Commit
·
21bee68
1
Parent(s):
2707673
distillclip-different-moon-37
Browse files- last-checkpoint/config.json +0 -169
- last-checkpoint/merges.txt +0 -0
- last-checkpoint/model.safetensors +0 -3
- last-checkpoint/optimizer.pt +0 -3
- last-checkpoint/preprocessor_config.json +0 -29
- last-checkpoint/rng_state.pth +0 -3
- last-checkpoint/scaler.pt +0 -3
- last-checkpoint/scheduler.pt +0 -3
- last-checkpoint/special_tokens_map.json +0 -24
- last-checkpoint/tokenizer.json +0 -0
- last-checkpoint/tokenizer_config.json +0 -34
- last-checkpoint/trainer_state.json +0 -3366
- last-checkpoint/training_args.bin +0 -3
- last-checkpoint/vocab.json +0 -0
- model.safetensors +1 -1
last-checkpoint/config.json
DELETED
|
@@ -1,169 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"_commit_hash": null,
|
| 3 |
-
"architectures": [
|
| 4 |
-
"CLIPForOfflineDistillation"
|
| 5 |
-
],
|
| 6 |
-
"initializer_factor": 1.0,
|
| 7 |
-
"logit_scale_init_value": 2.6592,
|
| 8 |
-
"loss_type": "mse",
|
| 9 |
-
"model_type": "clip",
|
| 10 |
-
"projection_dim": 256,
|
| 11 |
-
"teacher_name_or_path": "openai/clip-vit-base-patch32",
|
| 12 |
-
"text_config": {
|
| 13 |
-
"_name_or_path": "",
|
| 14 |
-
"add_cross_attention": false,
|
| 15 |
-
"architectures": null,
|
| 16 |
-
"attention_dropout": 0.0,
|
| 17 |
-
"bad_words_ids": null,
|
| 18 |
-
"begin_suppress_tokens": null,
|
| 19 |
-
"bos_token_id": 0,
|
| 20 |
-
"chunk_size_feed_forward": 0,
|
| 21 |
-
"cross_attention_hidden_size": null,
|
| 22 |
-
"decoder_start_token_id": null,
|
| 23 |
-
"diversity_penalty": 0.0,
|
| 24 |
-
"do_sample": false,
|
| 25 |
-
"early_stopping": false,
|
| 26 |
-
"encoder_no_repeat_ngram_size": 0,
|
| 27 |
-
"eos_token_id": 2,
|
| 28 |
-
"exponential_decay_length_penalty": null,
|
| 29 |
-
"finetuning_task": null,
|
| 30 |
-
"forced_bos_token_id": null,
|
| 31 |
-
"forced_eos_token_id": null,
|
| 32 |
-
"hidden_act": "quick_gelu",
|
| 33 |
-
"hidden_size": 512,
|
| 34 |
-
"id2label": {
|
| 35 |
-
"0": "LABEL_0",
|
| 36 |
-
"1": "LABEL_1"
|
| 37 |
-
},
|
| 38 |
-
"initializer_factor": 1.0,
|
| 39 |
-
"initializer_range": 0.02,
|
| 40 |
-
"intermediate_size": 2048,
|
| 41 |
-
"is_decoder": false,
|
| 42 |
-
"is_encoder_decoder": false,
|
| 43 |
-
"label2id": {
|
| 44 |
-
"LABEL_0": 0,
|
| 45 |
-
"LABEL_1": 1
|
| 46 |
-
},
|
| 47 |
-
"layer_norm_eps": 1e-05,
|
| 48 |
-
"length_penalty": 1.0,
|
| 49 |
-
"max_length": 20,
|
| 50 |
-
"max_position_embeddings": 77,
|
| 51 |
-
"min_length": 0,
|
| 52 |
-
"model_type": "clip_text_model",
|
| 53 |
-
"no_repeat_ngram_size": 0,
|
| 54 |
-
"num_attention_heads": 8,
|
| 55 |
-
"num_beam_groups": 1,
|
| 56 |
-
"num_beams": 1,
|
| 57 |
-
"num_hidden_layers": 6,
|
| 58 |
-
"num_return_sequences": 1,
|
| 59 |
-
"output_attentions": false,
|
| 60 |
-
"output_hidden_states": false,
|
| 61 |
-
"output_scores": false,
|
| 62 |
-
"pad_token_id": 1,
|
| 63 |
-
"prefix": null,
|
| 64 |
-
"problem_type": null,
|
| 65 |
-
"projection_dim": 512,
|
| 66 |
-
"pruned_heads": {},
|
| 67 |
-
"remove_invalid_values": false,
|
| 68 |
-
"repetition_penalty": 1.0,
|
| 69 |
-
"return_dict": true,
|
| 70 |
-
"return_dict_in_generate": false,
|
| 71 |
-
"sep_token_id": null,
|
| 72 |
-
"suppress_tokens": null,
|
| 73 |
-
"task_specific_params": null,
|
| 74 |
-
"temperature": 1.0,
|
| 75 |
-
"tf_legacy_loss": false,
|
| 76 |
-
"tie_encoder_decoder": false,
|
| 77 |
-
"tie_word_embeddings": true,
|
| 78 |
-
"tokenizer_class": null,
|
| 79 |
-
"top_k": 50,
|
| 80 |
-
"top_p": 1.0,
|
| 81 |
-
"torch_dtype": null,
|
| 82 |
-
"torchscript": false,
|
| 83 |
-
"transformers_version": "4.29.2",
|
| 84 |
-
"typical_p": 1.0,
|
| 85 |
-
"use_bfloat16": false,
|
| 86 |
-
"vocab_size": 49408
|
| 87 |
-
},
|
| 88 |
-
"torch_dtype": "float32",
|
| 89 |
-
"transformers_version": null,
|
| 90 |
-
"vision_config": {
|
| 91 |
-
"_name_or_path": "",
|
| 92 |
-
"add_cross_attention": false,
|
| 93 |
-
"architectures": null,
|
| 94 |
-
"attention_dropout": 0.0,
|
| 95 |
-
"bad_words_ids": null,
|
| 96 |
-
"begin_suppress_tokens": null,
|
| 97 |
-
"bos_token_id": null,
|
| 98 |
-
"chunk_size_feed_forward": 0,
|
| 99 |
-
"cross_attention_hidden_size": null,
|
| 100 |
-
"decoder_start_token_id": null,
|
| 101 |
-
"diversity_penalty": 0.0,
|
| 102 |
-
"do_sample": false,
|
| 103 |
-
"early_stopping": false,
|
| 104 |
-
"encoder_no_repeat_ngram_size": 0,
|
| 105 |
-
"eos_token_id": null,
|
| 106 |
-
"exponential_decay_length_penalty": null,
|
| 107 |
-
"finetuning_task": null,
|
| 108 |
-
"forced_bos_token_id": null,
|
| 109 |
-
"forced_eos_token_id": null,
|
| 110 |
-
"hidden_act": "quick_gelu",
|
| 111 |
-
"hidden_size": 384,
|
| 112 |
-
"id2label": {
|
| 113 |
-
"0": "LABEL_0",
|
| 114 |
-
"1": "LABEL_1"
|
| 115 |
-
},
|
| 116 |
-
"image_size": 224,
|
| 117 |
-
"initializer_factor": 1.0,
|
| 118 |
-
"initializer_range": 0.02,
|
| 119 |
-
"intermediate_size": 1536,
|
| 120 |
-
"is_decoder": false,
|
| 121 |
-
"is_encoder_decoder": false,
|
| 122 |
-
"label2id": {
|
| 123 |
-
"LABEL_0": 0,
|
| 124 |
-
"LABEL_1": 1
|
| 125 |
-
},
|
| 126 |
-
"layer_norm_eps": 1e-05,
|
| 127 |
-
"length_penalty": 1.0,
|
| 128 |
-
"max_length": 20,
|
| 129 |
-
"min_length": 0,
|
| 130 |
-
"model_type": "clip_vision_model",
|
| 131 |
-
"no_repeat_ngram_size": 0,
|
| 132 |
-
"num_attention_heads": 6,
|
| 133 |
-
"num_beam_groups": 1,
|
| 134 |
-
"num_beams": 1,
|
| 135 |
-
"num_channels": 3,
|
| 136 |
-
"num_hidden_layers": 12,
|
| 137 |
-
"num_return_sequences": 1,
|
| 138 |
-
"output_attentions": false,
|
| 139 |
-
"output_hidden_states": false,
|
| 140 |
-
"output_scores": false,
|
| 141 |
-
"pad_token_id": null,
|
| 142 |
-
"patch_size": 16,
|
| 143 |
-
"prefix": null,
|
| 144 |
-
"problem_type": null,
|
| 145 |
-
"projection_dim": 512,
|
| 146 |
-
"pruned_heads": {},
|
| 147 |
-
"remove_invalid_values": false,
|
| 148 |
-
"repetition_penalty": 1.0,
|
| 149 |
-
"return_dict": true,
|
| 150 |
-
"return_dict_in_generate": false,
|
| 151 |
-
"sep_token_id": null,
|
| 152 |
-
"suppress_tokens": null,
|
| 153 |
-
"task_specific_params": null,
|
| 154 |
-
"temperature": 1.0,
|
| 155 |
-
"tf_legacy_loss": false,
|
| 156 |
-
"tie_encoder_decoder": false,
|
| 157 |
-
"tie_word_embeddings": true,
|
| 158 |
-
"tokenizer_class": null,
|
| 159 |
-
"top_k": 50,
|
| 160 |
-
"top_p": 1.0,
|
| 161 |
-
"torch_dtype": null,
|
| 162 |
-
"torchscript": false,
|
| 163 |
-
"transformers_version": "4.29.2",
|
| 164 |
-
"typical_p": 1.0,
|
| 165 |
-
"use_bfloat16": false
|
| 166 |
-
},
|
| 167 |
-
"weight_r": 1.0,
|
| 168 |
-
"weight_s": 1.0
|
| 169 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
last-checkpoint/merges.txt
DELETED
|
The diff for this file is too large to render.
See raw diff
|
|
|
last-checkpoint/model.safetensors
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:359f0bb9c7fdaf6b9a47c37a482edc40df921b373339ce1a6c7b9f57147c1ae3
|
| 3 |
-
size 264627500
|
|
|
|
|
|
|
|
|
|
|
|
last-checkpoint/optimizer.pt
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:1d9a371842fd61230488662e5eb855b7be4e71789c97f32ee4d639b6966bcc40
|
| 3 |
-
size 529349637
|
|
|
|
|
|
|
|
|
|
|
|
last-checkpoint/preprocessor_config.json
DELETED
|
@@ -1,29 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"crop_size": {
|
| 3 |
-
"height": 224,
|
| 4 |
-
"width": 224
|
| 5 |
-
},
|
| 6 |
-
"do_center_crop": true,
|
| 7 |
-
"do_convert_rgb": true,
|
| 8 |
-
"do_normalize": true,
|
| 9 |
-
"do_rescale": true,
|
| 10 |
-
"do_resize": true,
|
| 11 |
-
"feature_extractor_type": "CLIPFeatureExtractor",
|
| 12 |
-
"image_mean": [
|
| 13 |
-
0.48145466,
|
| 14 |
-
0.4578275,
|
| 15 |
-
0.40821073
|
| 16 |
-
],
|
| 17 |
-
"image_processor_type": "CLIPImageProcessor",
|
| 18 |
-
"image_std": [
|
| 19 |
-
0.26862954,
|
| 20 |
-
0.26130258,
|
| 21 |
-
0.27577711
|
| 22 |
-
],
|
| 23 |
-
"processor_class": "CLIPProcessor",
|
| 24 |
-
"resample": 3,
|
| 25 |
-
"rescale_factor": 0.00392156862745098,
|
| 26 |
-
"size": {
|
| 27 |
-
"shortest_edge": 224
|
| 28 |
-
}
|
| 29 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
last-checkpoint/rng_state.pth
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:003dae60e1aad3bdb3943ababdac5d57f551939a6e1d0a3e9dee1a6ba7e680e3
|
| 3 |
-
size 14575
|
|
|
|
|
|
|
|
|
|
|
|
last-checkpoint/scaler.pt
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:456a4f0a789f4cbcc35bee1ee478eb09c60afb1bbecc6578d3c7ab655a844d1a
|
| 3 |
-
size 557
|
|
|
|
|
|
|
|
|
|
|
|
last-checkpoint/scheduler.pt
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:cad79c842bd93ea8f1ac897defc5b2d9d56cb0cbc0054b1a2ca367dd45b0c77a
|
| 3 |
-
size 627
|
|
|
|
|
|
|
|
|
|
|
|
last-checkpoint/special_tokens_map.json
DELETED
|
@@ -1,24 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"bos_token": {
|
| 3 |
-
"content": "<|startoftext|>",
|
| 4 |
-
"lstrip": false,
|
| 5 |
-
"normalized": true,
|
| 6 |
-
"rstrip": false,
|
| 7 |
-
"single_word": false
|
| 8 |
-
},
|
| 9 |
-
"eos_token": {
|
| 10 |
-
"content": "<|endoftext|>",
|
| 11 |
-
"lstrip": false,
|
| 12 |
-
"normalized": true,
|
| 13 |
-
"rstrip": false,
|
| 14 |
-
"single_word": false
|
| 15 |
-
},
|
| 16 |
-
"pad_token": "<|endoftext|>",
|
| 17 |
-
"unk_token": {
|
| 18 |
-
"content": "<|endoftext|>",
|
| 19 |
-
"lstrip": false,
|
| 20 |
-
"normalized": true,
|
| 21 |
-
"rstrip": false,
|
| 22 |
-
"single_word": false
|
| 23 |
-
}
|
| 24 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
last-checkpoint/tokenizer.json
DELETED
|
The diff for this file is too large to render.
See raw diff
|
|
|
last-checkpoint/tokenizer_config.json
DELETED
|
@@ -1,34 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"add_prefix_space": false,
|
| 3 |
-
"bos_token": {
|
| 4 |
-
"__type": "AddedToken",
|
| 5 |
-
"content": "<|startoftext|>",
|
| 6 |
-
"lstrip": false,
|
| 7 |
-
"normalized": true,
|
| 8 |
-
"rstrip": false,
|
| 9 |
-
"single_word": false
|
| 10 |
-
},
|
| 11 |
-
"clean_up_tokenization_spaces": true,
|
| 12 |
-
"do_lower_case": true,
|
| 13 |
-
"eos_token": {
|
| 14 |
-
"__type": "AddedToken",
|
| 15 |
-
"content": "<|endoftext|>",
|
| 16 |
-
"lstrip": false,
|
| 17 |
-
"normalized": true,
|
| 18 |
-
"rstrip": false,
|
| 19 |
-
"single_word": false
|
| 20 |
-
},
|
| 21 |
-
"errors": "replace",
|
| 22 |
-
"model_max_length": 77,
|
| 23 |
-
"pad_token": "<|endoftext|>",
|
| 24 |
-
"processor_class": "CLIPProcessor",
|
| 25 |
-
"tokenizer_class": "CLIPTokenizer",
|
| 26 |
-
"unk_token": {
|
| 27 |
-
"__type": "AddedToken",
|
| 28 |
-
"content": "<|endoftext|>",
|
| 29 |
-
"lstrip": false,
|
| 30 |
-
"normalized": true,
|
| 31 |
-
"rstrip": false,
|
| 32 |
-
"single_word": false
|
| 33 |
-
}
|
| 34 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
last-checkpoint/trainer_state.json
DELETED
|
@@ -1,3366 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"best_metric": null,
|
| 3 |
-
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 0.999612090830424,
|
| 5 |
-
"global_step": 33500,
|
| 6 |
-
"is_hyper_param_search": false,
|
| 7 |
-
"is_local_process_zero": true,
|
| 8 |
-
"is_world_process_zero": true,
|
| 9 |
-
"log_history": [
|
| 10 |
-
{
|
| 11 |
-
"epoch": 0.0,
|
| 12 |
-
"learning_rate": 3.0000000000000004e-07,
|
| 13 |
-
"loss": 0.1505,
|
| 14 |
-
"r_loss": 0.04637562483549118,
|
| 15 |
-
"s_loss": 0.008538205176591873,
|
| 16 |
-
"step": 100
|
| 17 |
-
},
|
| 18 |
-
{
|
| 19 |
-
"epoch": 0.01,
|
| 20 |
-
"learning_rate": 6.000000000000001e-07,
|
| 21 |
-
"loss": 0.0461,
|
| 22 |
-
"r_loss": 0.03377725929021835,
|
| 23 |
-
"s_loss": 0.003884976962581277,
|
| 24 |
-
"step": 200
|
| 25 |
-
},
|
| 26 |
-
{
|
| 27 |
-
"epoch": 0.01,
|
| 28 |
-
"learning_rate": 9e-07,
|
| 29 |
-
"loss": 0.034,
|
| 30 |
-
"r_loss": 0.028842270374298096,
|
| 31 |
-
"s_loss": 0.00349122891202569,
|
| 32 |
-
"step": 300
|
| 33 |
-
},
|
| 34 |
-
{
|
| 35 |
-
"epoch": 0.01,
|
| 36 |
-
"learning_rate": 1.2000000000000002e-06,
|
| 37 |
-
"loss": 0.0294,
|
| 38 |
-
"r_loss": 0.03719829022884369,
|
| 39 |
-
"s_loss": 0.0032082502730190754,
|
| 40 |
-
"step": 400
|
| 41 |
-
},
|
| 42 |
-
{
|
| 43 |
-
"epoch": 0.01,
|
| 44 |
-
"learning_rate": 1.5e-06,
|
| 45 |
-
"loss": 0.0259,
