norm": 1271.1109619140625,
|
| 280 |
+
"learning_rate": 4.5500000000000005e-06,
|
| 281 |
+
"loss": 10.5671,
|
| 282 |
+
"step": 19500
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.4577392259629689,
|
| 286 |
+
"grad_norm": 480.1098937988281,
|
| 287 |
+
"learning_rate": 4.666666666666666e-06,
|
| 288 |
+
"loss": 13.5673,
|
| 289 |
+
"step": 20000
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.46918270661204314,
|
| 293 |
+
"grad_norm": 1566.4591064453125,
|
| 294 |
+
"learning_rate": 4.783333333333333e-06,
|
| 295 |
+
"loss": 14.4705,
|
| 296 |
+
"step": 20500
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.48062618726111733,
|
| 300 |
+
"grad_norm": 0.0007110056467354298,
|
| 301 |
+
"learning_rate": 4.9e-06,
|
| 302 |
+
"loss": 11.6491,
|
| 303 |
+
"step": 21000
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.4920696679101916,
|
| 307 |
+
"grad_norm": 4.593759059906006,
|
| 308 |
+
"learning_rate": 5.016666666666666e-06,
|
| 309 |
+
"loss": 12.8944,
|
| 310 |
+
"step": 21500
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.5035131485592658,
|
| 314 |
+
"grad_norm": 0.0,
|
| 315 |
+
"learning_rate": 5.133333333333333e-06,
|
| 316 |
+
"loss": 13.6318,
|
| 317 |
+
"step": 22000
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.51495662920834,
|
| 321 |
+
"grad_norm": 132.595458984375,
|
| 322 |
+
"learning_rate": 5.25e-06,
|
| 323 |
+
"loss": 12.8382,
|
| 324 |
+
"step": 22500
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.5264001098574143,
|
| 328 |
+
"grad_norm": 347.5989074707031,
|
| 329 |
+
"learning_rate": 5.366666666666667e-06,
|
| 330 |
+
"loss": 11.8936,
|
| 331 |
+
"step": 23000
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.5378435905064884,
|
| 335 |
+
"grad_norm": 0.009264913387596607,
|
| 336 |
+
"learning_rate": 5.483333333333333e-06,
|
| 337 |
+
"loss": 14.0151,
|
| 338 |
+
"step": 23500
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.5492870711555626,
|
| 342 |
+
"grad_norm": 245.79705810546875,
|
| 343 |
+
"learning_rate": 5.600000000000001e-06,
|
| 344 |
+
"loss": 11.9611,
|
| 345 |
+
"step": 24000
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.5607305518046369,
|
| 349 |
+
"grad_norm": 1345.8895263671875,
|
| 350 |
+
"learning_rate": 5.7166666666666664e-06,
|
| 351 |
+
"loss": 13.2515,
|
| 352 |
+
"step": 24500
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 0.5721740324537111,
|
| 356 |
+
"grad_norm": 0.26148831844329834,
|
| 357 |
+
"learning_rate": 5.833333333333333e-06,
|
| 358 |
+
"loss": 13.5492,
|
| 359 |
+
"step": 25000
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 0.5836175131027853,
|
| 363 |
+
"grad_norm": 4638.9169921875,
|
| 364 |
+
"learning_rate": 5.95e-06,
|
| 365 |
+
"loss": 14.1596,
|
| 366 |
+
"step": 25500
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 0.5950609937518596,
|
| 370 |
+
"grad_norm": 184.8077392578125,
|
| 371 |
+
"learning_rate": 6.0666666666666665e-06,
|
| 372 |
+
"loss": 11.8104,
|
| 373 |
+
"step": 26000
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 0.6065044744009338,
|
| 377 |
+
"grad_norm": 140.49444580078125,
|
| 378 |
+
"learning_rate": 6.183333333333333e-06,
|
| 379 |
+
"loss": 11.2624,
|
| 380 |
+
"step": 26500
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 0.617947955050008,
|
| 384 |
+
"grad_norm": 0.0011962183052673936,
|
| 385 |
+
"learning_rate": 6.3e-06,
|
| 386 |
+
"loss": 12.7681,
|
| 387 |
+
"step": 27000
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 0.6293914356990822,
|
| 391 |
+
"grad_norm": 0.057224027812480927,
|
| 392 |
+
"learning_rate": 6.4166666666666665e-06,
|
| 393 |
+
"loss": 15.7726,
|
| 394 |
+
"step": 27500
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 0.6408349163481565,
|
| 398 |
+
"grad_norm": 0.04104612395167351,
|
| 399 |
+
"learning_rate": 6.533333333333333e-06,
|
| 400 |
+
"loss": 15.2471,
|
| 401 |
+
"step": 28000
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.6522783969972307,
|
| 405 |
+
"grad_norm": 0.003121469169855118,
|
| 406 |
+
"learning_rate": 6.65e-06,
|
| 407 |
+
"loss": 16.5911,
|
| 408 |
+
"step": 28500
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 0.6637218776463049,
|
| 412 |
+
"grad_norm": 0.0,
|
| 413 |
+
"learning_rate": 6.7666666666666665e-06,
|
| 414 |
+
"loss": 11.7935,
|
| 415 |
+
"step": 29000
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 0.6751653582953792,
|
| 419 |
+
"grad_norm": 1068.4923095703125,
|
| 420 |
+
"learning_rate": 6.883333333333333e-06,
|
| 421 |
+
"loss": 12.6445,
|
| 422 |
+
"step": 29500
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 0.6866088389444533,