|
| 46 |
-
"r_loss": 0.02185184881091118,
|
| 47 |
-
"s_loss": 0.002866600640118122,
|
| 48 |
-
"step": 500
|
| 49 |
-
},
|
| 50 |
-
{
|
| 51 |
-
"epoch": 0.01,
|
| 52 |
-
"eval_loss": 0.022319668903946877,
|
| 53 |
-
"eval_r_loss": 0.019447144120931625,
|
| 54 |
-
"eval_runtime": 5.725,
|
| 55 |
-
"eval_s_loss": 0.0028725252486765385,
|
| 56 |
-
"eval_samples_per_second": 117.38,
|
| 57 |
-
"eval_steps_per_second": 117.38,
|
| 58 |
-
"step": 500
|
| 59 |
-
},
|
| 60 |
-
{
|
| 61 |
-
"epoch": 0.02,
|
| 62 |
-
"learning_rate": 1.8e-06,
|
| 63 |
-
"loss": 0.0238,
|
| 64 |
-
"r_loss": 0.01666960120201111,
|
| 65 |
-
"s_loss": 0.0029884909745305777,
|
| 66 |
-
"step": 600
|
| 67 |
-
},
|
| 68 |
-
{
|
| 69 |
-
"epoch": 0.02,
|
| 70 |
-
"learning_rate": 2.1000000000000002e-06,
|
| 71 |
-
"loss": 0.0226,
|
| 72 |
-
"r_loss": 0.01765752211213112,
|
| 73 |
-
"s_loss": 0.003066658042371273,
|
| 74 |
-
"step": 700
|
| 75 |
-
},
|
| 76 |
-
{
|
| 77 |
-
"epoch": 0.02,
|
| 78 |
-
"learning_rate": 2.4000000000000003e-06,
|
| 79 |
-
"loss": 0.0212,
|
| 80 |
-
"r_loss": 0.01967948116362095,
|
| 81 |
-
"s_loss": 0.002733011730015278,
|
| 82 |
-
"step": 800
|
| 83 |
-
},
|
| 84 |
-
{
|
| 85 |
-
"epoch": 0.03,
|
| 86 |
-
"learning_rate": 2.7e-06,
|
| 87 |
-
"loss": 0.0205,
|
| 88 |
-
"r_loss": 0.016631901264190674,
|
| 89 |
-
"s_loss": 0.0027432385832071304,
|
| 90 |
-
"step": 900
|
| 91 |
-
},
|
| 92 |
-
{
|
| 93 |
-
"epoch": 0.03,
|
| 94 |
-
"learning_rate": 3e-06,
|
| 95 |
-
"loss": 0.0197,
|
| 96 |
-
"r_loss": 0.016916554421186447,
|
| 97 |
-
"s_loss": 0.002646064618602395,
|
| 98 |
-
"step": 1000
|
| 99 |
-
},
|
| 100 |
-
{
|
| 101 |
-
"epoch": 0.03,
|
| 102 |
-
"eval_loss": 0.01776285283267498,
|
| 103 |
-
"eval_r_loss": 0.015205347910523415,
|
| 104 |
-
"eval_runtime": 5.7797,
|
| 105 |
-
"eval_s_loss": 0.0025575056206434965,
|
| 106 |
-
"eval_samples_per_second": 116.268,
|
| 107 |
-
"eval_steps_per_second": 116.268,
|
| 108 |
-
"step": 1000
|
| 109 |
-
},
|
| 110 |
-
{
|
| 111 |
-
"epoch": 0.03,
|
| 112 |
-
"learning_rate": 3.3e-06,
|
| 113 |
-
"loss": 0.0189,
|
| 114 |
-
"r_loss": 0.016178447753190994,
|
| 115 |
-
"s_loss": 0.002462042961269617,
|
| 116 |
-
"step": 1100
|
| 117 |
-
},
|
| 118 |
-
{
|
| 119 |
-
"epoch": 0.04,
|
| 120 |
-
"learning_rate": 3.6e-06,
|
| 121 |
-
"loss": 0.0182,
|
| 122 |
-
"r_loss": 0.01640625111758709,
|
| 123 |
-
"s_loss": 0.0024527523200958967,
|
| 124 |
-
"step": 1200
|
| 125 |
-
},
|
| 126 |
-
{
|
| 127 |
-
"epoch": 0.04,
|
| 128 |
-
"learning_rate": 3.9e-06,
|
| 129 |
-
"loss": 0.0181,
|
| 130 |
-
"r_loss": 0.013397695496678352,
|
| 131 |
-
"s_loss": 0.002130940556526184,
|
| 132 |
-
"step": 1300
|
| 133 |
-
},
|
| 134 |
-
{
|
| 135 |
-
"epoch": 0.04,
|
| 136 |
-
"learning_rate": 4.2000000000000004e-06,
|
| 137 |
-
"loss": 0.017,
|
| 138 |
-
"r_loss": 0.015975676476955414,
|
| 139 |
-
"s_loss": 0.002383481478318572,
|
| 140 |
-
"step": 1400
|
| 141 |
-
},
|
| 142 |
-
{
|
| 143 |
-
"epoch": 0.04,
|
| 144 |
-
"learning_rate": 4.5e-06,
|
| 145 |
-
"loss": 0.017,
|
| 146 |
-
"r_loss": 0.013270380906760693,
|
| 147 |
-
"s_loss": 0.0021610523108392954,
|
| 148 |
-
"step": 1500
|
| 149 |
-
},
|
| 150 |
-
{
|
| 151 |
-
"epoch": 0.04,
|
| 152 |
-
"eval_loss": 0.015290064737200737,
|
| 153 |
-
"eval_r_loss": 0.012945108115673065,
|
| 154 |
-
"eval_runtime": 5.6931,
|
| 155 |
-
"eval_s_loss": 0.002344956621527672,
|
| 156 |
-
"eval_samples_per_second": 118.039,
|
| 157 |
-
"eval_steps_per_second": 118.039,
|
| 158 |
-
"step": 1500
|
| 159 |
-
},
|
| 160 |
-
{
|
| 161 |
-
"epoch": 0.05,
|
| 162 |
-
"learning_rate": 4.800000000000001e-06,
|
| 163 |
-
"loss": 0.0161,
|
| 164 |
-
"r_loss": 0.013699322938919067,
|
| 165 |
-
"s_loss": 0.0022580809891223907,
|
| 166 |
-
"step": 1600
|
| 167 |
-
},
|
| 168 |
-
{
|
| 169 |
-
"epoch": 0.05,
|
| 170 |
-
"learning_rate": 5.1e-06,
|
| 171 |
-
"loss": 0.016,
|
| 172 |
-
"r_loss": 0.014964626170694828,
|
| 173 |
-
"s_loss": 0.0021504017058759928,
|
| 174 |
-
"step": 1700
|
| 175 |
-
},
|
| 176 |
-
{
|
| 177 |
-
"epoch": 0.05,
|
| 178 |
-
"learning_rate": 5.4e-06,
|
| 179 |
-
"loss": 0.016,
|
| 180 |
-
"r_loss": 0.013486243784427643,
|
| 181 |
-
"s_loss": 0.0020603144075721502,
|
| 182 |
-
"step": 1800
|
| 183 |
-
},
|
| 184 |
-
{
|
| 185 |
-
"epoch": 0.06,
|
| 186 |
-
"learning_rate": 5.7000000000000005e-06,
|
| 187 |
-
"loss": 0.0154,
|
| 188 |
-
"r_loss": 0.012973317876458168,
|
| 189 |
-
"s_loss": 0.0021089715883135796,
|
| 190 |
-
"step": 1900
|
| 191 |
-
},
|
| 192 |
-
{
|
| 193 |
-
"epoch": 0.06,
|
| 194 |
-
"learning_rate": 6e-06,
|
| 195 |
-
"loss": 0.0153,
|
| 196 |
-
"r_loss": 0.015371004119515419,
|
| 197 |
-
"s_loss": 0.002191081177443266,
|
| 198 |
-
"step": 2000
|
| 199 |
-
},
|
| 200 |
-
{
|
| 201 |
-
"epoch": 0.06,
|
| 202 |
-
"eval_loss": 0.013254178687930107,
|
| 203 |
-
"eval_r_loss": 0.011181775480508804,
|
| 204 |
-
"eval_runtime": 5.6968,
|
| 205 |
-
"eval_s_loss": 0.0020724027417600155,
|
| 206 |
-
"eval_samples_per_second": 117.961,
|
| 207 |
-
"eval_steps_per_second": 117.961,
|
| 208 |
-
"step": 2000
|
| 209 |
-
},
|
| 210 |
-
{
|
| 211 |
-
"epoch": 0.06,
|
| 212 |
-
"learning_rate": 6.3e-06,
|
| 213 |
-
"loss": 0.015,
|
| 214 |
-
"r_loss": 0.011732030659914017,
|
| 215 |
-
"s_loss": 0.0019391687819734216,
|
| 216 |
-
"step": 2100
|
| 217 |
-
},
|
| 218 |
-
{
|
| 219 |
-
"epoch": 0.07,
|
| 220 |
-
"learning_rate": 6.6e-06,
|
| 221 |
-
"loss": 0.0147,
|
| 222 |
-
"r_loss": 0.015546457841992378,
|
| 223 |
-
"s_loss": 0.0020762544590979815,
|
| 224 |
-
"step": 2200
|
| 225 |
-
},
|
| 226 |
-
{
|
| 227 |
-
"epoch": 0.07,
|
| 228 |
-
"learning_rate": 6.900000000000001e-06,
|
| 229 |
-
"loss": 0.0146,
|
| 230 |
-
"r_loss": 0.009786777198314667,
|
| 231 |
-
"s_loss": 0.0019506356911733747,
|
| 232 |
-
"step": 2300
|
| 233 |
-
},
|
| 234 |
-
{
|
| 235 |
-
"epoch": 0.07,
|
| 236 |
-
"learning_rate": 7.2e-06,
|
| 237 |
-
"loss": 0.0144,
|
| 238 |
-
"r_loss": 0.012893063016235828,
|
| 239 |
-
"s_loss": 0.002081776736304164,
|
| 240 |
-
"step": 2400
|
| 241 |
-
},
|
| 242 |
-
{
|
| 243 |
-
"epoch": 0.07,
|
| 244 |
-
"learning_rate": 7.5e-06,
|
| 245 |
-
"loss": 0.0142,
|
| 246 |
-
"r_loss": 0.00976946298032999,
|
| 247 |
-
"s_loss": 0.0017642343882471323,
|
| 248 |
-
"step": 2500
|
| 249 |
-
},
|
| 250 |
-
{
|
| 251 |
-
"epoch": 0.07,
|
| 252 |
-
"eval_loss": 0.01350394356995821,
|
| 253 |
-
"eval_r_loss": 0.01156766340136528,
|
| 254 |
-
"eval_runtime": 5.7349,
|
| 255 |
-
"eval_s_loss": 0.001936280052177608,
|
| 256 |
-
"eval_samples_per_second": 117.177,
|
| 257 |
-
"eval_steps_per_second": 117.177,
|
| 258 |
-
"step": 2500
|
| 259 |
-
},
|
| 260 |
-
{
|
| 261 |
-
"epoch": 0.08,
|
| 262 |
-
"learning_rate": 7.8e-06,
|
| 263 |
-
"loss": 0.0136,
|
| 264 |
-
"r_loss": 0.01364838145673275,
|
| 265 |
-
"s_loss": 0.0019653863273561,
|
| 266 |
-
"step": 2600
|
| 267 |
-
},
|
| 268 |
-
{
|
| 269 |
-
"epoch": 0.08,
|
| 270 |
-
"learning_rate": 8.1e-06,
|
| 271 |
-
"loss": 0.014,
|
| 272 |
-
"r_loss": 0.013060592114925385,
|
| 273 |
-
"s_loss": 0.00197937642224133,
|
| 274 |
-
"step": 2700
|
| 275 |
-
},
|
| 276 |
-
{
|
| 277 |
-
"epoch": 0.08,
|
| 278 |
-
"learning_rate": 8.400000000000001e-06,
|
| 279 |
-
"loss": 0.0134,
|
| 280 |
-
"r_loss": 0.010259253904223442,
|
| 281 |
-
"s_loss": 0.0018942919559776783,
|
| 282 |
-
"step": 2800
|
| 283 |
-
},
|
| 284 |
-
{
|
| 285 |
-
"epoch": 0.09,
|
| 286 |
-
"learning_rate": 8.7e-06,
|
| 287 |
-
"loss": 0.0136,
|
| 288 |
-
"r_loss": 0.010634353384375572,
|
| 289 |
-
"s_loss": 0.002018529223278165,
|
| 290 |
-
"step": 2900
|
| 291 |
-
},
|
| 292 |
-
{
|
| 293 |
-
"epoch": 0.09,
|
| 294 |
-
"learning_rate": 9e-06,
|
| 295 |
-
"loss": 0.0134,
|
| 296 |
-
"r_loss": 0.01226745918393135,
|
| 297 |
-
"s_loss": 0.0018310793675482273,
|
| 298 |
-
"step": 3000
|
| 299 |
-
},
|
| 300 |
-
{
|
| 301 |
-
"epoch": 0.09,
|
| 302 |
-
"eval_loss": 0.01376401074230671,
|
| 303 |
-
"eval_r_loss": 0.011921562254428864,
|
| 304 |
-
"eval_runtime": 5.8031,
|
| 305 |
-
"eval_s_loss": 0.0018424488371238112,
|
| 306 |
-
"eval_samples_per_second": 115.8,
|
| 307 |
-
"eval_steps_per_second": 115.8,
|
| 308 |
-
"step": 3000
|
| 309 |
-
},
|
| 310 |
-
{
|
| 311 |
-
"epoch": 0.09,
|
| 312 |
-
"learning_rate": 9.3e-06,
|
| 313 |
-
"loss": 0.0131,
|
| 314 |
-
"r_loss": 0.011366230435669422,
|
| 315 |
-
"s_loss": 0.0018870271742343903,
|
| 316 |
-
"step": 3100
|
| 317 |
-
},
|
| 318 |
-
{
|
| 319 |
-
"epoch": 0.1,
|
| 320 |
-
"learning_rate": 9.600000000000001e-06,
|
| 321 |
-
"loss": 0.0129,
|
| 322 |
-
"r_loss": 0.015041200444102287,
|
| 323 |
-
"s_loss": 0.0019711025524884462,
|
| 324 |
-
"step": 3200
|
| 325 |
-
},
|
| 326 |
-
{
|
| 327 |
-
"epoch": 0.1,
|
| 328 |
-
"learning_rate": 9.9e-06,
|
| 329 |
-
"loss": 0.0126,
|
| 330 |
-
"r_loss": 0.010199671611189842,
|
| 331 |
-
"s_loss": 0.0018134403508156538,
|
| 332 |
-
"step": 3300
|
| 333 |
-
},
|
| 334 |
-
{
|
| 335 |
-
"epoch": 0.1,
|
| 336 |
-
"learning_rate": 1.02e-05,
|
| 337 |
-
"loss": 0.0127,
|
| 338 |
-
"r_loss": 0.009185859933495522,
|
| 339 |
-
"s_loss": 0.0016790288500487804,
|
| 340 |
-
"step": 3400
|
| 341 |
-
},
|
| 342 |
-
{
|
| 343 |
-
"epoch": 0.1,
|
| 344 |
-
"learning_rate": 1.05e-05,
|
| 345 |
-
"loss": 0.0127,
|
| 346 |
-
"r_loss": 0.011276321485638618,
|
| 347 |
-
"s_loss": 0.0016570865409448743,
|
| 348 |
-
"step": 3500
|
| 349 |
-
},
|
| 350 |
-
{
|
| 351 |
-
"epoch": 0.1,
|
| 352 |
-
"eval_loss": 0.011658551171422005,
|
| 353 |
-
"eval_r_loss": 0.009879879653453827,
|
| 354 |
-
"eval_runtime": 5.7067,
|
| 355 |
-
"eval_s_loss": 0.001778671983629465,
|
| 356 |
-
"eval_samples_per_second": 117.756,
|
| 357 |
-
"eval_steps_per_second": 117.756,
|
| 358 |
-
"step": 3500
|
| 359 |
-
},
|
| 360 |
-
{
|
| 361 |
-
"epoch": 0.11,
|
| 362 |
-
"learning_rate": 1.08e-05,
|
| 363 |
-
"loss": 0.0126,
|
| 364 |
-
"r_loss": 0.008440356701612473,
|
| 365 |
-
"s_loss": 0.0017231483943760395,
|
| 366 |
-
"step": 3600
|
| 367 |
-
},
|
| 368 |
-
{
|
| 369 |
-
"epoch": 0.11,
|
| 370 |
-
"learning_rate": 1.11e-05,
|
| 371 |
-
"loss": 0.0122,
|
| 372 |
-
"r_loss": 0.009153485298156738,
|
| 373 |
-
"s_loss": 0.0017831036821007729,
|
| 374 |
-
"step": 3700
|
| 375 |
-
},
|
| 376 |
-
{
|
| 377 |
-
"epoch": 0.11,
|
| 378 |
-
"learning_rate": 1.1400000000000001e-05,
|
| 379 |
-
"loss": 0.012,
|
| 380 |
-
"r_loss": 0.011471357196569443,
|
| 381 |
-
"s_loss": 0.001937872963026166,
|
| 382 |
-
"step": 3800
|
| 383 |
-
},
|
| 384 |
-
{
|
| 385 |
-
"epoch": 0.12,
|
| 386 |
-
"learning_rate": 1.1700000000000001e-05,
|
| 387 |
-
"loss": 0.0121,
|
| 388 |
-
"r_loss": 0.009361416101455688,
|
| 389 |
-
"s_loss": 0.001662532682530582,
|
| 390 |
-
"step": 3900
|
| 391 |
-
},
|
| 392 |
-
{
|
| 393 |
-
"epoch": 0.12,
|
| 394 |
-
"learning_rate": 1.2e-05,
|
| 395 |
-
"loss": 0.012,
|
| 396 |
-
"r_loss": 0.013415796682238579,
|
| 397 |
-
"s_loss": 0.0016665093135088682,
|
| 398 |
-
"step": 4000
|
| 399 |
-
},
|
| 400 |
-
{
|
| 401 |
-
"epoch": 0.12,
|
| 402 |
-
"eval_loss": 0.011559335514903069,
|
| 403 |
-
"eval_r_loss": 0.009901713579893112,
|
| 404 |
-
"eval_runtime": 5.7585,
|
| 405 |
-
"eval_s_loss": 0.0016576218185946345,
|
| 406 |
-
"eval_samples_per_second": 116.697,
|
| 407 |
-
"eval_steps_per_second": 116.697,
|
| 408 |
-
"step": 4000
|
| 409 |
-
},
|
| 410 |
-
{
|
| 411 |
-
"epoch": 0.12,
|
| 412 |
-
"learning_rate": 1.2299999999999999e-05,
|
| 413 |
-
"loss": 0.0122,
|
| 414 |
-
"r_loss": 0.009355029091238976,
|
| 415 |
-
"s_loss": 0.001734372228384018,
|
| 416 |
-
"step": 4100
|
| 417 |
-
},
|
| 418 |
-
{
|
| 419 |
-
"epoch": 0.13,
|
| 420 |
-
"learning_rate": 1.26e-05,
|
| 421 |
-
"loss": 0.0118,
|
| 422 |
-
"r_loss": 0.009848803281784058,
|
| 423 |
-
"s_loss": 0.0016517819603905082,
|
| 424 |
-
"step": 4200
|
| 425 |
-
},
|
| 426 |
-
{
|
| 427 |
-
"epoch": 0.13,
|
| 428 |
-
"learning_rate": 1.29e-05,
|
| 429 |
-
"loss": 0.0117,
|
| 430 |
-
"r_loss": 0.008925353176891804,
|
| 431 |
-
"s_loss": 0.0016535022296011448,
|
| 432 |
-
"step": 4300
|
| 433 |
-
},
|
| 434 |
-
{
|
| 435 |
-
"epoch": 0.13,
|
| 436 |
-
"learning_rate": 1.32e-05,
|
| 437 |
-
"loss": 0.0117,
|
| 438 |
-
"r_loss": 0.009858155623078346,
|
| 439 |
-
"s_loss": 0.001591067761182785,
|
| 440 |
-
"step": 4400
|
| 441 |
-
},
|
| 442 |
-
{
|
| 443 |
-
"epoch": 0.13,
|
| 444 |
-
"learning_rate": 1.3500000000000001e-05,
|
| 445 |
-
"loss": 0.0115,
|
| 446 |
-
"r_loss": 0.009000759571790695,
|
| 447 |
-
"s_loss": 0.001509728142991662,
|
| 448 |
-
"step": 4500
|
| 449 |
-
},
|
| 450 |
-
{
|
| 451 |
-
"epoch": 0.13,
|
| 452 |
-
"eval_loss": 0.011256811209022999,
|
| 453 |
-
"eval_r_loss": 0.009671147912740707,
|
| 454 |
-
"eval_runtime": 5.7598,
|
| 455 |
-
"eval_s_loss": 0.0015856630634516478,
|
| 456 |
-
"eval_samples_per_second": 116.67,
|
| 457 |
-
"eval_steps_per_second": 116.67,
|
| 458 |
-
"step": 4500
|
| 459 |
-
},
|
| 460 |
-
{
|
| 461 |
-
"epoch": 0.14,
|
| 462 |
-
"learning_rate": 1.3800000000000002e-05,
|
| 463 |
-
"loss": 0.0113,
|
| 464 |
-
"r_loss": 0.00949438102543354,
|
| 465 |
-
"s_loss": 0.0015365839935839176,
|
| 466 |
-
"step": 4600
|
| 467 |
-
},
|
| 468 |
-
{
|
| 469 |
-
"epoch": 0.14,
|
| 470 |
-
"learning_rate": 1.4099999999999999e-05,
|
| 471 |
-
"loss": 0.0114,
|
| 472 |
-
"r_loss": 0.01006004773080349,
|
| 473 |
-
"s_loss": 0.0015926426276564598,
|
| 474 |
-
"step": 4700
|
| 475 |
-
},
|
| 476 |
-
{
|
| 477 |
-
"epoch": 0.14,
|
| 478 |
-
"learning_rate": 1.44e-05,
|
| 479 |
-
"loss": 0.0114,
|
| 480 |
-
"r_loss": 0.010145231150090694,
|
| 481 |
-
"s_loss": 0.0015732853207737207,
|
| 482 |
-
"step": 4800
|
| 483 |
-
},
|
| 484 |
-
{
|
| 485 |
-
"epoch": 0.15,
|
| 486 |
-
"learning_rate": 1.47e-05,
|
| 487 |
-
"loss": 0.0112,
|
| 488 |
-
"r_loss": 0.00975881703197956,
|
| 489 |
-
"s_loss": 0.0014939116081222892,
|
| 490 |
-
"step": 4900
|
| 491 |
-
},
|
| 492 |
-
{
|
| 493 |
-
"epoch": 0.15,
|
| 494 |
-
"learning_rate": 1.5e-05,
|
| 495 |
-
"loss": 0.0111,
|
| 496 |
-
"r_loss": 0.009045148268342018,
|
| 497 |
-
"s_loss": 0.0014695697464048862,
|
| 498 |
-
"step": 5000
|
| 499 |
-
},
|
| 500 |
-
{
|
| 501 |
-
"epoch": 0.15,
|
| 502 |
-
"eval_loss": 0.011217299848794937,
|
| 503 |
-
"eval_r_loss": 0.009782791137695312,
|
| 504 |
-
"eval_runtime": 5.6702,
|
| 505 |
-
"eval_s_loss": 0.0014345089439302683,
|
| 506 |
-
"eval_samples_per_second": 118.514,
|
| 507 |
-
"eval_steps_per_second": 118.514,
|
| 508 |
-
"step": 5000
|
| 509 |
-
},
|
| 510 |
-
{
|
| 511 |
-
"epoch": 0.15,
|
| 512 |
-
"learning_rate": 1.53e-05,
|
| 513 |
-
"loss": 0.0111,
|
| 514 |
-
"r_loss": 0.009601066820323467,
|
| 515 |
-
"s_loss": 0.001514170435257256,
|
| 516 |
-
"step": 5100
|
| 517 |
-
},
|
| 518 |
-
{
|
| 519 |
-
"epoch": 0.16,
|
| 520 |
-
"learning_rate": 1.56e-05,
|
| 521 |
-
"loss": 0.0112,
|
| 522 |
-
"r_loss": 0.00870747584849596,
|
| 523 |
-
"s_loss": 0.0014461548998951912,
|
| 524 |
-
"step": 5200
|
| 525 |
-
},
|
| 526 |
-
{
|
| 527 |
-
"epoch": 0.16,
|
| 528 |
-
"learning_rate": 1.59e-05,
|
| 529 |
-
"loss": 0.0109,
|
| 530 |
-
"r_loss": 0.008360641077160835,
|
| 531 |
-
"s_loss": 0.0014414612669497728,
|
| 532 |
-
"step": 5300
|
| 533 |
-
},
|
| 534 |
-
{
|
| 535 |
-
"epoch": 0.16,
|
| 536 |
-
"learning_rate": 1.62e-05,
|
| 537 |
-
"loss": 0.0108,
|
| 538 |
-
"r_loss": 0.00921720638871193,
|
| 539 |
-
"s_loss": 0.0014027974102646112,
|
| 540 |
-
"step": 5400
|
| 541 |
-
},
|
| 542 |
-
{
|
| 543 |
-
"epoch": 0.16,
|
| 544 |
-
"learning_rate": 1.65e-05,
|
| 545 |
-
"loss": 0.0108,
|
| 546 |
-
"r_loss": 0.00895563792437315,
|
| 547 |
-
"s_loss": 0.0014760666526854038,
|
| 548 |
-
"step": 5500
|
| 549 |
-
},
|
| 550 |
-
{
|
| 551 |
-
"epoch": 0.16,
|
| 552 |
-
"eval_loss": 0.01118839718401432,
|
| 553 |
-
"eval_r_loss": 0.00973587017506361,
|
| 554 |
-
"eval_runtime": 5.7152,
|
| 555 |
-
"eval_s_loss": 0.0014525266597047448,
|
| 556 |
-
"eval_samples_per_second": 117.581,
|
| 557 |
-
"eval_steps_per_second": 117.581,
|
| 558 |
-
"step": 5500
|
| 559 |
-
},
|
| 560 |
-
{
|
| 561 |
-
"epoch": 0.17,
|
| 562 |
-
"learning_rate": 1.6800000000000002e-05,
|
| 563 |
-
"loss": 0.0107,
|
| 564 |
-
"r_loss": 0.007762259803712368,
|
| 565 |
-
"s_loss": 0.0015427314210683107,
|
| 566 |
-
"step": 5600
|
| 567 |
-
},
|
| 568 |
-
{
|
| 569 |
-
"epoch": 0.17,
|
| 570 |
-
"learning_rate": 1.71e-05,
|
| 571 |
-
"loss": 0.0108,
|
| 572 |
-
"r_loss": 0.009653204120695591,
|
| 573 |
-
"s_loss": 0.0015951216919347644,
|
| 574 |
-
"step": 5700
|
| 575 |
-
},
|
| 576 |
-
{
|
| 577 |
-
"epoch": 0.17,
|
| 578 |
-
"learning_rate": 1.74e-05,
|
| 579 |
-
"loss": 0.0106,
|
| 580 |
-
"r_loss": 0.007989178411662579,
|
| 581 |
-
"s_loss": 0.001397020067088306,
|
| 582 |
-
"step": 5800
|
| 583 |
-
},
|
| 584 |
-
{
|
| 585 |
-
"epoch": 0.18,
|
| 586 |
-
"learning_rate": 1.77e-05,
|
| 587 |
-
"loss": 0.0107,
|
| 588 |
-
"r_loss": 0.009275095537304878,
|
| 589 |
-
"s_loss": 0.001448939205147326,
|
| 590 |
-
"step": 5900
|
| 591 |
-
},
|
| 592 |
-
{
|
| 593 |
-
"epoch": 0.18,
|
| 594 |
-
"learning_rate": 1.8e-05,
|
| 595 |
-
"loss": 0.0106,
|
| 596 |
-
"r_loss": 0.008456457406282425,
|
| 597 |
-
"s_loss": 0.0014687306247651577,
|
| 598 |
-
"step": 6000
|
| 599 |
-
},
|
| 600 |
-
{
|
| 601 |
-
"epoch": 0.18,
|
| 602 |
-
"eval_loss": 0.010676538571715355,
|
| 603 |
-
"eval_r_loss": 0.009258019737899303,
|
| 604 |
-
"eval_runtime": 5.7202,
|
| 605 |
-
"eval_s_loss": 0.0014185188338160515,
|
| 606 |
-
"eval_samples_per_second": 117.477,
|
| 607 |
-
"eval_steps_per_second": 117.477,
|
| 608 |
-
"step": 6000
|
| 609 |
-
},
|
| 610 |
-
{
|
| 611 |
-
"epoch": 0.18,
|
| 612 |
-
"learning_rate": 1.83e-05,
|
| 613 |
-
"loss": 0.0105,
|
| 614 |
-
"r_loss": 0.008195489645004272,
|
| 615 |
-
"s_loss": 0.0014291288098320365,
|
| 616 |
-
"step": 6100
|
| 617 |
-
},
|
| 618 |
-
{
|
| 619 |
-
"epoch": 0.19,
|
| 620 |
-
"learning_rate": 1.86e-05,
|
| 621 |
-
"loss": 0.0102,
|
| 622 |
-
"r_loss": 0.008191170170903206,
|
| 623 |
-
"s_loss": 0.0015366484876722097,
|
| 624 |
-
"step": 6200
|
| 625 |
-
},
|
| 626 |
-
{
|
| 627 |
-
"epoch": 0.19,
|
| 628 |
-
"learning_rate": 1.8900000000000002e-05,
|
| 629 |
-
"loss": 0.0102,
|
| 630 |
-
"r_loss": 0.00822894275188446,
|
| 631 |
-
"s_loss": 0.0014636135892942548,
|
| 632 |
-
"step": 6300
|
| 633 |
-
},
|
| 634 |
-
{
|
| 635 |
-
"epoch": 0.19,
|
| 636 |
-
"learning_rate": 1.9200000000000003e-05,
|
| 637 |
-
"loss": 0.0103,
|
| 638 |
-
"r_loss": 0.0088454969227314,
|
| 639 |
-
"s_loss": 0.001345100230537355,
|
| 640 |
-
"step": 6400
|
| 641 |
-
},
|
| 642 |
-
{
|
| 643 |
-
"epoch": 0.19,
|
| 644 |
-
"learning_rate": 1.95e-05,
|
| 645 |
-
"loss": 0.0105,
|
| 646 |
-
"r_loss": 0.01023932732641697,
|
| 647 |
-
"s_loss": 0.001511400449089706,
|
| 648 |
-
"step": 6500
|
| 649 |
-
},
|
| 650 |
-
{
|
| 651 |
-
"epoch": 0.19,
|
| 652 |
-
"eval_loss": 0.010188945569097996,
|
| 653 |
-
"eval_r_loss": 0.00889565609395504,
|
| 654 |
-
"eval_runtime": 6.5641,
|
| 655 |
-
"eval_s_loss": 0.0012932894751429558,
|
| 656 |
-
"eval_samples_per_second": 102.375,
|
| 657 |
-
"eval_steps_per_second": 102.375,
|
| 658 |
-
"step": 6500
|
| 659 |
-
},
|
| 660 |
-
{
|
| 661 |
-
"epoch": 0.2,
|
| 662 |
-
"learning_rate": 1.98e-05,
|
| 663 |
-
"loss": 0.0104,
|
| 664 |
-
"r_loss": 0.00900462456047535,
|
| 665 |
-
"s_loss": 0.001456625759601593,
|
| 666 |
-
"step": 6600
|
| 667 |
-
},
|
| 668 |
-
{
|
| 669 |
-
"epoch": 0.2,
|
| 670 |
-
"learning_rate": 2.01e-05,
|
| 671 |
-
"loss": 0.0102,
|
| 672 |
-
"r_loss": 0.00964261218905449,
|
| 673 |
-
"s_loss": 0.001401584129780531,
|
| 674 |
-
"step": 6700
|
| 675 |
-
},
|
| 676 |
-
{
|
| 677 |
-
"epoch": 0.2,
|
| 678 |
-
"learning_rate": 2.04e-05,
|
| 679 |
-
"loss": 0.0101,
|
| 680 |
-
"r_loss": 0.007694972679018974,
|
| 681 |
-
"s_loss": 0.0014238519361242652,
|
| 682 |
-
"step": 6800
|
| 683 |
-
},
|
| 684 |
-
{
|
| 685 |
-
"epoch": 0.21,
|
| 686 |
-
"learning_rate": 2.07e-05,
|
| 687 |
-
"loss": 0.0102,
|
| 688 |
-
"r_loss": 0.009449545294046402,
|
| 689 |
-
"s_loss": 0.0014661129098385572,
|
| 690 |
-
"step": 6900
|
| 691 |
-
},
|
| 692 |
-
{
|
| 693 |
-
"epoch": 0.21,
|
| 694 |
-
"learning_rate": 2.1e-05,
|
| 695 |
-
"loss": 0.0101,
|
| 696 |
-
"r_loss": 0.00955035537481308,
|
| 697 |
-
"s_loss": 0.001392255537211895,
|
| 698 |
-
"step": 7000
|
| 699 |
-
},
|
| 700 |
-
{
|
| 701 |
-
"epoch": 0.21,
|
| 702 |
-
"eval_loss": 0.010023693554103374,
|
| 703 |
-
"eval_r_loss": 0.008697763085365295,
|
| 704 |
-
"eval_runtime": 5.7082,
|
| 705 |
-
"eval_s_loss": 0.001325930585153401,
|
| 706 |
-
"eval_samples_per_second": 117.726,
|
| 707 |
-
"eval_steps_per_second": 117.726,
|
| 708 |
-
"step": 7000
|
| 709 |
-
},
|
| 710 |
-
{
|
| 711 |
-
"epoch": 0.21,
|
| 712 |
-
"learning_rate": 2.13e-05,
|
| 713 |
-
"loss": 0.0102,
|
| 714 |
-
"r_loss": 0.00833200104534626,
|
| 715 |
-
"s_loss": 0.0014237057184800506,
|
| 716 |
-
"step": 7100
|
| 717 |
-
},
|
| 718 |
-
{
|
| 719 |
-
"epoch": 0.21,
|
| 720 |
-
"learning_rate": 2.16e-05,
|
| 721 |
-
"loss": 0.0099,
|
| 722 |
-
"r_loss": 0.008094298653304577,
|
| 723 |
-
"s_loss": 0.0014223785838112235,
|
| 724 |
-
"step": 7200
|
| 725 |
-
},
|
| 726 |
-
{
|
| 727 |
-
"epoch": 0.22,
|
| 728 |
-
"learning_rate": 2.19e-05,
|
| 729 |
-
"loss": 0.01,
|
| 730 |
-
"r_loss": 0.00752690713852644,
|
| 731 |
-
"s_loss": 0.0013399685267359018,
|
| 732 |
-
"step": 7300
|
| 733 |
-
},
|
| 734 |
-
{
|
| 735 |
-
"epoch": 0.22,
|
| 736 |
-
"learning_rate": 2.22e-05,
|
| 737 |
-
"loss": 0.0099,
|
| 738 |
-
"r_loss": 0.00953773781657219,
|
| 739 |
-
"s_loss": 0.001381666399538517,
|
| 740 |
-
"step": 7400
|
| 741 |
-
},
|
| 742 |
-
{
|
| 743 |
-
"epoch": 0.22,
|
| 744 |
-
"learning_rate": 2.25e-05,
|
| 745 |
-
"loss": 0.0098,
|
| 746 |
-
"r_loss": 0.008129315450787544,
|
| 747 |
-
"s_loss": 0.0011789320269599557,
|
| 748 |
-
"step": 7500
|
| 749 |
-
},
|
| 750 |
-
{
|
| 751 |
-
"epoch": 0.22,
|
| 752 |
-
"eval_loss": 0.010138859041035175,
|
| 753 |
-
"eval_r_loss": 0.00888618640601635,
|
| 754 |
-
"eval_runtime": 5.8833,
|
| 755 |
-
"eval_s_loss": 0.001252672984264791,
|
| 756 |
-
"eval_samples_per_second": 114.221,
|
| 757 |
-
"eval_steps_per_second": 114.221,
|
| 758 |
-
"step": 7500
|
| 759 |
-
},
|
| 760 |
-
{
|
| 761 |
-
"epoch": 0.23,
|
| 762 |
-
"learning_rate": 2.2800000000000002e-05,
|
| 763 |
-
"loss": 0.0098,
|
| 764 |
-
"r_loss": 0.009464412927627563,
|
| 765 |
-
"s_loss": 0.001395503873936832,
|
| 766 |
-
"step": 7600
|
| 767 |
-
},
|
| 768 |
-
{
|
| 769 |
-
"epoch": 0.23,
|
| 770 |
-
"learning_rate": 2.3100000000000002e-05,
|
| 771 |
-
"loss": 0.0097,
|
| 772 |
-
"r_loss": 0.00881454348564148,
|
| 773 |
-
"s_loss": 0.001235602656379342,
|
| 774 |
-
"step": 7700
|
| 775 |
-
},
|
| 776 |
-
{
|
| 777 |
-
"epoch": 0.23,
|
| 778 |
-
"learning_rate": 2.3400000000000003e-05,
|
| 779 |
-
"loss": 0.01,
|
| 780 |
-
"r_loss": 0.00844954326748848,
|
| 781 |
-
"s_loss": 0.0012883525341749191,
|
| 782 |
-
"step": 7800
|
| 783 |
-
},
|
| 784 |
-
{
|
| 785 |
-
"epoch": 0.24,
|
| 786 |
-
"learning_rate": 2.37e-05,
|
| 787 |
-
"loss": 0.0095,
|
| 788 |
-
"r_loss": 0.008428743109107018,
|
| 789 |
-
"s_loss": 0.001225842977873981,
|
| 790 |
-
"step": 7900
|
| 791 |
-
},
|
| 792 |
-
{
|
| 793 |
-
"epoch": 0.24,
|
| 794 |
-
"learning_rate": 2.4e-05,
|
| 795 |
-
"loss": 0.0098,
|
| 796 |
-
"r_loss": 0.008358178660273552,
|
| 797 |
-
"s_loss": 0.0013608363224193454,
|
| 798 |
-
"step": 8000
|
| 799 |
-
},
|
| 800 |
-
{
|
| 801 |
-
"epoch": 0.24,
|
| 802 |
-
"eval_loss": 0.010027415119111538,
|
| 803 |
-
"eval_r_loss": 0.008776991628110409,
|
| 804 |
-
"eval_runtime": 5.7172,
|
| 805 |
-
"eval_s_loss": 0.0012504233745858073,
|
| 806 |
-
"eval_samples_per_second": 117.54,
|
| 807 |
-
"eval_steps_per_second": 117.54,
|
| 808 |
-
"step": 8000
|
| 809 |
-
},
|
| 810 |
-
{
|
| 811 |
-
"epoch": 0.24,
|
| 812 |
-
"learning_rate": 2.43e-05,
|
| 813 |
-
"loss": 0.0097,
|
| 814 |
-
"r_loss": 0.007988743484020233,
|
| 815 |
-
"s_loss": 0.0012902095913887024,
|
| 816 |
-
"step": 8100
|
| 817 |
-
},
|
| 818 |
-
{
|
| 819 |
-
"epoch": 0.24,
|
| 820 |
-
"learning_rate": 2.4599999999999998e-05,
|
| 821 |
-
"loss": 0.0097,
|
| 822 |
-
"r_loss": 0.00881593581289053,
|
| 823 |
-
"s_loss": 0.0012137378798797727,
|
| 824 |
-
"step": 8200
|
| 825 |
-
},
|
| 826 |
-
{
|
| 827 |
-
"epoch": 0.25,
|
| 828 |
-
"learning_rate": 2.49e-05,
|
| 829 |
-
"loss": 0.0093,