|
| 426 |
+
"grad_norm": 49.82635498046875,
|
| 427 |
+
"learning_rate": 7e-06,
|
| 428 |
+
"loss": 13.4921,
|
| 429 |
+
"step": 30000
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.6980523195935275,
|
| 433 |
+
"grad_norm": 0.021331172436475754,
|
| 434 |
+
"learning_rate": 6.987037037037037e-06,
|
| 435 |
+
"loss": 13.8913,
|
| 436 |
+
"step": 30500
|
| 437 |
+
},
|
| 438 |
+
{
|
| 439 |
+
"epoch": 0.7094958002426018,
|
| 440 |
+
"grad_norm": 390.8102111816406,
|
| 441 |
+
"learning_rate": 6.974074074074074e-06,
|
| 442 |
+
"loss": 14.1043,
|
| 443 |
+
"step": 31000
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 0.720939280891676,
|
| 447 |
+
"grad_norm": 464.5948791503906,
|
| 448 |
+
"learning_rate": 6.9611111111111116e-06,
|
| 449 |
+
"loss": 14.1976,
|
| 450 |
+
"step": 31500
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 0.7323827615407502,
|
| 454 |
+
"grad_norm": 383.39373779296875,
|
| 455 |
+
"learning_rate": 6.948148148148148e-06,
|
| 456 |
+
"loss": 13.338,
|
| 457 |
+
"step": 32000
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 0.7438262421898244,
|
| 461 |
+
"grad_norm": 0.005027694161981344,
|
| 462 |
+
"learning_rate": 6.935185185185185e-06,
|
| 463 |
+
"loss": 13.6854,
|
| 464 |
+
"step": 32500
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 0.7552697228388987,
|
| 468 |
+
"grad_norm": 1620.7523193359375,
|
| 469 |
+
"learning_rate": 6.922222222222222e-06,
|
| 470 |
+
"loss": 11.7295,
|
| 471 |
+
"step": 33000
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 0.7667132034879729,
|
| 475 |
+
"grad_norm": 959.1514282226562,
|
| 476 |
+
"learning_rate": 6.90925925925926e-06,
|
| 477 |
+
"loss": 14.5347,
|
| 478 |
+
"step": 33500
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 0.7781566841370471,
|
| 482 |
+
"grad_norm": 778.1695556640625,
|
| 483 |
+
"learning_rate": 6.896296296296296e-06,
|
| 484 |
+
"loss": 12.9294,
|
| 485 |
+
"step": 34000
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 0.7896001647861214,
|
| 489 |
+
"grad_norm": 18.240928649902344,
|
| 490 |
+
"learning_rate": 6.883333333333333e-06,
|
| 491 |
+
"loss": 12.5578,
|
| 492 |
+
"step": 34500
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 0.8010436454351956,
|
| 496 |
+
"grad_norm": 0.060547858476638794,
|
| 497 |
+
"learning_rate": 6.8703703703703704e-06,
|
| 498 |
+
"loss": 11.7427,
|
| 499 |
+
"step": 35000
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 0.8124871260842698,
|
| 503 |
+
"grad_norm": 0.03452019393444061,
|
| 504 |
+
"learning_rate": 6.857407407407408e-06,
|
| 505 |
+
"loss": 13.3464,
|
| 506 |
+
"step": 35500
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 0.8239306067333441,
|
| 510 |
+
"grad_norm": 0.0,
|
| 511 |
+
"learning_rate": 6.844444444444444e-06,
|
| 512 |
+
"loss": 12.3813,
|
| 513 |
+
"step": 36000
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 0.8353740873824183,
|
| 517 |
+
"grad_norm": 0.01644955575466156,
|
| 518 |
+
"learning_rate": 6.831481481481482e-06,
|
| 519 |
+
"loss": 15.1999,
|
| 520 |
+
"step": 36500
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 0.8468175680314924,
|
| 524 |
+
"grad_norm": 0.041852571070194244,
|
| 525 |
+
"learning_rate": 6.8185185185185185e-06,
|
| 526 |
+
"loss": 14.611,
|
| 527 |
+
"step": 37000
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 0.8582610486805666,
|
| 531 |
+
"grad_norm": 0.03489886596798897,
|
| 532 |
+
"learning_rate": 6.805555555555556e-06,
|
| 533 |
+
"loss": 12.7151,
|
| 534 |
+
"step": 37500
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 0.8697045293296409,
|
| 538 |
+
"grad_norm": 1784.8714599609375,
|
| 539 |
+
"learning_rate": 6.792592592592592e-06,
|
| 540 |
+
"loss": 13.324,
|
| 541 |
+
"step": 38000
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 0.8811480099787151,
|
| 545 |
+
"grad_norm": 325.86151123046875,
|
| 546 |
+
"learning_rate": 6.77962962962963e-06,
|
| 547 |
+
"loss": 13.8547,
|
| 548 |
+
"step": 38500
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 0.8925914906277893,
|
| 552 |
+
"grad_norm": 209.41114807128906,
|
| 553 |
+
"learning_rate": 6.7666666666666665e-06,
|
| 554 |
+
"loss": 13.9472,
|
| 555 |
+
"step": 39000
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 0.9040349712768636,
|
| 559 |
+
"grad_norm": 0.0,
|
| 560 |
+
"learning_rate": 6.753703703703704e-06,
|
| 561 |
+
"loss": 14.3618,
|
| 562 |
+
"step": 39500
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 0.9154784519259378,
|
| 566 |
+
"grad_norm": 1665.116943359375,
|
| 567 |
+
"learning_rate": 6.74074074074074e-06,
|
| 568 |
+
"loss": 15.8669,
|
| 569 |
+
"step": 40000