|
| 830 |
-
"r_loss": 0.008240088820457458,
|
| 831 |
-
"s_loss": 0.0012537827715277672,
|
| 832 |
-
"step": 8300
|
| 833 |
-
},
|
| 834 |
-
{
|
| 835 |
-
"epoch": 0.25,
|
| 836 |
-
"learning_rate": 2.52e-05,
|
| 837 |
-
"loss": 0.0096,
|
| 838 |
-
"r_loss": 0.008193010464310646,
|
| 839 |
-
"s_loss": 0.0012971541145816445,
|
| 840 |
-
"step": 8400
|
| 841 |
-
},
|
| 842 |
-
{
|
| 843 |
-
"epoch": 0.25,
|
| 844 |
-
"learning_rate": 2.55e-05,
|
| 845 |
-
"loss": 0.0098,
|
| 846 |
-
"r_loss": 0.007697759196162224,
|
| 847 |
-
"s_loss": 0.0012005027383565903,
|
| 848 |
-
"step": 8500
|
| 849 |
-
},
|
| 850 |
-
{
|
| 851 |
-
"epoch": 0.25,
|
| 852 |
-
"eval_loss": 0.010029895231127739,
|
| 853 |
-
"eval_r_loss": 0.008869171142578125,
|
| 854 |
-
"eval_runtime": 5.779,
|
| 855 |
-
"eval_s_loss": 0.0011607238557189703,
|
| 856 |
-
"eval_samples_per_second": 116.282,
|
| 857 |
-
"eval_steps_per_second": 116.282,
|
| 858 |
-
"step": 8500
|
| 859 |
-
},
|
| 860 |
-
{
|
| 861 |
-
"epoch": 0.26,
|
| 862 |
-
"learning_rate": 2.58e-05,
|
| 863 |
-
"loss": 0.0094,
|
| 864 |
-
"r_loss": 0.00764896534383297,
|
| 865 |
-
"s_loss": 0.0012557113077491522,
|
| 866 |
-
"step": 8600
|
| 867 |
-
},
|
| 868 |
-
{
|
| 869 |
-
"epoch": 0.26,
|
| 870 |
-
"learning_rate": 2.61e-05,
|
| 871 |
-
"loss": 0.0096,
|
| 872 |
-
"r_loss": 0.007685758639127016,
|
| 873 |
-
"s_loss": 0.0011759819462895393,
|
| 874 |
-
"step": 8700
|
| 875 |
-
},
|
| 876 |
-
{
|
| 877 |
-
"epoch": 0.26,
|
| 878 |
-
"learning_rate": 2.64e-05,
|
| 879 |
-
"loss": 0.0095,
|
| 880 |
-
"r_loss": 0.008306157775223255,
|
| 881 |
-
"s_loss": 0.0011866830755025148,
|
| 882 |
-
"step": 8800
|
| 883 |
-
},
|
| 884 |
-
{
|
| 885 |
-
"epoch": 0.27,
|
| 886 |
-
"learning_rate": 2.6700000000000002e-05,
|
| 887 |
-
"loss": 0.0095,
|
| 888 |
-
"r_loss": 0.008763814345002174,
|
| 889 |
-
"s_loss": 0.0012581332121044397,
|
| 890 |
-
"step": 8900
|
| 891 |
-
},
|
| 892 |
-
{
|
| 893 |
-
"epoch": 0.27,
|
| 894 |
-
"learning_rate": 2.7000000000000002e-05,
|
| 895 |
-
"loss": 0.0094,
|
| 896 |
-
"r_loss": 0.007176906801760197,
|
| 897 |
-
"s_loss": 0.0012848544865846634,
|
| 898 |
-
"step": 9000
|
| 899 |
-
},
|
| 900 |
-
{
|
| 901 |
-
"epoch": 0.27,
|
| 902 |
-
"eval_loss": 0.009547967463731766,
|
| 903 |
-
"eval_r_loss": 0.008400797843933105,
|
| 904 |
-
"eval_runtime": 5.8725,
|
| 905 |
-
"eval_s_loss": 0.0011471696197986603,
|
| 906 |
-
"eval_samples_per_second": 114.432,
|
| 907 |
-
"eval_steps_per_second": 114.432,
|
| 908 |
-
"step": 9000
|
| 909 |
-
},
|
| 910 |
-
{
|
| 911 |
-
"epoch": 0.27,
|
| 912 |
-
"learning_rate": 2.7300000000000003e-05,
|
| 913 |
-
"loss": 0.0091,
|
| 914 |
-
"r_loss": 0.007933049462735653,
|
| 915 |
-
"s_loss": 0.0013153750915080309,
|
| 916 |
-
"step": 9100
|
| 917 |
-
},
|
| 918 |
-
{
|
| 919 |
-
"epoch": 0.27,
|
| 920 |
-
"learning_rate": 2.7600000000000003e-05,
|
| 921 |
-
"loss": 0.0093,
|
| 922 |
-
"r_loss": 0.008543523028492928,
|
| 923 |
-
"s_loss": 0.0012904139002785087,
|
| 924 |
-
"step": 9200
|
| 925 |
-
},
|
| 926 |
-
{
|
| 927 |
-
"epoch": 0.28,
|
| 928 |
-
"learning_rate": 2.79e-05,
|
| 929 |
-
"loss": 0.0093,
|
| 930 |
-
"r_loss": 0.007229233160614967,
|
| 931 |
-
"s_loss": 0.0011527976021170616,
|
| 932 |
-
"step": 9300
|
| 933 |
-
},
|
| 934 |
-
{
|
| 935 |
-
"epoch": 0.28,
|
| 936 |
-
"learning_rate": 2.8199999999999998e-05,
|
| 937 |
-
"loss": 0.0093,
|
| 938 |
-
"r_loss": 0.007913423702120781,
|
| 939 |
-
"s_loss": 0.0011675741989165545,
|
| 940 |
-
"step": 9400
|
| 941 |
-
},
|
| 942 |
-
{
|
| 943 |
-
"epoch": 0.28,
|
| 944 |
-
"learning_rate": 2.8499999999999998e-05,
|
| 945 |
-
"loss": 0.0092,
|
| 946 |
-
"r_loss": 0.007350780535489321,
|
| 947 |
-
"s_loss": 0.0012241221265867352,
|
| 948 |
-
"step": 9500
|
| 949 |
-
},
|
| 950 |
-
{
|
| 951 |
-
"epoch": 0.28,
|
| 952 |
-
"eval_loss": 0.009159870445728302,
|
| 953 |
-
"eval_r_loss": 0.008046709932386875,
|
| 954 |
-
"eval_runtime": 5.7652,
|
| 955 |
-
"eval_s_loss": 0.0011131602805107832,
|
| 956 |
-
"eval_samples_per_second": 116.562,
|
| 957 |
-
"eval_steps_per_second": 116.562,
|
| 958 |
-
"step": 9500
|
| 959 |
-
},
|
| 960 |
-
{
|
| 961 |
-
"epoch": 0.29,
|
| 962 |
-
"learning_rate": 2.88e-05,
|
| 963 |
-
"loss": 0.0092,
|
| 964 |
-
"r_loss": 0.007505511865019798,
|
| 965 |
-
"s_loss": 0.0012209609849378467,
|
| 966 |
-
"step": 9600
|
| 967 |
-
},
|
| 968 |
-
{
|
| 969 |
-
"epoch": 0.29,
|
| 970 |
-
"learning_rate": 2.91e-05,
|
| 971 |
-
"loss": 0.0092,
|
| 972 |
-
"r_loss": 0.00815967470407486,
|
| 973 |
-
"s_loss": 0.0011755439918488264,
|
| 974 |
-
"step": 9700
|
| 975 |
-
},
|
| 976 |
-
{
|
| 977 |
-
"epoch": 0.29,
|
| 978 |
-
"learning_rate": 2.94e-05,
|
| 979 |
-
"loss": 0.0092,
|
| 980 |
-
"r_loss": 0.007525102701038122,
|
| 981 |
-
"s_loss": 0.0012379743857309222,
|
| 982 |
-
"step": 9800
|
| 983 |
-
},
|
| 984 |
-
{
|
| 985 |
-
"epoch": 0.3,
|
| 986 |
-
"learning_rate": 2.97e-05,
|
| 987 |
-
"loss": 0.009,
|
| 988 |
-
"r_loss": 0.008029448799788952,
|
| 989 |
-
"s_loss": 0.0012208997504785657,
|
| 990 |
-
"step": 9900
|
| 991 |
-
},
|
| 992 |
-
{
|
| 993 |
-
"epoch": 0.3,
|
| 994 |
-
"learning_rate": 3e-05,
|
| 995 |
-
"loss": 0.0091,
|
| 996 |
-
"r_loss": 0.007686637807637453,
|
| 997 |
-
"s_loss": 0.0012359985848888755,
|
| 998 |
-
"step": 10000
|
| 999 |
-
},
|
| 1000 |
-
{
|
| 1001 |
-
"epoch": 0.3,
|
| 1002 |
-
"eval_loss": 0.009688720107078552,
|
| 1003 |
-
"eval_r_loss": 0.008566668257117271,
|
| 1004 |
-
"eval_runtime": 5.8537,
|
| 1005 |
-
"eval_s_loss": 0.0011220521992072463,
|
| 1006 |
-
"eval_samples_per_second": 114.8,
|
| 1007 |
-
"eval_steps_per_second": 114.8,
|
| 1008 |
-
"step": 10000
|
| 1009 |
-
},
|
| 1010 |
-
{
|
| 1011 |
-
"epoch": 0.3,
|
| 1012 |
-
"learning_rate": 2.9998687772974054e-05,
|
| 1013 |
-
"loss": 0.0091,
|
| 1014 |
-
"r_loss": 0.007961354218423367,
|
| 1015 |
-
"s_loss": 0.0010860951151698828,
|
| 1016 |
-
"step": 10100
|
| 1017 |
-
},
|
| 1018 |
-
{
|
| 1019 |
-
"epoch": 0.3,
|
| 1020 |
-
"learning_rate": 2.9994698173783606e-05,
|
| 1021 |
-
"loss": 0.0091,
|
| 1022 |
-
"r_loss": 0.008185433223843575,
|
| 1023 |
-
"s_loss": 0.0011867923894897103,
|
| 1024 |
-
"step": 10200
|
| 1025 |
-
},
|
| 1026 |
-
{
|
| 1027 |
-
"epoch": 0.31,
|
| 1028 |
-
"learning_rate": 2.998803178074702e-05,
|
| 1029 |
-
"loss": 0.0092,
|
| 1030 |
-
"r_loss": 0.008926862850785255,
|
| 1031 |
-
"s_loss": 0.0011962838470935822,
|
| 1032 |
-
"step": 10300
|
| 1033 |
-
},
|
| 1034 |
-
{
|
| 1035 |
-
"epoch": 0.31,
|
| 1036 |
-
"learning_rate": 2.997868978392226e-05,
|
| 1037 |
-
"loss": 0.009,
|
| 1038 |
-
"r_loss": 0.008092904463410378,
|
| 1039 |
-
"s_loss": 0.0011486653238534927,
|
| 1040 |
-
"step": 10400
|
| 1041 |
-
},
|
| 1042 |
-
{
|
| 1043 |
-
"epoch": 0.31,
|
| 1044 |
-
"learning_rate": 2.996667385100541e-05,
|
| 1045 |
-
"loss": 0.0091,
|
| 1046 |
-
"r_loss": 0.007265549618750811,
|
| 1047 |
-
"s_loss": 0.0010849842801690102,
|
| 1048 |
-
"step": 10500
|
| 1049 |
-
},
|
| 1050 |
-
{
|
| 1051 |
-
"epoch": 0.31,
|
| 1052 |
-
"eval_loss": 0.0097579974681139,
|
| 1053 |
-
"eval_r_loss": 0.008664416149258614,
|
| 1054 |
-
"eval_runtime": 5.8205,
|
| 1055 |
-
"eval_s_loss": 0.0010935813188552856,
|
| 1056 |
-
"eval_samples_per_second": 115.453,
|
| 1057 |
-
"eval_steps_per_second": 115.453,
|
| 1058 |
-
"step": 10500
|
| 1059 |
-
},
|
| 1060 |
-
{
|
| 1061 |
-
"epoch": 0.32,
|
| 1062 |
-
"learning_rate": 2.995198612703301e-05,
|
| 1063 |
-
"loss": 0.0091,
|
| 1064 |
-
"r_loss": 0.007716472260653973,
|
| 1065 |
-
"s_loss": 0.0011130705242976546,
|
| 1066 |
-
"step": 10600
|
| 1067 |
-
},
|
| 1068 |
-
{
|
| 1069 |
-
"epoch": 0.32,
|
| 1070 |
-
"learning_rate": 2.9934629233999088e-05,
|
| 1071 |
-
"loss": 0.0088,
|
| 1072 |
-
"r_loss": 0.0071815671399235725,
|
| 1073 |
-
"s_loss": 0.0010680295526981354,
|
| 1074 |
-
"step": 10700
|
| 1075 |
-
},
|
| 1076 |
-
{
|
| 1077 |
-
"epoch": 0.32,
|
| 1078 |
-
"learning_rate": 2.991460627038711e-05,
|
| 1079 |
-
"loss": 0.0089,
|
| 1080 |
-
"r_loss": 0.008051994256675243,
|
| 1081 |
-
"s_loss": 0.0011568560730665922,
|
| 1082 |
-
"step": 10800
|
| 1083 |
-
},
|
| 1084 |
-
{
|
| 1085 |
-
"epoch": 0.33,
|
| 1086 |
-
"learning_rate": 2.9891920810616865e-05,
|
| 1087 |
-
"loss": 0.0089,
|
| 1088 |
-
"r_loss": 0.007709968835115433,
|
| 1089 |
-
"s_loss": 0.0010994401527568698,
|
| 1090 |
-
"step": 10900
|
| 1091 |
-
},
|
| 1092 |
-
{
|
| 1093 |
-
"epoch": 0.33,
|
| 1094 |
-
"learning_rate": 2.986657690440635e-05,
|
| 1095 |
-
"loss": 0.0087,
|
| 1096 |
-
"r_loss": 0.007149288430809975,
|
| 1097 |
-
"s_loss": 0.0010991152375936508,
|
| 1098 |
-
"step": 11000
|
| 1099 |
-
},
|
| 1100 |
-
{
|
| 1101 |
-
"epoch": 0.33,
|
| 1102 |
-
"eval_loss": 0.008960756473243237,
|
| 1103 |
-
"eval_r_loss": 0.0079101687297225,
|
| 1104 |
-
"eval_runtime": 5.8735,
|
| 1105 |
-
"eval_s_loss": 0.001050587510690093,
|
| 1106 |
-
"eval_samples_per_second": 114.412,
|
| 1107 |
-
"eval_steps_per_second": 114.412,
|
| 1108 |
-
"step": 11000
|
| 1109 |
-
},
|
| 1110 |
-
{
|
| 1111 |
-
"epoch": 0.33,
|
| 1112 |
-
"learning_rate": 2.983857907604885e-05,
|
| 1113 |
-
"loss": 0.0087,
|
| 1114 |
-
"r_loss": 0.006731455214321613,
|
| 1115 |
-
"s_loss": 0.0011426934506744146,
|
| 1116 |
-
"step": 11100
|
| 1117 |
-
},
|
| 1118 |
-
{
|
| 1119 |
-
"epoch": 0.33,
|
| 1120 |
-
"learning_rate": 2.9807932323605262e-05,
|
| 1121 |
-
"loss": 0.009,
|
| 1122 |
-
"r_loss": 0.008175029419362545,
|
| 1123 |
-
"s_loss": 0.0010848107049241662,
|
| 1124 |
-
"step": 11200
|
| 1125 |
-
},
|
| 1126 |
-
{
|
| 1127 |
-
"epoch": 0.34,
|
| 1128 |
-
"learning_rate": 2.977464211801187e-05,
|
| 1129 |
-
"loss": 0.0085,
|
| 1130 |
-
"r_loss": 0.007222716696560383,
|
| 1131 |
-
"s_loss": 0.0010968766873702407,
|
| 1132 |
-
"step": 11300
|
| 1133 |
-
},
|
| 1134 |
-
{
|
| 1135 |
-
"epoch": 0.34,
|
| 1136 |
-
"learning_rate": 2.9738714402103696e-05,
|
| 1137 |
-
"loss": 0.0086,
|
| 1138 |
-
"r_loss": 0.007525290362536907,
|
| 1139 |
-
"s_loss": 0.0010664989240467548,
|
| 1140 |
-
"step": 11400
|
| 1141 |
-
},
|
| 1142 |
-
{
|
| 1143 |
-
"epoch": 0.34,
|
| 1144 |
-
"learning_rate": 2.9700155589553614e-05,
|
| 1145 |
-
"loss": 0.0085,
|
| 1146 |
-
"r_loss": 0.00836112443357706,
|
| 1147 |
-
"s_loss": 0.0012076541315764189,
|
| 1148 |
-
"step": 11500
|
| 1149 |
-
},
|
| 1150 |
-
{
|
| 1151 |
-
"epoch": 0.34,
|
| 1152 |
-
"eval_loss": 0.008930221199989319,
|
| 1153 |
-
"eval_r_loss": 0.007907573133707047,
|
| 1154 |
-
"eval_runtime": 5.8343,
|
| 1155 |
-
"eval_s_loss": 0.0010226481826975942,
|
| 1156 |
-
"eval_samples_per_second": 115.18,
|
| 1157 |
-
"eval_steps_per_second": 115.18,
|
| 1158 |
-
"step": 11500
|
| 1159 |
-
},
|
| 1160 |
-
{
|
| 1161 |
-
"epoch": 0.35,
|
| 1162 |
-
"learning_rate": 2.9658972563727394e-05,
|
| 1163 |
-
"loss": 0.0087,
|
| 1164 |
-
"r_loss": 0.007679732982069254,
|
| 1165 |
-
"s_loss": 0.000998866162262857,
|
| 1166 |
-
"step": 11600
|
| 1167 |
-
},
|
| 1168 |
-
{
|
| 1169 |
-
"epoch": 0.35,
|
| 1170 |
-
"learning_rate": 2.9615172676454915e-05,
|
| 1171 |
-
"loss": 0.0085,
|
| 1172 |
-
"r_loss": 0.007193954661488533,
|
| 1173 |
-
"s_loss": 0.00100530288182199,
|
| 1174 |
-
"step": 11700
|
| 1175 |
-
},
|
| 1176 |
-
{
|
| 1177 |
-
"epoch": 0.35,
|
| 1178 |
-
"learning_rate": 2.956876374671775e-05,
|
| 1179 |
-
"loss": 0.0086,
|
| 1180 |
-
"r_loss": 0.008202124387025833,
|
| 1181 |
-
"s_loss": 0.001100401277653873,
|
| 1182 |
-
"step": 11800
|
| 1183 |
-
},
|
| 1184 |
-
{
|
| 1185 |
-
"epoch": 0.36,
|
| 1186 |
-
"learning_rate": 2.9519754059253352e-05,
|
| 1187 |
-
"loss": 0.0086,
|
| 1188 |
-
"r_loss": 0.005764150992035866,
|
| 1189 |
-
"s_loss": 0.001019550021737814,
|
| 1190 |
-
"step": 11900
|
| 1191 |
-
},
|
| 1192 |
-
{
|
| 1193 |
-
"epoch": 0.36,
|
| 1194 |
-
"learning_rate": 2.946815236307609e-05,
|
| 1195 |
-
"loss": 0.0088,
|
| 1196 |
-
"r_loss": 0.006989379413425922,
|
| 1197 |
-
"s_loss": 0.0010334283579140902,
|
| 1198 |
-
"step": 12000
|
| 1199 |
-
},
|
| 1200 |
-
{
|
| 1201 |
-
"epoch": 0.36,
|
| 1202 |
-
"eval_loss": 0.008558472618460655,
|
| 1203 |
-
"eval_r_loss": 0.007531056646257639,
|
| 1204 |
-
"eval_runtime": 5.8831,
|
| 1205 |
-