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 0.926921932575012,
|
| 573 |
+
"grad_norm": 0.0066319056786596775,
|
| 574 |
+
"learning_rate": 6.727777777777778e-06,
|
| 575 |
+
"loss": 14.8973,
|
| 576 |
+
"step": 40500
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 0.9383654132240863,
|
| 580 |
+
"grad_norm": 738.2325439453125,
|
| 581 |
+
"learning_rate": 6.714814814814815e-06,
|
| 582 |
+
"loss": 13.7844,
|
| 583 |
+
"step": 41000
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 0.9498088938731605,
|
| 587 |
+
"grad_norm": 276.5779113769531,
|
| 588 |
+
"learning_rate": 6.701851851851852e-06,
|
| 589 |
+
"loss": 12.9653,
|
| 590 |
+
"step": 41500
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 0.9612523745222347,
|
| 594 |
+
"grad_norm": 1572.8726806640625,
|
| 595 |
+
"learning_rate": 6.688888888888889e-06,
|
| 596 |
+
"loss": 12.8457,
|
| 597 |
+
"step": 42000
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 0.9726958551713089,
|
| 601 |
+
"grad_norm": 1.0475573539733887,
|
| 602 |
+
"learning_rate": 6.675925925925926e-06,
|
| 603 |
+
"loss": 12.2727,
|
| 604 |
+
"step": 42500
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 0.9841393358203832,
|
| 608 |
+
"grad_norm": 12.392683029174805,
|
| 609 |
+
"learning_rate": 6.662962962962963e-06,
|
| 610 |
+
"loss": 14.6495,
|
| 611 |
+
"step": 43000
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 0.9955828164694573,
|
| 615 |
+
"grad_norm": 0.0,
|
| 616 |
+
"learning_rate": 6.65e-06,
|
| 617 |
+
"loss": 13.6766,
|
| 618 |
+
"step": 43500
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 1.0070262971185315,
|
| 622 |
+
"grad_norm": 178.05191040039062,
|
| 623 |
+
"learning_rate": 6.637037037037037e-06,
|
| 624 |
+
"loss": 12.0881,
|
| 625 |
+
"step": 44000
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 1.0184697777676057,
|
| 629 |
+
"grad_norm": 287.11749267578125,
|
| 630 |
+
"learning_rate": 6.624074074074074e-06,
|
| 631 |
+
"loss": 10.3373,
|
| 632 |
+
"step": 44500
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 1.02991325841668,
|
| 636 |
+
"grad_norm": 229.68655395507812,
|
| 637 |
+
"learning_rate": 6.611111111111111e-06,
|
| 638 |
+
"loss": 12.0917,
|
| 639 |
+
"step": 45000
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 1.0413567390657543,
|
| 643 |
+
"grad_norm": 0.005943186115473509,
|
| 644 |
+
"learning_rate": 6.598148148148148e-06,
|
| 645 |
+
"loss": 11.5574,
|
| 646 |
+
"step": 45500
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"epoch": 1.0528002197148285,
|
| 650 |
+
"grad_norm": 0.00034078938188031316,
|
| 651 |
+
"learning_rate": 6.585185185185185e-06,
|
| 652 |
+
"loss": 13.1375,
|
| 653 |
+
"step": 46000
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 1.0642437003639027,
|
| 657 |
+
"grad_norm": 0.3885180354118347,
|
| 658 |
+
"learning_rate": 6.572222222222222e-06,
|
| 659 |
+
"loss": 11.2181,
|
| 660 |
+
"step": 46500
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 1.075687181012977,
|
| 664 |
+
"grad_norm": 1127.3404541015625,
|
| 665 |
+
"learning_rate": 6.559259259259259e-06,
|
| 666 |
+
"loss": 12.5768,
|
| 667 |
+
"step": 47000
|
| 668 |
+
},
|
| 669 |
+
{
|
| 670 |
+
"epoch": 1.087130661662051,
|
| 671 |
+
"grad_norm": 1220.1749267578125,
|
| 672 |
+
"learning_rate": 6.546296296296297e-06,
|
| 673 |
+
"loss": 13.4144,
|
| 674 |
+
"step": 47500
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 1.0985741423111253,
|
| 678 |
+
"grad_norm": 0.0005302530480548739,
|
| 679 |
+
"learning_rate": 6.533333333333333e-06,
|
| 680 |
+
"loss": 11.0923,
|
| 681 |
+
"step": 48000
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 1.1100176229601995,
|
| 685 |
+
"grad_norm": 661.6603393554688,
|
| 686 |
+
"learning_rate": 6.52037037037037e-06,
|
| 687 |
+
"loss": 11.0951,
|
| 688 |
+
"step": 48500
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"epoch": 1.1214611036092739,
|
| 692 |
+
"grad_norm": 0.08705360442399979,
|
| 693 |
+
"learning_rate": 6.507407407407407e-06,
|
| 694 |
+
"loss": 11.5883,
|
| 695 |
+
"step": 49000
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 1.132904584258348,
|
| 699 |
+
"grad_norm": 171.4398651123047,
|
| 700 |
+
"learning_rate": 6.494444444444445e-06,
|
| 701 |
+
"loss": 13.5062,
|
| 702 |
+
"step": 49500
|
| 703 |
+
},
|
| 704 |
+
{
|
| 705 |
+
"epoch": 1.1443480649074222,
|
| 706 |
+
"grad_norm": 0.04586634412407875,
|
| 707 |
+
"learning_rate": 6.481481481481481e-06,
|
| 708 |
+
"loss": 12.4203,
|
| 709 |
+
"step": 50000
|
| 710 |
+
},
|
| 711 |
+
{
|
| 712 |
+
"epoch": 1.1557915455564964,
|
| 713 |
+
"grad_norm": 0.0418708510696888,