"eval_s_loss": 0.0010274164378643036,
|
| 1206 |
-
"eval_samples_per_second": 114.225,
|
| 1207 |
-
"eval_steps_per_second": 114.225,
|
| 1208 |
-
"step": 12000
|
| 1209 |
-
},
|
| 1210 |
-
{
|
| 1211 |
-
"epoch": 0.36,
|
| 1212 |
-
"learning_rate": 2.941396786991542e-05,
|
| 1213 |
-
"loss": 0.0084,
|
| 1214 |
-
"r_loss": 0.007328535430133343,
|
| 1215 |
-
"s_loss": 0.0010166720021516085,
|
| 1216 |
-
"step": 12100
|
| 1217 |
-
},
|
| 1218 |
-
{
|
| 1219 |
-
"epoch": 0.36,
|
| 1220 |
-
"learning_rate": 2.9357210252571423e-05,
|
| 1221 |
-
"loss": 0.0084,
|
| 1222 |
-
"r_loss": 0.006314602214843035,
|
| 1223 |
-
"s_loss": 0.0010744825704023242,
|
| 1224 |
-
"step": 12200
|
| 1225 |
-
},
|
| 1226 |
-
{
|
| 1227 |
-
"epoch": 0.37,
|
| 1228 |
-
"learning_rate": 2.929788964318808e-05,
|
| 1229 |
-
"loss": 0.0084,
|
| 1230 |
-
"r_loss": 0.007112974300980568,
|
| 1231 |
-
"s_loss": 0.0010199513053521514,
|
| 1232 |
-
"step": 12300
|
| 1233 |
-
},
|
| 1234 |
-
{
|
| 1235 |
-
"epoch": 0.37,
|
| 1236 |
-
"learning_rate": 2.923601663144452e-05,
|
| 1237 |
-
"loss": 0.0083,
|
| 1238 |
-
"r_loss": 0.0075445109978318214,
|
| 1239 |
-
"s_loss": 0.001125972718000412,
|
| 1240 |
-
"step": 12400
|
| 1241 |
-
},
|
| 1242 |
-
{
|
| 1243 |
-
"epoch": 0.37,
|
| 1244 |
-
"learning_rate": 2.9171602262664564e-05,
|
| 1245 |
-
"loss": 0.0082,
|
| 1246 |
-
"r_loss": 0.006783840246498585,
|
| 1247 |
-
"s_loss": 0.0010000347392633557,
|
| 1248 |
-
"step": 12500
|
| 1249 |
-
},
|
| 1250 |
-
{
|
| 1251 |
-
"epoch": 0.37,
|
| 1252 |
-
"eval_loss": 0.008438749238848686,
|
| 1253 |
-
"eval_r_loss": 0.007456656079739332,
|
| 1254 |
-
"eval_runtime": 6.6399,
|
| 1255 |
-
"eval_s_loss": 0.0009820933919399977,
|
| 1256 |
-
"eval_samples_per_second": 101.206,
|
| 1257 |
-
"eval_steps_per_second": 101.206,
|
| 1258 |
-
"step": 12500
|
| 1259 |
-
},
|
| 1260 |
-
{
|
| 1261 |
-
"epoch": 0.38,
|
| 1262 |
-
"learning_rate": 2.9104658035844992e-05,
|
| 1263 |
-
"loss": 0.0084,
|
| 1264 |
-
"r_loss": 0.008811071515083313,
|
| 1265 |
-
"s_loss": 0.0010695005767047405,
|
| 1266 |
-
"step": 12600
|
| 1267 |
-
},
|
| 1268 |
-
{
|
| 1269 |
-
"epoch": 0.38,
|
| 1270 |
-
"learning_rate": 2.9035195901602766e-05,
|
| 1271 |
-
"loss": 0.0081,
|
| 1272 |
-
"r_loss": 0.00737602636218071,
|
| 1273 |
-
"s_loss": 0.001059257541783154,
|
| 1274 |
-
"step": 12700
|
| 1275 |
-
},
|
| 1276 |
-
{
|
| 1277 |
-
"epoch": 0.38,
|
| 1278 |
-
"learning_rate": 2.896322826004167e-05,
|
| 1279 |
-
"loss": 0.0083,
|
| 1280 |
-
"r_loss": 0.006479831412434578,
|
| 1281 |
-
"s_loss": 0.00109660136513412,
|
| 1282 |
-
"step": 12800
|
| 1283 |
-
},
|
| 1284 |
-
{
|
| 1285 |
-
"epoch": 0.38,
|
| 1286 |
-
"learning_rate": 2.8888767958538672e-05,
|
| 1287 |
-
"loss": 0.0081,
|
| 1288 |
-
"r_loss": 0.006183864548802376,
|
| 1289 |
-
"s_loss": 0.000993464607745409,
|
| 1290 |
-
"step": 12900
|
| 1291 |
-
},
|
| 1292 |
-
{
|
| 1293 |
-
"epoch": 0.39,
|
| 1294 |
-
"learning_rate": 2.881182828945048e-05,
|
| 1295 |
-
"loss": 0.0082,
|
| 1296 |
-
"r_loss": 0.007777994964271784,
|
| 1297 |
-
"s_loss": 0.0010871184058487415,
|
| 1298 |
-
"step": 13000
|
| 1299 |
-
},
|
| 1300 |
-
{
|
| 1301 |
-
"epoch": 0.39,
|
| 1302 |
-
"eval_loss": 0.007983732037246227,
|
| 1303 |
-
"eval_r_loss": 0.007039351388812065,
|
| 1304 |
-
"eval_runtime": 5.8966,
|
| 1305 |
-
"eval_s_loss": 0.0009443803573958576,
|
| 1306 |
-
"eval_samples_per_second": 113.965,
|
| 1307 |
-
"eval_steps_per_second": 113.965,
|
| 1308 |
-
"step": 13000
|
| 1309 |
-
},
|
| 1310 |
-
{
|
| 1311 |
-
"epoch": 0.39,
|
| 1312 |
-
"learning_rate": 2.873242298774064e-05,
|
| 1313 |
-
"loss": 0.0082,
|
| 1314 |
-
"r_loss": 0.010438592173159122,
|
| 1315 |
-
"s_loss": 0.0010730213252827525,
|
| 1316 |
-
"step": 13100
|
| 1317 |
-
},
|
| 1318 |
-
{
|
| 1319 |
-
"epoch": 0.39,
|
| 1320 |
-
"learning_rate": 2.865056622852762e-05,
|
| 1321 |
-
"loss": 0.0081,
|
| 1322 |
-
"r_loss": 0.006384184584021568,
|
| 1323 |
-
"s_loss": 0.0008851269376464188,
|
| 1324 |
-
"step": 13200
|
| 1325 |
-
},
|
| 1326 |
-
{
|
| 1327 |
-
"epoch": 0.4,
|
| 1328 |
-
"learning_rate": 2.8566272624554314e-05,
|
| 1329 |
-
"loss": 0.0081,
|
| 1330 |
-
"r_loss": 0.007597366347908974,
|
| 1331 |
-
"s_loss": 0.001074151019565761,
|
| 1332 |
-
"step": 13300
|
| 1333 |
-
},
|
| 1334 |
-
{
|
| 1335 |
-
"epoch": 0.4,
|
| 1336 |
-
"learning_rate": 2.847955722357946e-05,
|
| 1337 |
-
"loss": 0.0081,
|
| 1338 |
-
"r_loss": 0.006059112958610058,
|
| 1339 |
-
"s_loss": 0.0010072380537167192,
|
| 1340 |
-
"step": 13400
|
| 1341 |
-
},
|
| 1342 |
-
{
|
| 1343 |
-
"epoch": 0.4,
|
| 1344 |
-
"learning_rate": 2.8390435505691352e-05,
|
| 1345 |
-
"loss": 0.008,
|
| 1346 |
-
"r_loss": 0.006421985570341349,
|
| 1347 |
-
"s_loss": 0.0009825531160458922,
|
| 1348 |
-
"step": 13500
|
| 1349 |
-
},
|
| 1350 |
-
{
|
| 1351 |
-
"epoch": 0.4,
|
| 1352 |
-
"eval_loss": 0.008024024777114391,
|
| 1353 |
-
"eval_r_loss": 0.007059826515614986,
|
| 1354 |
-
"eval_runtime": 5.9301,
|
| 1355 |
-
"eval_s_loss": 0.0009641979122534394,
|
| 1356 |
-
"eval_samples_per_second": 113.321,
|
| 1357 |
-
"eval_steps_per_second": 113.321,
|
| 1358 |
-
"step": 13500
|
| 1359 |
-
},
|
| 1360 |
-
{
|
| 1361 |
-
"epoch": 0.41,
|
| 1362 |
-
"learning_rate": 2.8298923380544406e-05,
|
| 1363 |
-
"loss": 0.008,
|
| 1364 |
-
"r_loss": 0.006581292487680912,
|
| 1365 |
-
"s_loss": 0.0010704277083277702,
|
| 1366 |
-
"step": 13600
|
| 1367 |
-
},
|
| 1368 |
-
{
|
| 1369 |
-
"epoch": 0.41,
|
| 1370 |
-
"learning_rate": 2.8205037184519026e-05,
|
| 1371 |
-
"loss": 0.008,
|
| 1372 |
-
"r_loss": 0.008079946041107178,
|
| 1373 |
-
"s_loss": 0.0009939800947904587,
|
| 1374 |
-
"step": 13700
|
| 1375 |
-
},
|
| 1376 |
-
{
|
| 1377 |
-
"epoch": 0.41,
|
| 1378 |
-
"learning_rate": 2.8108793677805307e-05,
|
| 1379 |
-
"loss": 0.0079,
|
| 1380 |
-
"r_loss": 0.005696495994925499,
|
| 1381 |
-
"s_loss": 0.0009651000145822763,
|
| 1382 |
-
"step": 13800
|
| 1383 |
-
},
|
| 1384 |
-
{
|
| 1385 |
-
"epoch": 0.41,
|
| 1386 |
-
"learning_rate": 2.8010210041411057e-05,
|
| 1387 |
-
"loss": 0.008,
|
| 1388 |
-
"r_loss": 0.006050314754247665,
|
| 1389 |
-
"s_loss": 0.0010551176965236664,
|
| 1390 |
-
"step": 13900
|
| 1391 |
-
},
|
| 1392 |
-
{
|
| 1393 |
-
"epoch": 0.42,
|
| 1394 |
-
"learning_rate": 2.7909303874094737e-05,
|
| 1395 |
-
"loss": 0.008,
|
| 1396 |
-
"r_loss": 0.007171120494604111,
|
| 1397 |
-
"s_loss": 0.001050697872415185,
|
| 1398 |
-
"step": 14000
|
| 1399 |
-
},
|
| 1400 |
-
{
|
| 1401 |
-
"epoch": 0.42,
|
| 1402 |
-
"eval_loss": 0.0087936632335186,
|
| 1403 |
-
"eval_r_loss": 0.007837551645934582,
|
| 1404 |
-
"eval_runtime": 5.8165,
|
| 1405 |
-
"eval_s_loss": 0.0009561114711686969,
|
| 1406 |
-
"eval_samples_per_second": 115.534,
|
| 1407 |
-
"eval_steps_per_second": 115.534,
|
| 1408 |
-
"step": 14000
|
| 1409 |
-
},
|
| 1410 |
-
{
|
| 1411 |
-
"epoch": 0.42,
|
| 1412 |
-
"learning_rate": 2.7806093189223774e-05,
|
| 1413 |
-
"loss": 0.0079,
|
| 1414 |
-
"r_loss": 0.007168681360781193,
|
| 1415 |
-
"s_loss": 0.0011435865890234709,
|
| 1416 |
-
"step": 14100
|
| 1417 |
-
},
|
| 1418 |
-
{
|
| 1419 |
-
"epoch": 0.42,
|
| 1420 |
-
"learning_rate": 2.7700596411558902e-05,
|
| 1421 |
-
"loss": 0.0078,
|
| 1422 |
-
"r_loss": 0.007676262408494949,
|
| 1423 |
-
"s_loss": 0.0010355566628277302,
|
| 1424 |
-
"step": 14200
|
| 1425 |
-
},
|
| 1426 |
-
{
|
| 1427 |
-
"epoch": 0.43,
|
| 1428 |
-
"learning_rate": 2.7592832373965038e-05,
|
| 1429 |
-
"loss": 0.008,
|
| 1430 |
-
"r_loss": 0.007139571011066437,
|
| 1431 |
-
"s_loss": 0.001015349174849689,
|
| 1432 |
-
"step": 14300
|
| 1433 |
-
},
|
| 1434 |
-
{
|
| 1435 |
-
"epoch": 0.43,
|
| 1436 |
-
"learning_rate": 2.7482820314049326e-05,
|
| 1437 |
-
"loss": 0.0079,
|
| 1438 |
-
"r_loss": 0.00660574808716774,
|
| 1439 |
-
"s_loss": 0.000994799891486764,
|
| 1440 |
-
"step": 14400
|
| 1441 |
-
},
|
| 1442 |
-
{
|
| 1443 |
-
"epoch": 0.43,
|
| 1444 |
-
"learning_rate": 2.7370579870726906e-05,
|
| 1445 |
-
"loss": 0.0078,
|
| 1446 |
-
"r_loss": 0.006421282887458801,
|
| 1447 |
-
"s_loss": 0.0009415894746780396,
|
| 1448 |
-
"step": 14500
|
| 1449 |
-
},
|
| 1450 |
-
{
|
| 1451 |
-
"epoch": 0.43,
|
| 1452 |
-
"eval_loss": 0.008639033883810043,
|
| 1453 |
-
"eval_r_loss": 0.007638509385287762,
|
| 1454 |
-
"eval_runtime": 5.8664,
|
| 1455 |
-
"eval_s_loss": 0.0010005244985222816,
|
| 1456 |
-
"eval_samples_per_second": 114.55,
|
| 1457 |
-
"eval_steps_per_second": 114.55,
|
| 1458 |
-
"step": 14500
|
| 1459 |
-
},
|
| 1460 |
-
{
|
| 1461 |
-
"epoch": 0.44,
|
| 1462 |
-
"learning_rate": 2.7256131080715053e-05,
|
| 1463 |
-
"loss": 0.0078,
|
| 1464 |
-
"r_loss": 0.006864764261990786,
|
| 1465 |
-
"s_loss": 0.0010434763971716166,
|
| 1466 |
-
"step": 14600
|
| 1467 |
-
},
|
| 1468 |
-
{
|
| 1469 |
-
"epoch": 0.44,
|
| 1470 |
-
"learning_rate": 2.7139494374956316e-05,
|
| 1471 |
-
"loss": 0.0078,
|
| 1472 |
-
"r_loss": 0.00723334401845932,
|
| 1473 |
-
"s_loss": 0.0010841034818440676,
|
| 1474 |
-
"step": 14700
|
| 1475 |
-
},
|
| 1476 |
-
{
|
| 1477 |
-
"epoch": 0.44,
|
| 1478 |
-
"learning_rate": 2.7020690574971236e-05,
|
| 1479 |
-
"loss": 0.0078,
|
| 1480 |
-
"r_loss": 0.006216096691787243,
|
| 1481 |
-
"s_loss": 0.0008721616468392313,
|
| 1482 |
-
"step": 14800
|
| 1483 |
-
},
|
| 1484 |
-
{
|
| 1485 |
-
"epoch": 0.44,
|
| 1486 |
-
"learning_rate": 2.6899740889141407e-05,
|
| 1487 |
-
"loss": 0.0077,
|
| 1488 |
-
"r_loss": 0.006137696094810963,
|
| 1489 |
-
"s_loss": 0.000969752436503768,
|
| 1490 |
-
"step": 14900
|
| 1491 |
-
},
|
| 1492 |
-
{
|
| 1493 |
-
"epoch": 0.45,
|
| 1494 |
-
"learning_rate": 2.677666690892343e-05,
|
| 1495 |
-
"loss": 0.0077,
|
| 1496 |
-
"r_loss": 0.006585408002138138,
|
| 1497 |
-
"s_loss": 0.0009770711185410619,
|
| 1498 |
-
"step": 15000
|
| 1499 |
-
},
|
| 1500 |
-
{
|
| 1501 |
-
"epoch": 0.45,
|
| 1502 |
-
"eval_loss": 0.008090370334684849,
|
| 1503 |
-
"eval_r_loss": 0.007137539330869913,
|
| 1504 |
-
"eval_runtime": 5.9249,
|
| 1505 |
-
"eval_s_loss": 0.0009528312948532403,
|
| 1506 |
-
"eval_samples_per_second": 113.42,
|
| 1507 |
-
"eval_steps_per_second": 113.42,
|
| 1508 |
-
"step": 15000
|
| 1509 |
-
},
|
| 1510 |
-
{
|
| 1511 |
-
"epoch": 0.45,
|
| 1512 |
-
"learning_rate": 2.6651490604994458e-05,
|
| 1513 |
-
"loss": 0.0076,
|
| 1514 |
-
"r_loss": 0.006115331780165434,
|
| 1515 |
-
"s_loss": 0.0010219502728432417,
|
| 1516 |
-
"step": 15100
|
| 1517 |
-
},
|
| 1518 |
-
{
|
| 1519 |
-
"epoch": 0.45,
|
| 1520 |
-
"learning_rate": 2.6524234323330147e-05,
|
| 1521 |
-
"loss": 0.0076,
|
| 1522 |
-
"r_loss": 0.006430969573557377,
|
| 1523 |
-
"s_loss": 0.0010282087605446577,
|
| 1524 |
-
"step": 15200
|
| 1525 |
-
},
|
| 1526 |
-
{
|
| 1527 |
-
"epoch": 0.46,
|
| 1528 |
-
"learning_rate": 2.6394920781215467e-05,
|
| 1529 |
-
"loss": 0.0076,
|
| 1530 |
-
"r_loss": 0.010935202240943909,
|
| 1531 |
-
"s_loss": 0.001126307644881308,
|
| 1532 |
-
"step": 15300
|
| 1533 |
-
},
|
| 1534 |
-
{
|
| 1535 |
-
"epoch": 0.46,
|
| 1536 |
-
"learning_rate": 2.6264896532848944e-05,
|
| 1537 |
-
"loss": 0.0076,
|
| 1538 |
-
"r_loss": 0.006224375218153,
|
| 1539 |
-
"s_loss": 0.0009270138689316809,
|
| 1540 |
-
"step": 15400
|
| 1541 |
-
},
|
| 1542 |
-
{
|
| 1543 |
-
"epoch": 0.46,
|
| 1544 |
-
"learning_rate": 2.6131558076617624e-05,
|
| 1545 |
-
"loss": 0.0076,
|
| 1546 |
-
"r_loss": 0.007503915578126907,
|
| 1547 |
-
"s_loss": 0.001000251155346632,
|
| 1548 |
-
"step": 15500
|
| 1549 |
-
},
|
| 1550 |
-
{
|
| 1551 |
-
"epoch": 0.46,
|
| 1552 |
-
"eval_loss": 0.007729613222181797,
|
| 1553 |
-
"eval_r_loss": 0.006781161762773991,
|
| 1554 |
-
"eval_runtime": 8.5719,
|
| 1555 |
-
"eval_s_loss": 0.0009484515758231282,
|
| 1556 |
-
"eval_samples_per_second": 78.396,
|
| 1557 |
-
"eval_steps_per_second": 78.396,
|
| 1558 |
-
"step": 15500
|
| 1559 |
-
},
|
| 1560 |
-
{
|
| 1561 |
-
"epoch": 0.47,
|
| 1562 |
-
"learning_rate": 2.5996232458936835e-05,
|
| 1563 |
-
"loss": 0.0077,
|
| 1564 |
-
"r_loss": 0.0065051475539803505,
|
| 1565 |
-
"s_loss": 0.0009143096976913512,
|
| 1566 |
-
"step": 15600
|
| 1567 |
-
},
|
| 1568 |
-
{
|
| 1569 |
-
"epoch": 0.47,
|
| 1570 |
-
"learning_rate": 2.5858943837597314e-05,
|
| 1571 |
-
"loss": 0.0075,
|
| 1572 |
-
"r_loss": 0.0064110783860087395,
|
| 1573 |
-
"s_loss": 0.0009684975957497954,
|
| 1574 |
-
"step": 15700
|
| 1575 |
-
},
|
| 1576 |
-
{
|
| 1577 |
-
"epoch": 0.47,
|
| 1578 |
-
"learning_rate": 2.5719716720817392e-05,
|
| 1579 |
-
"loss": 0.0074,
|
| 1580 |
-
"r_loss": 0.007918891496956348,
|
| 1581 |
-
"s_loss": 0.001000194693915546,
|
| 1582 |
-
"step": 15800
|
| 1583 |
-
},
|
| 1584 |
-
{
|
| 1585 |
-
"epoch": 0.47,
|
| 1586 |
-
"learning_rate": 2.5578575962867906e-05,
|
| 1587 |
-
"loss": 0.0075,
|
| 1588 |
-
"r_loss": 0.007117274217307568,
|
| 1589 |
-
"s_loss": 0.0009594152215868235,
|
| 1590 |
-
"step": 15900
|
| 1591 |
-
},
|
| 1592 |
-
{
|
| 1593 |
-
"epoch": 0.48,
|
| 1594 |
-
"learning_rate": 2.543554675963528e-05,
|
| 1595 |
-
"loss": 0.0075,
|
| 1596 |
-
"r_loss": 0.006600790657103062,
|
| 1597 |
-
"s_loss": 0.0009607726242393255,
|
| 1598 |
-
"step": 16000
|
| 1599 |
-
},
|
| 1600 |
-
{
|
| 1601 |
-
"epoch": 0.48,
|
| 1602 |
-
"eval_loss": 0.007644696161150932,
|
| 1603 |
-
"eval_r_loss": 0.006735973991453648,
|
| 1604 |
-
"eval_runtime": 8.488,
|
| 1605 |
-
"eval_s_loss": 0.0009087221696972847,
|
| 1606 |
-
"eval_samples_per_second": 79.171,
|
| 1607 |
-
"eval_steps_per_second": 79.171,
|
| 1608 |
-
"step": 16000
|
| 1609 |
-
},
|
| 1610 |
-
{
|
| 1611 |
-
"epoch": 0.48,
|
| 1612 |
-
"learning_rate": 2.5290654644123703e-05,
|
| 1613 |
-
"loss": 0.0075,
|
| 1614 |
-
"r_loss": 0.005946667864918709,
|
| 1615 |
-
"s_loss": 0.0009118504240177572,
|
| 1616 |
-
"step": 16100
|
| 1617 |
-
},
|
| 1618 |
-
{
|
| 1619 |
-
"epoch": 0.48,
|
| 1620 |
-
"learning_rate": 2.5143925481897017e-05,
|
| 1621 |
-
"loss": 0.0076,
|
| 1622 |
-
"r_loss": 0.005937603302299976,
|
| 1623 |
-
"s_loss": 0.0009287346038036048,
|
| 1624 |
-
"step": 16200
|
| 1625 |
-
},
|
| 1626 |
-
{
|
| 1627 |
-
"epoch": 0.49,
|
| 1628 |
-
"learning_rate": 2.499538546646136e-05,
|
| 1629 |
-
"loss": 0.0073,
|
| 1630 |
-
"r_loss": 0.0064862994477152824,
|
| 1631 |
-
"s_loss": 0.0008813185268081725,
|
| 1632 |
-
"step": 16300
|
| 1633 |
-
},
|
| 1634 |
-
{
|
| 1635 |
-
"epoch": 0.49,
|
| 1636 |
-
"learning_rate": 2.4845061114589165e-05,
|
| 1637 |
-
"loss": 0.0074,
|
| 1638 |
-
"r_loss": 0.006060485728085041,
|
| 1639 |
-
"s_loss": 0.0009768878808245063,
|
| 1640 |
-
"step": 16400
|
| 1641 |
-
},
|
| 1642 |
-
{
|
| 1643 |
-
"epoch": 0.49,
|
| 1644 |
-
"learning_rate": 2.4692979261585507e-05,
|
| 1645 |
-
"loss": 0.0074,
|
| 1646 |
-
"r_loss": 0.005972530692815781,
|
| 1647 |
-
"s_loss": 0.0009232640732079744,
|
| 1648 |
-
"step": 16500
|
| 1649 |
-
},
|
| 1650 |
-
{
|
| 1651 |
-
"epoch": 0.49,
|
| 1652 |
-
"eval_loss": 0.007525917608290911,
|
| 1653 |
-
"eval_r_loss": 0.006600276567041874,
|
| 1654 |
-
"eval_runtime": 8.5748,
|
| 1655 |
-
"eval_s_loss": 0.000925640866626054,
|
| 1656 |
-
"eval_samples_per_second": 78.369,
|
| 1657 |
-
"eval_steps_per_second": 78.369,
|
| 1658 |
-
"step": 16500
|
| 1659 |
-
},
|
| 1660 |
-
{
|
| 1661 |
-
"epoch": 0.5,
|
| 1662 |
-
"learning_rate": 2.4539167056497572e-05,
|
| 1663 |
-
"loss": 0.0075,
|
| 1664 |
-
"r_loss": 0.006335953716188669,
|
| 1665 |
-
"s_loss": 0.0009909672662615776,
|
| 1666 |
-
"step": 16600
|
| 1667 |
-
},
|
| 1668 |
-
{
|
| 1669 |
-
"epoch": 0.5,
|
| 1670 |
-
"learning_rate": 2.4383651957268106e-05,
|
| 1671 |
-
"loss": 0.0074,
|
| 1672 |
-
"r_loss": 0.0072190104983747005,
|
| 1673 |
-
"s_loss": 0.0009832193609327078,
|
| 1674 |
-
"step": 16700
|
| 1675 |
-
},
|
| 1676 |
-
{
|
| 1677 |
-
"epoch": 0.5,
|
| 1678 |
-
"learning_rate": 2.4226461725833757e-05,
|
| 1679 |
-
"loss": 0.0075,
|
| 1680 |
-
"r_loss": 0.0063974312506616116,
|
| 1681 |
-
"s_loss": 0.0009214280871674418,
|
| 1682 |
-
"step": 16800
|
| 1683 |
-
},
|
| 1684 |
-
{
|
| 1685 |
-
"epoch": 0.5,
|
| 1686 |
-
"learning_rate": 2.4067624423169087e-05,
|
| 1687 |
-
"loss": 0.0074,
|
| 1688 |
-
"r_loss": 0.006921318359673023,
|
| 1689 |
-
"s_loss": 0.000975217146333307,
|
| 1690 |
-
"step": 16900
|
| 1691 |
-
},
|
| 1692 |
-
{
|
| 1693 |
-
"epoch": 0.51,
|
| 1694 |
-
"learning_rate": 2.3907168404277275e-05,
|
| 1695 |
-
"loss": 0.0072,
|
| 1696 |
-
"r_loss": 0.005746101029217243,
|
| 1697 |
-
"s_loss": 0.0009283066028729081,
|
| 1698 |
-
"step": 17000
|
| 1699 |
-
},
|
| 1700 |
-
{
|
| 1701 |
-
"epoch": 0.51,
|
| 1702 |
-
"eval_loss": 0.006981786340475082,
|
| 1703 |
-
"eval_r_loss": 0.006118214689195156,
|
| 1704 |
-
"eval_runtime": 9.4285,
|
| 1705 |
-
"eval_s_loss": 0.0008635715930722654,
|
| 1706 |
-
"eval_samples_per_second": 71.273,
|
| 1707 |
-
"eval_steps_per_second": 71.273,
|
| 1708 |
-
"step": 17000
|
| 1709 |
-
},
|
| 1710 |
-
{
|
| 1711 |
-
"epoch": 0.51,
|
| 1712 |
-
"learning_rate": 2.3745122313128274e-05,
|
| 1713 |
-
"loss": 0.0072,
|
| 1714 |
-
"r_loss": 0.005779191851615906,
|
| 1715 |
-
"s_loss": 0.0009009677451103926,
|
| 1716 |
-
"step": 17100
|
| 1717 |
-
},
|
| 1718 |
-
{
|
| 1719 |
-
"epoch": 0.51,
|
| 1720 |
-
"learning_rate": 2.3581515077545418e-05,
|
| 1721 |
-
"loss": 0.0072,
|
| 1722 |
-
"r_loss": 0.007839716970920563,
|
| 1723 |
-
"s_loss": 0.0010175108909606934,
|
| 1724 |
-
"step": 17200
|
| 1725 |
-
},
|
| 1726 |
-
{
|
| 1727 |
-
"epoch": 0.52,
|
| 1728 |
-
"learning_rate": 2.34163759040413e-05,
|
| 1729 |
-
"loss": 0.0074,
|
| 1730 |
-
"r_loss": 0.006432846188545227,
|
| 1731 |
-
"s_loss": 0.001021060859784484,
|
| 1732 |
-
"step": 17300
|
| 1733 |
-
},
|
| 1734 |
-
{
|
| 1735 |
-
"epoch": 0.52,
|
| 1736 |
-
"learning_rate": 2.324973427260402e-05,
|
| 1737 |
-
"loss": 0.0073,
|
| 1738 |
-
"r_loss": 0.0062257954850792885,
|
| 1739 |
-
"s_loss": 0.0009748931624926627,
|
| 1740 |
-
"step": 17400
|
| 1741 |
-
},
|
| 1742 |
-
{
|
| 1743 |
-
"epoch": 0.52,
|
| 1744 |
-
"learning_rate": 2.3081619931434452e-05,
|
| 1745 |
-
"loss": 0.0072,
|
| 1746 |
-
"r_loss": 0.006158421281725168,
|
| 1747 |
-
"s_loss": 0.0010701502906158566,
|
| 1748 |
-
"step": 17500
|
| 1749 |
-
},
|
| 1750 |
-
{
|
| 1751 |
-
"epoch": 0.52,
|
| 1752 |
-
"eval_loss": 0.007478422485291958,
|
| 1753 |
-
"eval_r_loss": 0.006587797310203314,
|
| 1754 |
-
"eval_runtime": 10.1462,
|
| 1755 |
-
"eval_s_loss": 0.0008906250586733222,
|
| 1756 |
-
"eval_samples_per_second": 66.232,
|
| 1757 |
-
"eval_steps_per_second": 66.232,
|
| 1758 |
-
"step": 17500
|
| 1759 |
-
},
|
| 1760 |
-
{
|
| 1761 |
-
"epoch": 0.53,
|
| 1762 |
-
"learning_rate": 2.2912062891635778e-05,
|
| 1763 |
-
"loss": 0.0072,
|
| 1764 |
-
"r_loss": 0.006115030962973833,
|
| 1765 |
-
"s_loss": 0.0009346662554889917,
|
| 1766 |
-
"step": 17600
|
| 1767 |
-
},
|
| 1768 |
-
{
|
| 1769 |
-
"epoch": 0.53,
|
| 1770 |
-
"learning_rate": 2.274109342185598e-05,
|
| 1771 |
-
"loss": 0.0074,
|
| 1772 |
-
"r_loss": 0.00592977087944746,
|
| 1773 |
-
"s_loss": 0.000866447517182678,
|
| 1774 |
-
"step": 17700
|
| 1775 |
-
},
|
| 1776 |
-
{
|
| 1777 |
-
"epoch": 0.53,
|
| 1778 |
-
"learning_rate": 2.256874204288442e-05,
|
| 1779 |
-
"loss": 0.0071,
|
| 1780 |
-
"r_loss": 0.006099973805248737,
|
| 1781 |
-
"s_loss": 0.0008897872176021338,
|
| 1782 |
-
"step": 17800
|
| 1783 |
-
},
|
| 1784 |
-
{
|
| 1785 |
-
"epoch": 0.53,
|
| 1786 |
-
"learning_rate": 2.2395039522203403e-05,
|
| 1787 |
-
"loss": 0.0073,
|
| 1788 |
-
"r_loss": 0.006193576380610466,
|
| 1789 |
-
"s_loss": 0.0008475868962705135,
|
| 1790 |
-
"step": 17900
|
| 1791 |
-
},
|
| 1792 |
-
{
|
| 1793 |
-
"epoch": 0.54,
|
| 1794 |
-
"learning_rate": 2.222001686849566e-05,
|
| 1795 |
-
"loss": 0.0071,
|
| 1796 |
-
"r_loss": 0.0067910789512097836,
|
| 1797 |
-
"s_loss": 0.0010297299595549703,
|
| 1798 |
-
"step": 18000
|
| 1799 |
-
},
|
| 1800 |
-
{
|
| 1801 |
-
"epoch": 0.54,
|
| 1802 |
-
"eval_loss": 0.007175224833190441,
|
| 1803 |
-
"eval_r_loss": 0.006267193704843521,
|
| 1804 |
-
"eval_runtime": 8.8162,
|
| 1805 |
-
"eval_s_loss": 0.0009080312447622418,
|
| 1806 |
-
"eval_samples_per_second": 76.223,
|
| 1807 |
-
"eval_steps_per_second": 76.223,
|
| 1808 |
-
"step": 18000
|
| 1809 |
-
},
|
| 1810 |
-
{
|
| 1811 |
-
"epoch": 0.54,
|
| 1812 |
-
"learning_rate": 2.2043705326108824e-05,
|
| 1813 |
-
"loss": 0.0072,
|
| 1814 |
-
"r_loss": 0.006607139483094215,
|
| 1815 |
-
"s_loss": 0.0009428428602404892,
|
| 1816 |
-
"step": 18100
|
| 1817 |
-
},
|
| 1818 |
-
{
|
| 1819 |
-
"epoch": 0.54,
|
| 1820 |
-
"learning_rate": 2.1866136369477807e-05,
|
| 1821 |
-
"loss": 0.0071,
|
| 1822 |
-
"r_loss": 0.0053857965394854546,
|
| 1823 |
-
"s_loss": 0.0008554364321753383,
|
| 1824 |
-
"step": 18200
|
| 1825 |
-
},
|
| 1826 |
-
{
|
| 1827 |
-
"epoch": 0.55,
|
| 1828 |
-
"learning_rate": 2.1687341697506106e-05,
|
| 1829 |
-
"loss": 0.0071,
|
| 1830 |
-
"r_loss": 0.005985291674733162,
|
| 1831 |
-
"s_loss": 0.0008826229604892433,
|
| 1832 |
-
"step": 18300
|
| 1833 |
-
},
|
| 1834 |
-
{
|
| 1835 |
-
"epoch": 0.55,
|
| 1836 |
-
"learning_rate": 2.150735322790704e-05,
|
| 1837 |
-
"loss": 0.0071,
|
| 1838 |
-
"r_loss": 0.005803743377327919,
|
| 1839 |
-
"s_loss": 0.0009314118069596589,
|
| 1840 |
-
"step": 18400
|
| 1841 |
-
},
|
| 1842 |
-
{
|
| 1843 |
-
"epoch": 0.55,
|
| 1844 |
-
"learning_rate": 2.1326203091505936e-05,
|
| 1845 |
-
"loss": 0.0071,
|
| 1846 |
-
"r_loss": 0.007132242433726788,
|
| 1847 |
-
"s_loss": 0.0009220225038006902,
|
| 1848 |
-
"step": 18500
|
| 1849 |
-
},
|
| 1850 |
-
{
|
| 1851 |
-
"epoch": 0.55,
|
| 1852 |
-
"eval_loss": 0.007128148805350065,
|
| 1853 |
-
"eval_r_loss": 0.006268758792430162,
|
| 1854 |
-
"eval_runtime": 8.6423,
|
| 1855 |
-
"eval_s_loss": 0.0008593900711275637,
|
| 1856 |
-
"eval_samples_per_second": 77.757,
|
| 1857 |
-
"eval_steps_per_second": 77.757,
|
| 1858 |
-
"step": 18500
|
| 1859 |
-
},
|
| 1860 |
-
{
|
| 1861 |
-
"epoch": 0.56,
|
| 1862 |
-
"learning_rate": 2.114392362650425e-05,
|
| 1863 |
-
"loss": 0.007,
|
| 1864 |
-
"r_loss": 0.006902765482664108,
|
| 1865 |
-
"s_loss": 0.0009045482147485018,
|
| 1866 |
-
"step": 18600
|
| 1867 |
-
},
|
| 1868 |
-
{
|
| 1869 |
-
"epoch": 0.56,
|
| 1870 |
-
"learning_rate": 2.096054737270669e-05,
|
| 1871 |
-
"loss": 0.0071,
|
| 1872 |
-
"r_loss": 0.006586451083421707,
|
| 1873 |
-
"s_loss": 0.000991364591754973,
|
| 1874 |
-
"step": 18700
|
| 1875 |
-
},
|
| 1876 |
-
{
|
| 1877 |
-
"epoch": 0.56,
|
| 1878 |
-
"learning_rate": 2.0776107065712326e-05,
|
| 1879 |
-
"loss": 0.007,
|
| 1880 |
-
"r_loss": 0.004980746190994978,
|
| 1881 |
-
"s_loss": 0.0007370096282102168,
|
| 1882 |
-
"step": 18800
|
| 1883 |
-
},
|
| 1884 |
-
{
|
| 1885 |
-
"epoch": 0.56,
|
| 1886 |
-
"learning_rate": 2.059063563107079e-05,
|
| 1887 |
-
"loss": 0.0071,
|
| 1888 |
-
"r_loss": 0.006382007151842117,
|
| 1889 |
-
"s_loss": 0.0008771989960223436,
|
| 1890 |
-
"step": 18900
|
| 1891 |
-
},
|
| 1892 |
-
{
|
| 1893 |
-
"epoch": 0.57,
|
| 1894 |
-
"learning_rate": 2.040416617840449e-05,
|
| 1895 |
-
"loss": 0.007,
|
| 1896 |
-
"r_loss": 0.0058512031100690365,
|
| 1897 |
-
"s_loss": 0.0008522871066816151,
|
| 1898 |
-
"step": 19000
|
| 1899 |
-
},
|
| 1900 |
-
{
|
| 1901 |
-
"epoch": 0.57,
|
| 1902 |
-
"eval_loss": 0.007616510149091482,
|
| 1903 |
-
"eval_r_loss": 0.006736051291227341,
|
| 1904 |
-
"eval_runtime": 8.8015,
|
| 1905 |
-
"eval_s_loss": 0.0008804587414488196,
|
| 1906 |
-
"eval_samples_per_second": 76.35,
|
| 1907 |
-
"eval_steps_per_second": 76.35,
|
| 1908 |
-
"step": 19000
|
| 1909 |
-
},
|
| 1910 |
-
{
|
| 1911 |
-
"epoch": 0.57,
|
| 1912 |
-
"learning_rate": 2.021673199549806e-05,
|
| 1913 |
-
"loss": 0.007,
|
| 1914 |
-
"r_loss": 0.005660332273691893,
|
| 1915 |
-
"s_loss": 0.0009998377645388246,
|
| 1916 |
-
"step": 19100
|
| 1917 |
-
},
|
| 1918 |
-
{
|
| 1919 |
-
"epoch": 0.57,
|
| 1920 |
-
"learning_rate": 2.002836654235594e-05,
|
| 1921 |
-
"loss": 0.0069,
|
| 1922 |
-
"r_loss": 0.005866233725100756,
|
| 1923 |
-
"s_loss": 0.0009188687545247376,
|
| 1924 |
-
"step": 19200
|
| 1925 |
-
},
|
| 1926 |
-
{
|
| 1927 |
-
"epoch": 0.58,
|
| 1928 |
-
"learning_rate": 1.983910344522921e-05,
|
| 1929 |
-
"loss": 0.0069,
|
| 1930 |
-
"r_loss": 0.0057606166228652,
|
| 1931 |
-
"s_loss": 0.000834951177239418,
|
| 1932 |
-
"step": 19300
|
| 1933 |
-
},
|
| 1934 |
-
{
|
| 1935 |
-
"epoch": 0.58,
|
| 1936 |
-