|
| 714 |
+
"learning_rate": 6.4685185185185185e-06,
|
| 715 |
+
"loss": 12.8418,
|
| 716 |
+
"step": 50500
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 1.1672350262055706,
|
| 720 |
+
"grad_norm": 635.2694091796875,
|
| 721 |
+
"learning_rate": 6.455555555555556e-06,
|
| 722 |
+
"loss": 11.7328,
|
| 723 |
+
"step": 51000
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 1.1786785068546448,
|
| 727 |
+
"grad_norm": 216.6066131591797,
|
| 728 |
+
"learning_rate": 6.442592592592593e-06,
|
| 729 |
+
"loss": 13.6954,
|
| 730 |
+
"step": 51500
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 1.1901219875037192,
|
| 734 |
+
"grad_norm": 120.59862518310547,
|
| 735 |
+
"learning_rate": 6.429629629629629e-06,
|
| 736 |
+
"loss": 11.0234,
|
| 737 |
+
"step": 52000
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 1.2015654681527934,
|
| 741 |
+
"grad_norm": 647.171875,
|
| 742 |
+
"learning_rate": 6.4166666666666665e-06,
|
| 743 |
+
"loss": 14.4181,
|
| 744 |
+
"step": 52500
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 1.2130089488018676,
|
| 748 |
+
"grad_norm": 0.8411855697631836,
|
| 749 |
+
"learning_rate": 6.403703703703704e-06,
|
| 750 |
+
"loss": 13.5984,
|
| 751 |
+
"step": 53000
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 1.2244524294509418,
|
| 755 |
+
"grad_norm": 0.0012393624056130648,
|
| 756 |
+
"learning_rate": 6.390740740740741e-06,
|
| 757 |
+
"loss": 12.1035,
|
| 758 |
+
"step": 53500
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 1.235895910100016,
|
| 762 |
+
"grad_norm": 0.0,
|
| 763 |
+
"learning_rate": 6.377777777777777e-06,
|
| 764 |
+
"loss": 12.8851,
|
| 765 |
+
"step": 54000
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 1.2473393907490902,
|
| 769 |
+
"grad_norm": 0.0,
|
| 770 |
+
"learning_rate": 6.3648148148148145e-06,
|
| 771 |
+
"loss": 15.1293,
|
| 772 |
+
"step": 54500
|
| 773 |
+
},
|
| 774 |
+
{
|
| 775 |
+
"epoch": 1.2587828713981644,
|
| 776 |
+
"grad_norm": 2496.68505859375,
|
| 777 |
+
"learning_rate": 6.351851851851852e-06,
|
| 778 |
+
"loss": 11.7267,
|
| 779 |
+
"step": 55000
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 1.2702263520472385,
|
| 783 |
+
"grad_norm": 603.0614013671875,
|
| 784 |
+
"learning_rate": 6.338888888888889e-06,
|
| 785 |
+
"loss": 12.6656,
|
| 786 |
+
"step": 55500
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 1.281669832696313,
|
| 790 |
+
"grad_norm": 0.1624222695827484,
|
| 791 |
+
"learning_rate": 6.325925925925925e-06,
|
| 792 |
+
"loss": 12.3149,
|
| 793 |
+
"step": 56000
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 1.2931133133453871,
|
| 797 |
+
"grad_norm": 0.0846194475889206,
|
| 798 |
+
"learning_rate": 6.3129629629629634e-06,
|
| 799 |
+
"loss": 14.1863,
|
| 800 |
+
"step": 56500
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 1.3045567939944613,
|
| 804 |
+
"grad_norm": 0.0,
|
| 805 |
+
"learning_rate": 6.3e-06,
|
| 806 |
+
"loss": 13.3858,
|
| 807 |
+
"step": 57000
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 1.3160002746435355,
|
| 811 |
+
"grad_norm": 0.0017423485405743122,
|
| 812 |
+
"learning_rate": 6.287037037037037e-06,
|
| 813 |
+
"loss": 10.2554,
|
| 814 |
+
"step": 57500
|
| 815 |
+
},
|
| 816 |
+
{
|
| 817 |
+
"epoch": 1.3274437552926097,
|
| 818 |
+
"grad_norm": 1450.389892578125,
|
| 819 |
+
"learning_rate": 6.274074074074073e-06,
|
| 820 |
+
"loss": 12.7777,
|
| 821 |
+
"step": 58000
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 1.3388872359416841,
|
| 825 |
+
"grad_norm": 0.043471284210681915,
|
| 826 |
+
"learning_rate": 6.2611111111111115e-06,
|
| 827 |
+
"loss": 11.1225,
|
| 828 |
+
"step": 58500
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"epoch": 1.3503307165907583,
|
| 832 |
+
"grad_norm": 0.0038315041456371546,
|
| 833 |
+
"learning_rate": 6.248148148148148e-06,
|
| 834 |
+
"loss": 10.2773,
|
| 835 |
+
"step": 59000
|
| 836 |
+
},
|
| 837 |
+
{
|
| 838 |
+
"epoch": 1.3617741972398325,
|
| 839 |
+
"grad_norm": 2073.392578125,
|
| 840 |
+
"learning_rate": 6.235185185185185e-06,
|
| 841 |
+
"loss": 13.5407,
|
| 842 |
+
"step": 59500
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 1.3732176778889067,
|
| 846 |
+
"grad_norm": 4.363699913024902,
|
| 847 |
+
"learning_rate": 6.2222222222222215e-06,
|
| 848 |
+
"loss": 14.8187,
|
| 849 |
+
"step": 60000
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 1.3846611585379809,
|
| 853 |
+
"grad_norm": 0.010874895378947258,
|
| 854 |
+
"learning_rate": 6.2092592592592595e-06,
|
| 855 |
+
"loss": 11.7766,
|
| 856 |
+
"step": 60500
|
| 857 |
+
},
|
| 858 |
+
{
|
| 859 |
+
"epoch": 1.396104639187055,
|
| 860 |
+
"grad_norm": 5.799459457397461,
|
| 861 |
+
"learning_rate": 6.196296296296296e-06,
|
| 862 |
+
"loss": 12.6413,
|
| 863 |
+
"step": 61000
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 1.4075481198361293,
|
| 867 |
+
"grad_norm": 0.0036441339179873466,
|
| 868 |
+
"learning_rate": 6.183333333333333e-06,
|
| 869 |
+
"loss": 11.2557,
|
| 870 |
+
"step": 61500
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"epoch": 1.4189916004852035,
|
| 874 |
+
"grad_norm": 1275.374755859375,
|
| 875 |
+
"learning_rate": 6.17037037037037e-06,
|
| 876 |
+
"loss": 11.7272,
|
| 877 |
+
"step": 62000
|
| 878 |
+
},
|
| 879 |
+
{
|
| 880 |
+
"epoch": 1.4304350811342779,
|
| 881 |
+
"grad_norm": 4855.68212890625,
|
| 882 |
+
"learning_rate": 6.157407407407408e-06,
|
| 883 |
+
"loss": 12.1518,
|
| 884 |
+
"step": 62500
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 1.441878561783352,
|
| 888 |
+
"grad_norm": 523.7196655273438,
|
| 889 |
+
"learning_rate": 6.144444444444444e-06,
|
| 890 |
+
"loss": 12.6978,
|
| 891 |
+
"step": 63000
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 1.4533220424324262,
|
| 895 |
+
"grad_norm": 1098.2850341796875,
|
| 896 |
+
"learning_rate": 6.131481481481481e-06,
|
| 897 |
+
"loss": 11.8028,
|
| 898 |
+
"step": 63500
|
| 899 |
+
},
|
| 900 |
+
{
|
| 901 |
+
"epoch": 1.4647655230815004,
|
| 902 |
+
"grad_norm": 1153.9100341796875,
|
| 903 |
+
"learning_rate": 6.118518518518518e-06,
|
| 904 |
+
"loss": 11.3937,
|
| 905 |
+
"step": 64000
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 1.4762090037305746,
|
| 909 |
+
"grad_norm": 1573.587890625,
|
| 910 |
+
"learning_rate": 6.105555555555556e-06,
|
| 911 |
+
"loss": 13.9497,
|
| 912 |
+
"step": 64500
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 1.487652484379649,
|
| 916 |
+
"grad_norm": 908.9359130859375,
|
| 917 |
+
"learning_rate": 6.092592592592592e-06,
|
| 918 |
+
"loss": 11.6579,
|
| 919 |
+
"step": 65000
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 1.4990959650287232,
|
| 923 |
+
"grad_norm": 1250.6552734375,
|
| 924 |
+
"learning_rate": 6.079629629629629e-06,
|
| 925 |
+
"loss": 12.2497,
|
| 926 |
+
"step": 65500
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 1.5105394456777974,
|
| 930 |
+
"grad_norm": 0.030488021671772003,
|
| 931 |
+
"learning_rate": 6.0666666666666665e-06,
|
| 932 |
+
"loss": 13.4229,
|
| 933 |
+
"step": 66000
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 1.5219829263268716,
|
| 937 |
+
"grad_norm": 3663.417724609375,
|
| 938 |
+
"learning_rate": 6.053703703703704e-06,
|
| 939 |
+
"loss": 11.9166,
|
| 940 |
+
"step": 66500
|
| 941 |
+
},
|
| 942 |
+
{
|
| 943 |
+
"epoch": 1.5334264069759458,
|
| 944 |
+
"grad_norm": 0.10368392616510391,
|
| 945 |
+
"learning_rate": 6.04074074074074e-06,
|
| 946 |
+
"loss": 11.7133,
|
| 947 |
+
"step": 67000
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 1.54486988762502,
|
| 951 |
+
"grad_norm": 23.950149536132812,
|
| 952 |
+
"learning_rate": 6.027777777777778e-06,
|
| 953 |
+
"loss": 10.2979,
|
| 954 |
+
"step": 67500
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 1.5563133682740942,
|
| 958 |
+
"grad_norm": 1314.4285888671875,
|
| 959 |
+
"learning_rate": 6.0148148148148145e-06,
|
| 960 |
+
"loss": 11.5517,
|
| 961 |
+
"step": 68000
|
| 962 |
+
},
|
| 963 |
+
{
|
| 964 |
+
"epoch": 1.5677568489231684,
|
| 965 |
+
"grad_norm": 0.0,
|
| 966 |
+
"learning_rate": 6.001851851851852e-06,
|
| 967 |
+
"loss": 12.2986,
|
| 968 |
+
"step": 68500
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 1.5792003295722425,
|
| 972 |
+
"grad_norm": 558.1203002929688,
|
| 973 |
+
"learning_rate": 5.988888888888888e-06,
|
| 974 |
+
"loss": 11.5298,
|
| 975 |
+
"step": 69000
|
| 976 |
+
},
|
| 977 |
+
{
|
| 978 |
+
"epoch": 1.590643810221317,
|
| 979 |
+
"grad_norm": 0.5483641624450684,
|
| 980 |
+
"learning_rate": 5.975925925925926e-06,
|
| 981 |
+
"loss": 11.2464,
|
| 982 |
+
"step": 69500
|
| 983 |
+
},
|
| 984 |
+
{
|
| 985 |
+
"epoch": 1.6020872908703911,
|
| 986 |
+
"grad_norm": 0.009755097329616547,
|
| 987 |
+
"learning_rate": 5.9629629629629626e-06,
|
| 988 |
+
"loss": 13.0815,
|
| 989 |
+
"step": 70000
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 1.6135307715194653,
|
| 993 |
+
"grad_norm": 0.007829127833247185,
|
| 994 |
+
"learning_rate": 5.95e-06,
|
| 995 |
+
"loss": 10.083,
|
| 996 |
+
"step": 70500
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 1.6249742521685397,
|
| 1000 |
+
"grad_norm": 435.7135314941406,
|
| 1001 |
+
"learning_rate": 5.937037037037036e-06,
|
| 1002 |
+
"loss": 11.4477,
|
| 1003 |