"learning_rate": 1.9648976490612795e-05,
|
| 1937 |
-
"loss": 0.0067,
|
| 1938 |
-
"r_loss": 0.004965795204043388,
|
| 1939 |
-
"s_loss": 0.0008518850081600249,
|
| 1940 |
-
"step": 19400
|
| 1941 |
-
},
|
| 1942 |
-
{
|
| 1943 |
-
"epoch": 0.58,
|
| 1944 |
-
"learning_rate": 1.9459933184205116e-05,
|
| 1945 |
-
"loss": 0.0069,
|
| 1946 |
-
"r_loss": 0.006777866743505001,
|
| 1947 |
-
"s_loss": 0.0010352524695917964,
|
| 1948 |
-
"step": 19500
|
| 1949 |
-
},
|
| 1950 |
-
{
|
| 1951 |
-
"epoch": 0.58,
|
| 1952 |
-
"eval_loss": 0.007355178706347942,
|
| 1953 |
-
"eval_r_loss": 0.006485129706561565,
|
| 1954 |
-
"eval_runtime": 8.9082,
|
| 1955 |
-
"eval_s_loss": 0.0008700488251633942,
|
| 1956 |
-
"eval_samples_per_second": 75.436,
|
| 1957 |
-
"eval_steps_per_second": 75.436,
|
| 1958 |
-
"step": 19500
|
| 1959 |
-
},
|
| 1960 |
-
{
|
| 1961 |
-
"epoch": 0.58,
|
| 1962 |
-
"learning_rate": 1.9268188273952553e-05,
|
| 1963 |
-
"loss": 0.0068,
|
| 1964 |
-
"r_loss": 0.007018654141575098,
|
| 1965 |
-
"s_loss": 0.0009793075732886791,
|
| 1966 |
-
"step": 19600
|
| 1967 |
-
},
|
| 1968 |
-
{
|
| 1969 |
-
"epoch": 0.59,
|
| 1970 |
-
"learning_rate": 1.9075681423714705e-05,
|
| 1971 |
-
"loss": 0.0069,
|
| 1972 |
-
"r_loss": 0.006022634916007519,
|
| 1973 |
-
"s_loss": 0.0008647244540043175,
|
| 1974 |
-
"step": 19700
|
| 1975 |
-
},
|
| 1976 |
-
{
|
| 1977 |
-
"epoch": 0.59,
|
| 1978 |
-
"learning_rate": 1.8882446999048395e-05,
|
| 1979 |
-
"loss": 0.0068,
|
| 1980 |
-
"r_loss": 0.005010883789509535,
|
| 1981 |
-
"s_loss": 0.0008228466031141579,
|
| 1982 |
-
"step": 19800
|
| 1983 |
-
},
|
| 1984 |
-
{
|
| 1985 |
-
"epoch": 0.59,
|
| 1986 |
-
"learning_rate": 1.8688519495394125e-05,
|
| 1987 |
-
"loss": 0.0069,
|
| 1988 |
-
"r_loss": 0.006728707812726498,
|
| 1989 |
-
"s_loss": 0.000956969684921205,
|
| 1990 |
-
"step": 19900
|
| 1991 |
-
},
|
| 1992 |
-
{
|
| 1993 |
-
"epoch": 0.6,
|
| 1994 |
-
"learning_rate": 1.8493933531918117e-05,
|
| 1995 |
-
"loss": 0.0068,
|
| 1996 |
-
"r_loss": 0.007086700294166803,
|
| 1997 |
-
"s_loss": 0.0008907000883482397,
|
| 1998 |
-
"step": 20000
|
| 1999 |
-
},
|
| 2000 |
-
{
|
| 2001 |
-
"epoch": 0.6,
|
| 2002 |
-
"eval_loss": 0.006719064898788929,
|
| 2003 |
-
"eval_r_loss": 0.005855009891092777,
|
| 2004 |
-
"eval_runtime": 8.5414,
|
| 2005 |
-
"eval_s_loss": 0.0008640547748655081,
|
| 2006 |
-
"eval_samples_per_second": 78.676,
|
| 2007 |
-
"eval_steps_per_second": 78.676,
|
| 2008 |
-
"step": 20000
|
| 2009 |
-
},
|
| 2010 |
-
{
|
| 2011 |
-
"epoch": 0.6,
|
| 2012 |
-
"learning_rate": 1.8298723845332198e-05,
|
| 2013 |
-
"loss": 0.0067,
|
| 2014 |
-
"r_loss": 0.005535440053790808,
|
| 2015 |
-
"s_loss": 0.0009100943570956588,
|
| 2016 |
-
"step": 20100
|
| 2017 |
-
},
|
| 2018 |
-
{
|
| 2019 |
-
"epoch": 0.6,
|
| 2020 |
-
"learning_rate": 1.8102925283692782e-05,
|
| 2021 |
-
"loss": 0.0068,
|
| 2022 |
-
"r_loss": 0.006036119069904089,
|
| 2023 |
-
"s_loss": 0.0008750570705160499,
|
| 2024 |
-
"step": 20200
|
| 2025 |
-
},
|
| 2026 |
-
{
|
| 2027 |
-
"epoch": 0.61,
|
| 2028 |
-
"learning_rate": 1.79065728001799e-05,
|
| 2029 |
-
"loss": 0.0068,
|
| 2030 |
-
"r_loss": 0.005460575222969055,
|
| 2031 |
-
"s_loss": 0.0009543396299704909,
|
| 2032 |
-
"step": 20300
|
| 2033 |
-
},
|
| 2034 |
-
{
|
| 2035 |
-
"epoch": 0.61,
|
| 2036 |
-
"learning_rate": 1.7709701446857527e-05,
|
| 2037 |
-
"loss": 0.0068,
|
| 2038 |
-
"r_loss": 0.005476498045027256,
|
| 2039 |
-
"s_loss": 0.0008106306777335703,
|
| 2040 |
-
"step": 20400
|
| 2041 |
-
},
|
| 2042 |
-
{
|
| 2043 |
-
"epoch": 0.61,
|
| 2044 |
-
"learning_rate": 1.75123463684162e-05,
|
| 2045 |
-
"loss": 0.0069,
|
| 2046 |
-
"r_loss": 0.006093060597777367,
|
| 2047 |
-
"s_loss": 0.0009121097973547876,
|
| 2048 |
-
"step": 20500
|
| 2049 |
-
},
|
| 2050 |
-
{
|
| 2051 |
-
"epoch": 0.61,
|
| 2052 |
-
"eval_loss": 0.006673221942037344,
|
| 2053 |
-
"eval_r_loss": 0.005839366465806961,
|
| 2054 |
-
"eval_runtime": 8.4759,
|
| 2055 |
-
"eval_s_loss": 0.0008338554762303829,
|
| 2056 |
-
"eval_samples_per_second": 79.284,
|
| 2057 |
-
"eval_steps_per_second": 79.284,
|
| 2058 |
-
"step": 20500
|
| 2059 |
-
},
|
| 2060 |
-
{
|
| 2061 |
-
"epoch": 0.61,
|
| 2062 |
-
"learning_rate": 1.7314542795899137e-05,
|
| 2063 |
-
"loss": 0.0067,
|
| 2064 |
-
"r_loss": 0.005582145415246487,
|
| 2065 |
-
"s_loss": 0.0008446918218396604,
|
| 2066 |
-
"step": 20600
|
| 2067 |
-
},
|
| 2068 |
-
{
|
| 2069 |
-
"epoch": 0.62,
|
| 2070 |
-
"learning_rate": 1.7116326040412943e-05,
|
| 2071 |
-
"loss": 0.0067,
|
| 2072 |
-
"r_loss": 0.005406418815255165,
|
| 2073 |
-
"s_loss": 0.0010036693420261145,
|
| 2074 |
-
"step": 20700
|
| 2075 |
-
},
|
| 2076 |
-
{
|
| 2077 |
-
"epoch": 0.62,
|
| 2078 |
-
"learning_rate": 1.6917731486823998e-05,
|
| 2079 |
-
"loss": 0.0068,
|
| 2080 |
-
"r_loss": 0.005483163520693779,
|
| 2081 |
-
"s_loss": 0.0009342134580947459,
|
| 2082 |
-
"step": 20800
|
| 2083 |
-
},
|
| 2084 |
-
{
|
| 2085 |
-
"epoch": 0.62,
|
| 2086 |
-
"learning_rate": 1.6718794587441696e-05,
|
| 2087 |
-
"loss": 0.0067,
|
| 2088 |
-
"r_loss": 0.005454606376588345,
|
| 2089 |
-
"s_loss": 0.0008848806610330939,
|
| 2090 |
-
"step": 20900
|
| 2091 |
-
},
|
| 2092 |
-
{
|
| 2093 |
-
"epoch": 0.63,
|
| 2094 |
-
"learning_rate": 1.6519550855689638e-05,
|
| 2095 |
-
"loss": 0.0067,
|
| 2096 |
-
"r_loss": 0.005080068949609995,
|
| 2097 |
-
"s_loss": 0.0008759861811995506,
|
| 2098 |
-
"step": 21000
|
| 2099 |
-
},
|
| 2100 |
-
{
|
| 2101 |
-
"epoch": 0.63,
|
| 2102 |
-
"eval_loss": 0.006933785974979401,
|
| 2103 |
-
"eval_r_loss": 0.0060889944434165955,
|
| 2104 |
-
"eval_runtime": 8.6029,
|
| 2105 |
-
"eval_s_loss": 0.0008447913569398224,
|
| 2106 |
-
"eval_samples_per_second": 78.113,
|
| 2107 |
-
"eval_steps_per_second": 78.113,
|
| 2108 |
-
"step": 21000
|
| 2109 |
-
},
|
| 2110 |
-
{
|
| 2111 |
-
"epoch": 0.63,
|
| 2112 |
-
"learning_rate": 1.6320035859765918e-05,
|
| 2113 |
-
"loss": 0.0068,
|
| 2114 |
-
"r_loss": 0.006143931299448013,
|
| 2115 |
-
"s_loss": 0.000945593579672277,
|
| 2116 |
-
"step": 21100
|
| 2117 |
-
},
|
| 2118 |
-
{
|
| 2119 |
-
"epoch": 0.63,
|
| 2120 |
-
"learning_rate": 1.612028521629364e-05,
|
| 2121 |
-
"loss": 0.0069,
|
| 2122 |
-
"r_loss": 0.006441016681492329,
|
| 2123 |
-
"s_loss": 0.0009456037660129368,
|
| 2124 |
-
"step": 21200
|
| 2125 |
-
},
|
| 2126 |
-
{
|
| 2127 |
-
"epoch": 0.64,
|
| 2128 |
-
"learning_rate": 1.5920334583962753e-05,
|
| 2129 |
-
"loss": 0.0067,
|
| 2130 |
-
"r_loss": 0.006160522345453501,
|
| 2131 |
-
"s_loss": 0.0008229748345911503,
|
| 2132 |
-
"step": 21300
|
| 2133 |
-
},
|
| 2134 |
-
{
|
| 2135 |
-
"epoch": 0.64,
|
| 2136 |
-
"learning_rate": 1.5720219657164435e-05,
|
| 2137 |
-
"loss": 0.0067,
|
| 2138 |
-
"r_loss": 0.005891709588468075,
|
| 2139 |
-
"s_loss": 0.000853882054798305,
|
| 2140 |
-
"step": 21400
|
| 2141 |
-
},
|
| 2142 |
-
{
|
| 2143 |
-
"epoch": 0.64,
|
| 2144 |
-
"learning_rate": 1.551997615961906e-05,
|
| 2145 |
-
"loss": 0.0067,
|
| 2146 |
-
"r_loss": 0.005442744120955467,
|
| 2147 |
-
"s_loss": 0.0008191297529265285,
|
| 2148 |
-
"step": 21500
|
| 2149 |
-
},
|
| 2150 |
-
{
|
| 2151 |
-
"epoch": 0.64,
|
| 2152 |
-
"eval_loss": 0.007066499907523394,
|
| 2153 |
-
"eval_r_loss": 0.006228615529835224,
|
| 2154 |
-
"eval_runtime": 10.9272,
|
| 2155 |
-
"eval_s_loss": 0.0008378842030651867,
|
| 2156 |
-
"eval_samples_per_second": 61.498,
|
| 2157 |
-
"eval_steps_per_second": 61.498,
|
| 2158 |
-
"step": 21500
|
| 2159 |
-
},
|
| 2160 |
-
{
|
| 2161 |
-
"epoch": 0.64,
|
| 2162 |
-
"learning_rate": 1.5319639837998926e-05,
|
| 2163 |
-
"loss": 0.0066,
|
| 2164 |
-
"r_loss": 0.006707844324409962,
|
| 2165 |
-
"s_loss": 0.0009414084488525987,
|
| 2166 |
-
"step": 21600
|
| 2167 |
-
},
|
| 2168 |
-
{
|
| 2169 |
-
"epoch": 0.65,
|
| 2170 |
-
"learning_rate": 1.5119246455546931e-05,
|
| 2171 |
-
"loss": 0.0066,
|
| 2172 |
-
"r_loss": 0.0060582030564546585,
|
| 2173 |
-
"s_loss": 0.0009419742273166776,
|
| 2174 |
-
"step": 21700
|
| 2175 |
-
},
|
| 2176 |
-
{
|
| 2177 |
-
"epoch": 0.65,
|
| 2178 |
-
"learning_rate": 1.4918831785692232e-05,
|
| 2179 |
-
"loss": 0.0067,
|
| 2180 |
-
"r_loss": 0.006246947217732668,
|
| 2181 |
-
"s_loss": 0.0008474804344587028,
|
| 2182 |
-
"step": 21800
|
| 2183 |
-
},
|
| 2184 |
-
{
|
| 2185 |
-
"epoch": 0.65,
|
| 2186 |
-
"learning_rate": 1.4718431605664146e-05,
|
| 2187 |
-
"loss": 0.0066,
|
| 2188 |
-
"r_loss": 0.00554366409778595,
|
| 2189 |
-
"s_loss": 0.000851424119900912,
|
| 2190 |
-
"step": 21900
|
| 2191 |
-
},
|
| 2192 |
-
{
|
| 2193 |
-
"epoch": 0.66,
|
| 2194 |
-
"learning_rate": 1.4518081690105308e-05,
|
| 2195 |
-
"loss": 0.0065,
|
| 2196 |
-
"r_loss": 0.005924141500145197,
|
| 2197 |
-
"s_loss": 0.0008292071870528162,
|
| 2198 |
-
"step": 22000
|
| 2199 |
-
},
|
| 2200 |
-
{
|
| 2201 |
-
"epoch": 0.66,
|
| 2202 |
-
"eval_loss": 0.006922111380845308,
|
| 2203 |
-
"eval_r_loss": 0.00609211903065443,
|
| 2204 |
-
"eval_runtime": 8.5409,
|
| 2205 |
-
"eval_s_loss": 0.0008299925248138607,
|
| 2206 |
-
"eval_samples_per_second": 78.68,
|
| 2207 |
-
"eval_steps_per_second": 78.68,
|
| 2208 |
-
"step": 22000
|
| 2209 |
-
},
|
| 2210 |
-
{
|
| 2211 |
-
"epoch": 0.66,
|
| 2212 |
-
"learning_rate": 1.431781780468534e-05,
|
| 2213 |
-
"loss": 0.0066,
|
| 2214 |
-
"r_loss": 0.006054874509572983,
|
| 2215 |
-
"s_loss": 0.0008940041880123317,
|
| 2216 |
-
"step": 22100
|
| 2217 |
-
},
|
| 2218 |
-
{
|
| 2219 |
-
"epoch": 0.66,
|
| 2220 |
-
"learning_rate": 1.4119676400616625e-05,
|
| 2221 |
-
"loss": 0.0065,
|
| 2222 |
-
"r_loss": 0.005405670963227749,
|
| 2223 |
-
"s_loss": 0.0009109702077694237,
|
| 2224 |
-
"step": 22200
|
| 2225 |
-
},
|
| 2226 |
-
{
|
| 2227 |
-
"epoch": 0.67,
|
| 2228 |
-
"learning_rate": 1.3919690052810628e-05,
|
| 2229 |
-
"loss": 0.0064,
|
| 2230 |
-
"r_loss": 0.004920615814626217,
|
| 2231 |
-
"s_loss": 0.0008736539166420698,
|
| 2232 |
-
"step": 22300
|
| 2233 |
-
},
|
| 2234 |
-
{
|
| 2235 |
-
"epoch": 0.67,
|
| 2236 |
-
"learning_rate": 1.3719896557637283e-05,
|
| 2237 |
-
"loss": 0.0064,
|
| 2238 |
-
"r_loss": 0.0057085175067186356,
|
| 2239 |
-
"s_loss": 0.0008784402743913233,
|
| 2240 |
-
"step": 22400
|
| 2241 |
-
},
|
| 2242 |
-
{
|
| 2243 |
-
"epoch": 0.67,
|
| 2244 |
-
"learning_rate": 1.3520331581436254e-05,
|
| 2245 |
-
"loss": 0.0065,
|
| 2246 |
-
"r_loss": 0.004970056004822254,
|
| 2247 |
-
"s_loss": 0.0008474025526084006,
|
| 2248 |
-
"step": 22500
|
| 2249 |
-
},
|
| 2250 |
-
{
|
| 2251 |
-
"epoch": 0.67,
|
| 2252 |
-
"eval_loss": 0.00664812745526433,
|
| 2253 |
-
"eval_r_loss": 0.005824685096740723,
|
| 2254 |
-
"eval_runtime": 8.4868,
|
| 2255 |
-
"eval_s_loss": 0.0008234424167312682,
|
| 2256 |
-
"eval_samples_per_second": 79.181,
|
| 2257 |
-
"eval_steps_per_second": 79.181,
|
| 2258 |
-
"step": 22500
|
| 2259 |
-
},
|
| 2260 |
-
{
|
| 2261 |
-
"epoch": 0.67,
|
| 2262 |
-
"learning_rate": 1.3321030749752928e-05,
|
| 2263 |
-
"loss": 0.0065,
|
| 2264 |
-
"r_loss": 0.006657294929027557,
|
| 2265 |
-
"s_loss": 0.0008960987906903028,
|
| 2266 |
-
"step": 22600
|
| 2267 |
-
},
|
| 2268 |
-
{
|
| 2269 |
-
"epoch": 0.68,
|
| 2270 |
-
"learning_rate": 1.3122029640978642e-05,
|
| 2271 |
-
"loss": 0.0065,
|
| 2272 |
-
"r_loss": 0.0066483840346336365,
|
| 2273 |
-
"s_loss": 0.0008470122702419758,
|
| 2274 |
-
"step": 22700
|
| 2275 |
-
},
|
| 2276 |
-
{
|
| 2277 |
-
"epoch": 0.68,
|
| 2278 |
-
"learning_rate": 1.2923363779999415e-05,
|
| 2279 |
-
"loss": 0.0064,
|
| 2280 |
-
"r_loss": 0.006104170344769955,
|
| 2281 |
-
"s_loss": 0.000873480923473835,
|
| 2282 |
-
"step": 22800
|
| 2283 |
-
},
|
| 2284 |
-
{
|
| 2285 |
-
"epoch": 0.68,
|
| 2286 |
-
"learning_rate": 1.2725068631854143e-05,
|
| 2287 |
-
"loss": 0.0065,
|
| 2288 |
-
"r_loss": 0.0065166400745511055,
|
| 2289 |
-
"s_loss": 0.0008359896601177752,
|
| 2290 |
-
"step": 22900
|
| 2291 |
-
},
|
| 2292 |
-
{
|
| 2293 |
-
"epoch": 0.69,
|
| 2294 |
-
"learning_rate": 1.2527179595403555e-05,
|
| 2295 |
-
"loss": 0.0065,
|
| 2296 |
-
"r_loss": 0.005165203474462032,
|
| 2297 |
-
"s_loss": 0.0008489371393807232,
|
| 2298 |
-
"step": 23000
|
| 2299 |
-
},
|
| 2300 |
-
{
|
| 2301 |
-
"epoch": 0.69,
|