+
"step": 71000
|
| 1004 |
+
},
|
| 1005 |
+
{
|
| 1006 |
+
"epoch": 1.636417732817614,
|
| 1007 |
+
"grad_norm": 5.702239513397217,
|
| 1008 |
+
"learning_rate": 5.924074074074074e-06,
|
| 1009 |
+
"loss": 11.0763,
|
| 1010 |
+
"step": 71500
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 1.6478612134666881,
|
| 1014 |
+
"grad_norm": 0.10179603844881058,
|
| 1015 |
+
"learning_rate": 5.911111111111111e-06,
|
| 1016 |
+
"loss": 12.8192,
|
| 1017 |
+
"step": 72000
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 1.6593046941157623,
|
| 1021 |
+
"grad_norm": 681.2998046875,
|
| 1022 |
+
"learning_rate": 5.898148148148148e-06,
|
| 1023 |
+
"loss": 11.8389,
|
| 1024 |
+
"step": 72500
|
| 1025 |
+
},
|
| 1026 |
+
{
|
| 1027 |
+
"epoch": 1.6707481747648365,
|
| 1028 |
+
"grad_norm": 168.10958862304688,
|
| 1029 |
+
"learning_rate": 5.885185185185185e-06,
|
| 1030 |
+
"loss": 9.9504,
|
| 1031 |
+
"step": 73000
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 1.6821916554139107,
|
| 1035 |
+
"grad_norm": 0.004441286437213421,
|
| 1036 |
+
"learning_rate": 5.872222222222222e-06,
|
| 1037 |
+
"loss": 14.6829,
|
| 1038 |
+
"step": 73500
|
| 1039 |
+
},
|
| 1040 |
+
{
|
| 1041 |
+
"epoch": 1.6936351360629849,
|
| 1042 |
+
"grad_norm": 0.020898401737213135,
|
| 1043 |
+
"learning_rate": 5.859259259259259e-06,
|
| 1044 |
+
"loss": 11.1156,
|
| 1045 |
+
"step": 74000
|
| 1046 |
+
},
|
| 1047 |
+
{
|
| 1048 |
+
"epoch": 1.705078616712059,
|
| 1049 |
+
"grad_norm": 0.021843064576387405,
|
| 1050 |
+
"learning_rate": 5.846296296296296e-06,
|
| 1051 |
+
"loss": 10.6854,
|
| 1052 |
+
"step": 74500
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 1.7165220973611333,
|
| 1056 |
+
"grad_norm": 33.806251525878906,
|
| 1057 |
+
"learning_rate": 5.833333333333333e-06,
|
| 1058 |
+
"loss": 13.1055,
|
| 1059 |
+
"step": 75000
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 1.7279655780102074,
|
| 1063 |
+
"grad_norm": 29.790348052978516,
|
| 1064 |
+
"learning_rate": 5.82037037037037e-06,
|
| 1065 |
+
"loss": 12.1967,
|
| 1066 |
+
"step": 75500
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 1.7394090586592819,
|
| 1070 |
+
"grad_norm": 2041.1212158203125,
|
| 1071 |
+
"learning_rate": 5.8074074074074076e-06,
|
| 1072 |
+
"loss": 12.7192,
|
| 1073 |
+
"step": 76000
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 1.750852539308356,
|
| 1077 |
+
"grad_norm": 1129.54833984375,
|
| 1078 |
+
"learning_rate": 5.794444444444444e-06,
|
| 1079 |
+
"loss": 12.4452,
|
| 1080 |
+
"step": 76500
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 1.7622960199574302,
|
| 1084 |
+
"grad_norm": 0.3305358290672302,
|
| 1085 |
+
"learning_rate": 5.781481481481482e-06,
|
| 1086 |
+
"loss": 12.6818,
|
| 1087 |
+
"step": 77000
|
| 1088 |
+
},
|
| 1089 |
+
{
|
| 1090 |
+
"epoch": 1.7737395006065044,
|
| 1091 |
+
"grad_norm": 5318.0498046875,
|
| 1092 |
+
"learning_rate": 5.768518518518518e-06,
|
| 1093 |
+
"loss": 11.3843,
|
| 1094 |
+
"step": 77500
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 1.7851829812555788,
|
| 1098 |
+
"grad_norm": 1354.53466796875,
|
| 1099 |
+
"learning_rate": 5.755555555555556e-06,
|
| 1100 |
+
"loss": 11.9361,
|
| 1101 |
+
"step": 78000
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 1.796626461904653,
|
| 1105 |
+
"grad_norm": 0.01847067102789879,
|
| 1106 |
+
"learning_rate": 5.742592592592593e-06,
|
| 1107 |
+
"loss": 14.4414,
|
| 1108 |
+
"step": 78500
|
| 1109 |
+
},
|
| 1110 |
+
{
|
| 1111 |
+
"epoch": 1.8080699425537272,
|
| 1112 |
+
"grad_norm": 0.0,
|
| 1113 |
+
"learning_rate": 5.72962962962963e-06,
|
| 1114 |
+
"loss": 12.2809,
|
| 1115 |
+
"step": 79000
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 1.8195134232028014,
|
| 1119 |
+
"grad_norm": 0.0010120772058144212,
|
| 1120 |
+
"learning_rate": 5.7166666666666664e-06,
|
| 1121 |
+
"loss": 11.9991,
|
| 1122 |
+
"step": 79500
|
| 1123 |
+
},
|
| 1124 |
+
{
|
| 1125 |
+
"epoch": 1.8309569038518756,
|
| 1126 |
+
"grad_norm": 0.0,
|
| 1127 |
+
"learning_rate": 5.703703703703704e-06,
|
| 1128 |
+
"loss": 10.9662,
|
| 1129 |
+
"step": 80000
|
| 1130 |
+
},
|
| 1131 |
+
{
|
| 1132 |
+
"epoch": 1.8424003845009498,
|
| 1133 |
+
"grad_norm": 279.47137451171875,
|
| 1134 |
+
"learning_rate": 5.690740740740741e-06,
|
| 1135 |
+
"loss": 11.1481,
|
| 1136 |
+
"step": 80500
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 1.853843865150024,
|
| 1140 |
+
"grad_norm": 0.0,
|
| 1141 |
+
"learning_rate": 5.677777777777778e-06,
|
| 1142 |
+
"loss": 12.123,
|
| 1143 |
+
"step": 81000
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 1.8652873457990982,
|
| 1147 |
+
"grad_norm": 807.4469604492188,
|
| 1148 |
+
"learning_rate": 5.6648148148148145e-06,
|
| 1149 |
+
"loss": 13.1075,
|
| 1150 |
+
"step": 81500
|
| 1151 |
+
},
|
| 1152 |
+
{
|
| 1153 |
+
"epoch": 1.8767308264481724,
|
| 1154 |
+
"grad_norm": 173.2933349609375,
|
| 1155 |
+
"learning_rate": 5.6518518518518525e-06,
|
| 1156 |
+
"loss": 12.5405,
|
| 1157 |
+
"step": 82000
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 1.8881743070972465,
|
| 1161 |
+
"grad_norm": 0.0,
|
| 1162 |
+
"learning_rate": 5.638888888888889e-06,
|
| 1163 |
+
"loss": 13.9821,
|
| 1164 |
+
"step": 82500
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"epoch": 1.899617787746321,
|
| 1168 |
+
"grad_norm": 2014.615478515625,
|
| 1169 |
+
"learning_rate": 5.625925925925926e-06,
|
| 1170 |
+
"loss": 9.4421,
|
| 1171 |
+
"step": 83000
|
| 1172 |
+
},
|
| 1173 |
+
{
|
| 1174 |
+
"epoch": 1.9110612683953951,
|
| 1175 |
+
"grad_norm": 0.0,
|
| 1176 |
+
"learning_rate": 5.6129629629629625e-06,
|
| 1177 |
+
"loss": 10.8762,
|
| 1178 |
+
"step": 83500
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 1.9225047490444693,
|
| 1182 |
+
"grad_norm": 70.37690734863281,
|
| 1183 |
+
"learning_rate": 5.600000000000001e-06,
|
| 1184 |
+
"loss": 12.3405,
|
| 1185 |
+
"step": 84000
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 1.9339482296935437,
|
| 1189 |
+
"grad_norm": 0.5092810392379761,
|
| 1190 |
+
"learning_rate": 5.587037037037037e-06,
|
| 1191 |
+
"loss": 10.5346,
|
| 1192 |
+
"step": 84500
|
| 1193 |
+
},
|
| 1194 |
+
{
|
| 1195 |
+
"epoch": 1.945391710342618,
|
| 1196 |
+
"grad_norm": 1005.1751098632812,
|
| 1197 |
+
"learning_rate": 5.574074074074074e-06,
|
| 1198 |
+
"loss": 13.8286,
|
| 1199 |
+
"step": 85000
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 1.9568351909916921,
|
| 1203 |
+
"grad_norm": 0.003633589716628194,
|
| 1204 |
+
"learning_rate": 5.5611111111111106e-06,
|
| 1205 |
+
"loss": 12.5279,
|
| 1206 |
+
"step": 85500
|
| 1207 |
+
},
|
| 1208 |
+
{
|
| 1209 |
+
"epoch": 1.9682786716407663,
|
| 1210 |
+
"grad_norm": 398.6470642089844,
|
| 1211 |
+
"learning_rate": 5.548148148148149e-06,
|
| 1212 |
+
"loss": 10.5745,
|
| 1213 |
+
"step": 86000
|
| 1214 |
+
},
|
| 1215 |
+
{
|
| 1216 |
+
"epoch": 1.9797221522898405,
|
| 1217 |
+
"grad_norm": 538.2572021484375,
|
| 1218 |
+
"learning_rate": 5.535185185185185e-06,
|
| 1219 |
+
"loss": 14.3954,
|
| 1220 |
+
"step": 86500
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 1.9911656329389147,
|
| 1224 |
+
"grad_norm": 656.5657348632812,
|
| 1225 |
+
"learning_rate": 5.522222222222222e-06,
|
| 1226 |
+
"loss": 13.2243,
|
| 1227 |
+
"step": 87000
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 2.002609113587989,
|
| 1231 |
+
"grad_norm": 195.96044921875,
|
| 1232 |
+
"learning_rate": 5.5092592592592595e-06,
|
| 1233 |
+
"loss": 13.1255,
|
| 1234 |
+
"step": 87500
|
| 1235 |
+
},
|
| 1236 |
+
{
|
| 1237 |
+
"epoch": 2.014052594237063,
|
| 1238 |
+
"grad_norm": 1494.8990478515625,
|
| 1239 |
+
"learning_rate": 5.496296296296297e-06,
|
| 1240 |
+
"loss": 8.6693,
|
| 1241 |
+
"step": 88000
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 2.0254960748861373,
|
| 1245 |
+
"grad_norm": 0.0,
|
| 1246 |
+
"learning_rate": 5.483333333333333e-06,
|
| 1247 |
+
"loss": 10.1512,
|
| 1248 |
+
"step": 88500
|
| 1249 |
+
},
|
| 1250 |
+
{
|
| 1251 |
+
"epoch": 2.0369395555352114,
|
| 1252 |
+
"grad_norm": 3263.656982421875,
|
| 1253 |
+
"learning_rate": 5.47037037037037e-06,
|
| 1254 |
+
"loss": 11.7968,
|
| 1255 |
+
"step": 89000
|
| 1256 |
+
},
|
| 1257 |
+
{
|
| 1258 |
+
"epoch": 2.0483830361842856,
|
| 1259 |
+
"grad_norm": 0.5306602716445923,
|
| 1260 |
+
"learning_rate": 5.4574074074074075e-06,
|
| 1261 |
+
"loss": 10.1574,
|
| 1262 |
+
"step": 89500
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 2.05982651683336,
|
| 1266 |
+
"grad_norm": 955.903076171875,
|
| 1267 |
+
"learning_rate": 5.444444444444445e-06,
|
| 1268 |
+
"loss": 8.7719,
|
| 1269 |
+
"step": 90000
|
| 1270 |
+
},
|
| 1271 |
+
{
|
| 1272 |
+
"epoch": 2.0712699974824345,
|
| 1273 |
+
"grad_norm": 28.916826248168945,
|
| 1274 |
+
"learning_rate": 5.431481481481481e-06,
|
| 1275 |
+
"loss": 7.6421,
|
| 1276 |
+
"step": 90500
|
| 1277 |
+
},
|
| 1278 |
+
{
|
| 1279 |
+
"epoch": 2.0827134781315086,
|
| 1280 |
+
"grad_norm": 0.018332993611693382,
|
| 1281 |
+
"learning_rate": 5.418518518518518e-06,
|
| 1282 |
+
"loss": 9.0135,
|
| 1283 |
+
"step": 91000
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 2.094156958780583,
|
| 1287 |
+
"grad_norm": 240.7179412841797,
|
| 1288 |
+
"learning_rate": 5.4055555555555556e-06,
|
| 1289 |
+
"loss": 8.8454,
|
| 1290 |
+
"step": 91500
|
| 1291 |
+
},
|
| 1292 |
+
{
|
| 1293 |
+
"epoch": 2.105600439429657,
|
| 1294 |
+
"grad_norm": 237.4253692626953,
|
| 1295 |
+
"learning_rate": 5.392592592592593e-06,
|
| 1296 |
+
"loss": 9.7903,
|
| 1297 |
+
"step": 92000
|
| 1298 |
+
},
|
| 1299 |
+
{
|
| 1300 |
+
"epoch": 2.117043920078731,
|
| 1301 |
+
"grad_norm": 2.053232192993164,
|
| 1302 |
+
"learning_rate": 5.379629629629629e-06,
|
| 1303 |
+
"loss": 11.0608,
|
| 1304 |
+
"step": 92500
|
| 1305 |
+
},
|
| 1306 |
+
{
|
| 1307 |
+
"epoch": 2.1284874007278054,
|
| 1308 |
+
"grad_norm": 862.31201171875,
|
| 1309 |
+
"learning_rate": 5.366666666666667e-06,
|
| 1310 |
+
"loss": 10.3125,
|
| 1311 |
+
"step": 93000
|
| 1312 |
+
},
|
| 1313 |
+
{
|
| 1314 |
+
"epoch": 2.1399308813768796,
|
| 1315 |
+
"grad_norm": 0.03503242880105972,
|
| 1316 |
+
"learning_rate": 5.353703703703704e-06,
|
| 1317 |
+
"loss": 12.3845,
|
| 1318 |
+
"step": 93500
|
| 1319 |
+
},
|
| 1320 |
+
{
|
| 1321 |
+
"epoch": 2.151374362025954,
|
| 1322 |
+
"grad_norm": 0.035824213176965714,
|
| 1323 |
+
"learning_rate": 5.340740740740741e-06,
|
| 1324 |
+
"loss": 9.1687,
|
| 1325 |
+
"step": 94000
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 2.162817842675028,
|
| 1329 |
+
"grad_norm": 0.0,
|
| 1330 |
+
"learning_rate": 5.327777777777777e-06,
|
| 1331 |
+
"loss": 9.9659,
|
| 1332 |
+
"step": 94500
|
| 1333 |
+
},
|
| 1334 |
+
{
|
| 1335 |
+
"epoch": 2.174261323324102,
|
| 1336 |
+
"grad_norm": 87.5394515991211,
|
| 1337 |
+
"learning_rate": 5.314814814814815e-06,
|
| 1338 |
+
"loss": 9.6173,
|
| 1339 |
+
"step": 95000
|
| 1340 |
+
},
|
| 1341 |
+
{
|
| 1342 |
+
"epoch": 2.1857048039731763,
|
| 1343 |
+
"grad_norm": 0.0,
|
| 1344 |
+
"learning_rate": 5.301851851851852e-06,
|
| 1345 |
+
"loss": 9.4475,
|
| 1346 |
+
"step": 95500
|
| 1347 |
+
},
|
| 1348 |
+
{
|
| 1349 |
+
"epoch": 2.1971482846222505,
|
| 1350 |
+
"grad_norm": 1670.8548583984375,
|
| 1351 |
+
"learning_rate": 5.288888888888889e-06,
|
| 1352 |
+
"loss": 9.944,
|
| 1353 |
+
"step": 96000
|
| 1354 |
+
},
|
| 1355 |
+
{
|
| 1356 |
+
"epoch": 2.2085917652713247,
|
| 1357 |
+
"grad_norm": 0.0,
|
| 1358 |
+
"learning_rate": 5.275925925925925e-06,
|
| 1359 |
+
"loss": 9.0547,
|
| 1360 |
+
"step": 96500
|
| 1361 |
+
},
|
| 1362 |
+
{
|
| 1363 |
+
"epoch": 2.220035245920399,
|
| 1364 |
+
"grad_norm": 0.06572415679693222,
|
| 1365 |
+
"learning_rate": 5.262962962962963e-06,
|
| 1366 |
+
"loss": 8.8515,
|
| 1367 |
+
"step": 97000
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 2.2314787265694735,
|
| 1371 |
+
"grad_norm": 0.009067896753549576,
|
| 1372 |
+
"learning_rate": 5.25e-06,
|
| 1373 |
+
"loss": 10.3228,
|
| 1374 |
+
"step": 97500
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"epoch": 2.2429222072185477,
|
| 1378 |
+
"grad_norm": 0.0,
|
| 1379 |
+
"learning_rate": 5.237037037037037e-06,
|
| 1380 |
+
"loss": 10.091,
|
| 1381 |
+
"step": 98000
|
| 1382 |
+
},
|
| 1383 |
+
{
|
| 1384 |
+
"epoch": 2.254365687867622,
|
| 1385 |
+
"grad_norm": 12197.4453125,
|
| 1386 |
+
"learning_rate": 5.224074074074074e-06,
|
| 1387 |
+
"loss": 10.6507,
|
| 1388 |
+
"step": 98500
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 2.265809168516696,
|
| 1392 |
+
"grad_norm": 0.006908933632075787,
|
| 1393 |
+
"learning_rate": 5.211111111111111e-06,
|
| 1394 |
+
"loss": 9.9997,
|
| 1395 |
+
"step": 99000
|
| 1396 |
+
},
|
| 1397 |
+
{
|
| 1398 |
+
"epoch": 2.2772526491657703,
|
| 1399 |
+
"grad_norm": 786.2550659179688,
|
| 1400 |
+
"learning_rate": 5.198148148148148e-06,
|
| 1401 |
+
"loss": 10.4626,
|
| 1402 |
+
"step": 99500
|
| 1403 |
+
},
|
| 1404 |
+
{
|
| 1405 |
+
"epoch": 2.2886961298148445,
|
| 1406 |
+
"grad_norm": 206.81629943847656,
|
| 1407 |
+
"learning_rate": 5.185185185185185e-06,
|
| 1408 |
+
"loss": 9.6626,
|
| 1409 |
+
"step": 100000
|
| 1410 |
+
}
|
| 1411 |
+
],
|
| 1412 |
+
"logging_steps": 500,
|
| 1413 |
+
"max_steps": 300000,
|
| 1414 |
+
"num_input_tokens_seen": 0,
|
| 1415 |
+
"num_train_epochs": 7,
|
| 1416 |
+
"save_steps": 5000,
|
| 1417 |
+
"stateful_callbacks": {
|
| 1418 |
+
"TrainerControl": {
|
| 1419 |
+
"args": {
|
| 1420 |
+
"should_epoch_stop": false,
|
| 1421 |
+
"should_evaluate": false,
|
| 1422 |
+
"should_log": false,
|
| 1423 |
+
"should_save": true,
|
| 1424 |
+
"should_training_stop": false
|
| 1425 |
+
},
|
| 1426 |
+
"attributes": {}
|
| 1427 |
+
}
|
| 1428 |
+
},
|
| 1429 |
+
"total_flos": 0.0,
|
| 1430 |
+
"train_batch_size": 1,
|
| 1431 |
+
"trial_name": null,
|
| 1432 |
+
"trial_params": null
|
| 1433 |
+
}
|