| 2302 |
-
"eval_loss": 0.007038415875285864,
|
| 2303 |
-
"eval_r_loss": 0.006207308266311884,
|
| 2304 |
-
"eval_runtime": 8.684,
|
| 2305 |
-
"eval_s_loss": 0.00083110760897398,
|
| 2306 |
-
"eval_samples_per_second": 77.384,
|
| 2307 |
-
"eval_steps_per_second": 77.384,
|
| 2308 |
-
"step": 23000
|
| 2309 |
-
},
|
| 2310 |
-
{
|
| 2311 |
-
"epoch": 0.69,
|
| 2312 |
-
"learning_rate": 1.2329731997010932e-05,
|
| 2313 |
-
"loss": 0.0064,
|
| 2314 |
-
"r_loss": 0.005528077483177185,
|
| 2315 |
-
"s_loss": 0.0008984919404610991,
|
| 2316 |
-
"step": 23100
|
| 2317 |
-
},
|
| 2318 |
-
{
|
| 2319 |
-
"epoch": 0.69,
|
| 2320 |
-
"learning_rate": 1.2132761084235799e-05,
|
| 2321 |
-
"loss": 0.0064,
|
| 2322 |
-
"r_loss": 0.005268075503408909,
|
| 2323 |
-
"s_loss": 0.0008533818763680756,
|
| 2324 |
-
"step": 23200
|
| 2325 |
-
},
|
| 2326 |
-
{
|
| 2327 |
-
"epoch": 0.7,
|
| 2328 |
-
"learning_rate": 1.1936302019541638e-05,
|
| 2329 |
-
"loss": 0.0065,
|
| 2330 |
-
"r_loss": 0.004923930391669273,
|
| 2331 |
-
"s_loss": 0.0008330261334776878,
|
| 2332 |
-
"step": 23300
|
| 2333 |
-
},
|
| 2334 |
-
{
|
| 2335 |
-
"epoch": 0.7,
|
| 2336 |
-
"learning_rate": 1.1740389874018872e-05,
|
| 2337 |
-
"loss": 0.0064,
|
| 2338 |
-
"r_loss": 0.0052954284474253654,
|
| 2339 |
-
"s_loss": 0.0008716843440197408,
|
| 2340 |
-
"step": 23400
|
| 2341 |
-
},
|
| 2342 |
-
{
|
| 2343 |
-
"epoch": 0.7,
|
| 2344 |
-
"learning_rate": 1.1545059621124078e-05,
|
| 2345 |
-
"loss": 0.0064,
|
| 2346 |
-
"r_loss": 0.005905309226363897,
|
| 2347 |
-
"s_loss": 0.0008605217444710433,
|
| 2348 |
-
"step": 23500
|
| 2349 |
-
},
|
| 2350 |
-
{
|
| 2351 |
-
"epoch": 0.7,
|
| 2352 |
-
"eval_loss": 0.006766035221517086,
|
| 2353 |
-
"eval_r_loss": 0.005948369391262531,
|
| 2354 |
-
"eval_runtime": 9.2172,
|
| 2355 |
-
"eval_s_loss": 0.0008176658302545547,
|
| 2356 |
-
"eval_samples_per_second": 72.908,
|
| 2357 |
-
"eval_steps_per_second": 72.908,
|
| 2358 |
-
"step": 23500
|
| 2359 |
-
},
|
| 2360 |
-
{
|
| 2361 |
-
"epoch": 0.7,
|
| 2362 |
-
"learning_rate": 1.1350346130436692e-05,
|
| 2363 |
-
"loss": 0.0064,
|
| 2364 |
-
"r_loss": 0.005245131440460682,
|
| 2365 |
-
"s_loss": 0.0008455686620436609,
|
| 2366 |
-
"step": 23600
|
| 2367 |
-
},
|
| 2368 |
-
{
|
| 2369 |
-
"epoch": 0.71,
|
| 2370 |
-
"learning_rate": 1.1156284161434186e-05,
|
| 2371 |
-
"loss": 0.0064,
|
| 2372 |
-
"r_loss": 0.006068192422389984,
|
| 2373 |
-
"s_loss": 0.0008869940065778792,
|
| 2374 |
-
"step": 23700
|
| 2375 |
-
},
|
| 2376 |
-
{
|
| 2377 |
-
"epoch": 0.71,
|
| 2378 |
-
"learning_rate": 1.0962908357287e-05,
|
| 2379 |
-
"loss": 0.0063,
|
| 2380 |
-
"r_loss": 0.00627498421818018,
|
| 2381 |
-
"s_loss": 0.0008408837020397186,
|
| 2382 |
-
"step": 23800
|
| 2383 |
-
},
|
| 2384 |
-
{
|
| 2385 |
-
"epoch": 0.71,
|
| 2386 |
-
"learning_rate": 1.0770253238674148e-05,
|
| 2387 |
-
"loss": 0.0065,
|
| 2388 |
-
"r_loss": 0.006341907661408186,
|
| 2389 |
-
"s_loss": 0.0008763980586081743,
|
| 2390 |
-
"step": 23900
|
| 2391 |
-
},
|
| 2392 |
-
{
|
| 2393 |
-
"epoch": 0.72,
|
| 2394 |
-
"learning_rate": 1.0578353197620722e-05,
|
| 2395 |
-
"loss": 0.0064,
|
| 2396 |
-
"r_loss": 0.005438666325062513,
|
| 2397 |
-
"s_loss": 0.0008361585787497461,
|
| 2398 |
-
"step": 24000
|
| 2399 |
-
},
|
| 2400 |
-
{
|
| 2401 |
-
"epoch": 0.72,
|
| 2402 |
-
"eval_loss": 0.006421332713216543,
|
| 2403 |
-
"eval_r_loss": 0.005617132410407066,
|
| 2404 |
-
"eval_runtime": 8.6645,
|
| 2405 |
-
"eval_s_loss": 0.000804200186394155,
|
| 2406 |
-
"eval_samples_per_second": 77.558,
|
| 2407 |
-
"eval_steps_per_second": 77.558,
|
| 2408 |
-
"step": 24000
|
| 2409 |
-
},
|
| 2410 |
-
{
|
| 2411 |
-
"epoch": 0.72,
|
| 2412 |
-
"learning_rate": 1.0387242491358379e-05,
|
| 2413 |
-
"loss": 0.0064,
|
| 2414 |
-
"r_loss": 0.004842773545533419,
|
| 2415 |
-
"s_loss": 0.0008286428637802601,
|
| 2416 |
-
"step": 24100
|
| 2417 |
-
},
|
| 2418 |
-
{
|
| 2419 |
-
"epoch": 0.72,
|
| 2420 |
-
"learning_rate": 1.0196955236209875e-05,
|
| 2421 |
-
"loss": 0.0064,
|
| 2422 |
-
"r_loss": 0.005401437636464834,
|
| 2423 |
-
"s_loss": 0.0008410606533288956,
|
| 2424 |
-
"step": 24200
|
| 2425 |
-
},
|
| 2426 |
-
{
|
| 2427 |
-
"epoch": 0.73,
|
| 2428 |
-
"learning_rate": 1.0007525401498747e-05,
|
| 2429 |
-
"loss": 0.0064,
|
| 2430 |
-
"r_loss": 0.005490908399224281,
|
| 2431 |
-
"s_loss": 0.0009162276983261108,
|
| 2432 |
-
"step": 24300
|
| 2433 |
-
},
|
| 2434 |
-
{
|
| 2435 |
-
"epoch": 0.73,
|
| 2436 |
-
"learning_rate": 9.818986803485237e-06,
|
| 2437 |
-
"loss": 0.0064,
|
| 2438 |
-
"r_loss": 0.004949535708874464,
|
| 2439 |
-
"s_loss": 0.0008561740978620946,
|
| 2440 |
-
"step": 24400
|
| 2441 |
-
},
|
| 2442 |
-
{
|
| 2443 |
-
"epoch": 0.73,
|
| 2444 |
-
"learning_rate": 9.63137309932957e-06,
|
| 2445 |
-
"loss": 0.0063,
|
| 2446 |
-
"r_loss": 0.004786663688719273,
|
| 2447 |
-
"s_loss": 0.0008052530465647578,
|
| 2448 |
-
"step": 24500
|
| 2449 |
-
},
|
| 2450 |
-
{
|
| 2451 |
-
"epoch": 0.73,
|
| 2452 |
-
"eval_loss": 0.006642198655754328,
|
| 2453 |
-
"eval_r_loss": 0.005832642316818237,
|
| 2454 |
-
"eval_runtime": 8.6342,
|
| 2455 |
-
"eval_s_loss": 0.0008095565135590732,
|
| 2456 |
-
"eval_samples_per_second": 77.83,
|
| 2457 |
-
"eval_steps_per_second": 77.83,
|
| 2458 |
-
"step": 24500
|
| 2459 |
-
},
|
| 2460 |
-
{
|
| 2461 |
-
"epoch": 0.73,
|
| 2462 |
-
"learning_rate": 9.444717781083603e-06,
|
| 2463 |
-
"loss": 0.0064,
|
| 2464 |
-
"r_loss": 0.004507116507738829,
|
| 2465 |
-
"s_loss": 0.0007864255458116531,
|
| 2466 |
-
"step": 24600
|
| 2467 |
-
},
|
| 2468 |
-
{
|
| 2469 |
-
"epoch": 0.74,
|
| 2470 |
-
"learning_rate": 9.25905416971195e-06,
|
| 2471 |
-
"loss": 0.0063,
|
| 2472 |
-
"r_loss": 0.005326538346707821,
|
| 2473 |
-
"s_loss": 0.0008638648432679474,
|
| 2474 |
-
"step": 24700
|
| 2475 |
-
},
|
| 2476 |
-
{
|
| 2477 |
-
"epoch": 0.74,
|
| 2478 |
-
"learning_rate": 9.074415409143651e-06,
|
| 2479 |
-
"loss": 0.0064,
|
| 2480 |
-
"r_loss": 0.005989129655063152,
|
| 2481 |
-
"s_loss": 0.0009261829545721412,
|
| 2482 |
-
"step": 24800
|
| 2483 |
-
},
|
| 2484 |
-
{
|
| 2485 |
-
"epoch": 0.74,
|
| 2486 |
-
"learning_rate": 8.890834460355467e-06,
|
| 2487 |
-
"loss": 0.0063,
|
| 2488 |
-
"r_loss": 0.004806933458894491,
|
| 2489 |
-
"s_loss": 0.0008874195045791566,
|
| 2490 |
-
"step": 24900
|
| 2491 |
-
},
|
| 2492 |
-
{
|
| 2493 |
-
"epoch": 0.75,
|
| 2494 |
-
"learning_rate": 8.708344095487813e-06,
|
| 2495 |
-
"loss": 0.0063,
|
| 2496 |
-
"r_loss": 0.0051068831235170364,
|
| 2497 |
-
"s_loss": 0.000819625158328563,
|
| 2498 |
-
"step": 25000
|
| 2499 |
-
},
|
| 2500 |
-
{
|
| 2501 |
-
"epoch": 0.75,
|
| 2502 |
-
"eval_loss": 0.006493990775197744,
|
| 2503 |
-
"eval_r_loss": 0.005688361823558807,
|
| 2504 |
-
"eval_runtime": 8.6045,
|
| 2505 |
-
"eval_s_loss": 0.000805628951638937,
|
| 2506 |
-
"eval_samples_per_second": 78.098,
|
| 2507 |
-
"eval_steps_per_second": 78.098,
|
| 2508 |
-
"step": 25000
|
| 2509 |
-
},
|
| 2510 |
-
{
|
| 2511 |
-
"epoch": 0.75,
|
| 2512 |
-
"learning_rate": 8.526976891994414e-06,
|
| 2513 |
-
"loss": 0.0062,
|
| 2514 |
-
"r_loss": 0.006579666864126921,
|
| 2515 |
-
"s_loss": 0.0009487958159297705,
|
| 2516 |
-
"step": 25100
|
| 2517 |
-
},
|
| 2518 |
-
{
|
| 2519 |
-
"epoch": 0.75,
|
| 2520 |
-
"learning_rate": 8.346765226826655e-06,
|
| 2521 |
-
"loss": 0.0063,
|
| 2522 |
-
"r_loss": 0.004679057281464338,
|
| 2523 |
-
"s_loss": 0.0007369701052084565,
|
| 2524 |
-
"step": 25200
|
| 2525 |
-
},
|
| 2526 |
-
{
|
| 2527 |
-
"epoch": 0.75,
|
| 2528 |
-
"learning_rate": 8.167741270653863e-06,
|
| 2529 |
-
"loss": 0.0062,
|
| 2530 |
-
"r_loss": 0.005413247272372246,
|
| 2531 |
-
"s_loss": 0.000827790005132556,
|
| 2532 |
-
"step": 25300
|
| 2533 |
-
},
|
| 2534 |
-
{
|
| 2535 |
-
"epoch": 0.76,
|
| 2536 |
-
"learning_rate": 7.989936982120253e-06,
|
| 2537 |
-
"loss": 0.0062,
|
| 2538 |
-
"r_loss": 0.0051015885546803474,
|
| 2539 |
-
"s_loss": 0.0007679238333366811,
|
| 2540 |
-
"step": 25400
|
| 2541 |
-
},
|
| 2542 |
-
{
|
| 2543 |
-
"epoch": 0.76,
|
| 2544 |
-
"learning_rate": 7.813384102139837e-06,
|
| 2545 |
-
"loss": 0.0062,
|
| 2546 |
-
"r_loss": 0.007656958419829607,
|
| 2547 |
-
"s_loss": 0.0009585011284798384,
|
| 2548 |
-
"step": 25500
|
| 2549 |
-
},
|
| 2550 |
-
{
|
| 2551 |
-
"epoch": 0.76,
|
| 2552 |
-
"eval_loss": 0.006621798500418663,
|
| 2553 |
-
"eval_r_loss": 0.005829837638884783,
|
| 2554 |
-
"eval_runtime": 8.6093,
|
| 2555 |
-
"eval_s_loss": 0.0007919610943645239,
|
| 2556 |
-
"eval_samples_per_second": 78.055,
|
| 2557 |
-
"eval_steps_per_second": 78.055,
|
| 2558 |
-
"step": 25500
|
| 2559 |
-
},
|
| 2560 |
-
{
|
| 2561 |
-
"epoch": 0.76,
|
| 2562 |
-
"learning_rate": 7.63811414823016e-06,
|
| 2563 |
-
"loss": 0.0063,
|
| 2564 |
-
"r_loss": 0.00443157646805048,
|
| 2565 |
-
"s_loss": 0.0008406736305914819,
|
| 2566 |
-
"step": 25600
|
| 2567 |
-
},
|
| 2568 |
-
{
|
| 2569 |
-
"epoch": 0.77,
|
| 2570 |
-
"learning_rate": 7.464158408885898e-06,
|
| 2571 |
-
"loss": 0.0061,
|
| 2572 |
-
"r_loss": 0.005125071853399277,
|
| 2573 |
-
"s_loss": 0.0008355857571586967,
|
| 2574 |
-
"step": 25700
|
| 2575 |
-
},
|
| 2576 |
-
{
|
| 2577 |
-
"epoch": 0.77,
|
| 2578 |
-
"learning_rate": 7.291547937993373e-06,
|
| 2579 |
-
"loss": 0.0062,
|
| 2580 |
-
"r_loss": 0.00465528666973114,
|
| 2581 |
-
"s_loss": 0.000775384483858943,
|
| 2582 |
-
"step": 25800
|
| 2583 |
-
},
|
| 2584 |
-
{
|
| 2585 |
-
"epoch": 0.77,
|
| 2586 |
-
"learning_rate": 7.1203135492869385e-06,
|
| 2587 |
-
"loss": 0.0061,
|
| 2588 |
-
"r_loss": 0.004815374501049519,
|
| 2589 |
-
"s_loss": 0.0007990074809640646,
|
| 2590 |
-
"step": 25900
|
| 2591 |
-
},
|
| 2592 |
-
{
|
| 2593 |
-
"epoch": 0.78,
|
| 2594 |
-
"learning_rate": 6.952177025557144e-06,
|
| 2595 |
-
"loss": 0.0062,
|
| 2596 |
-
"r_loss": 0.004868610296398401,
|
| 2597 |
-
"s_loss": 0.0007737652049399912,
|
| 2598 |
-
"step": 26000
|
| 2599 |
-
},
|
| 2600 |
-
{
|
| 2601 |
-
"epoch": 0.78,
|
| 2602 |
-
"eval_loss": 0.0064270892180502415,
|
| 2603 |
-
"eval_r_loss": 0.005634445697069168,
|
| 2604 |
-
"eval_runtime": 9.0404,
|
| 2605 |
-
"eval_s_loss": 0.0007926435209810734,
|
| 2606 |
-
"eval_samples_per_second": 74.333,
|
| 2607 |
-
"eval_steps_per_second": 74.333,
|
| 2608 |
-
"step": 26000
|
| 2609 |
-
},
|
| 2610 |
-
{
|
| 2611 |
-
"epoch": 0.78,
|
| 2612 |
-
"learning_rate": 6.78377173545463e-06,
|
| 2613 |
-
"loss": 0.0062,
|
| 2614 |
-
"r_loss": 0.004927606321871281,
|
| 2615 |
-
"s_loss": 0.0008687089430168271,
|
| 2616 |
-
"step": 26100
|
| 2617 |
-
},
|
| 2618 |
-
{
|
| 2619 |
-
"epoch": 0.78,
|
| 2620 |
-
"learning_rate": 6.616833173725176e-06,
|
| 2621 |
-
"loss": 0.0062,
|
| 2622 |
-
"r_loss": 0.004880106542259455,
|
| 2623 |
-
"s_loss": 0.0008371942676603794,
|
| 2624 |
-
"step": 26200
|
| 2625 |
-
},
|
| 2626 |
-
{
|
| 2627 |
-
"epoch": 0.78,
|
| 2628 |
-
"learning_rate": 6.451391141576487e-06,
|
| 2629 |
-
"loss": 0.0062,
|
| 2630 |
-
"r_loss": 0.004924027249217033,
|
| 2631 |
-
"s_loss": 0.0008956523961387575,
|
| 2632 |
-
"step": 26300
|
| 2633 |
-
},
|
| 2634 |
-
{
|
| 2635 |
-
"epoch": 0.79,
|
| 2636 |
-
"learning_rate": 6.287475173061754e-06,
|
| 2637 |
-
"loss": 0.0061,
|
| 2638 |
-
"r_loss": 0.005853409878909588,
|
| 2639 |
-
"s_loss": 0.0008864904521033168,
|
| 2640 |
-
"step": 26400
|
| 2641 |
-
},
|
| 2642 |
-
{
|
| 2643 |
-
"epoch": 0.79,
|
| 2644 |
-
"learning_rate": 6.12511452980738e-06,
|
| 2645 |
-
"loss": 0.0062,
|
| 2646 |
-
"r_loss": 0.0052686696872115135,
|
| 2647 |
-
"s_loss": 0.000804445066023618,
|
| 2648 |
-
"step": 26500
|
| 2649 |
-
},
|
| 2650 |
-
{
|
| 2651 |
-
"epoch": 0.79,
|
| 2652 |
-
"eval_loss": 0.006505042780190706,
|
| 2653 |
-
"eval_r_loss": 0.005706477910280228,
|
| 2654 |
-
"eval_runtime": 8.801,
|
| 2655 |
-
"eval_s_loss": 0.0007985649281181395,
|
| 2656 |
-
"eval_samples_per_second": 76.355,
|
| 2657 |
-
"eval_steps_per_second": 76.355,
|
| 2658 |
-
"step": 26500
|
| 2659 |
-
},
|
| 2660 |
-
{
|
| 2661 |
-
"epoch": 0.79,
|
| 2662 |
-
"learning_rate": 5.9643381957892725e-06,
|
| 2663 |
-
"loss": 0.0062,
|
| 2664 |
-
"r_loss": 0.0047812857665121555,
|
| 2665 |
-
"s_loss": 0.0008621865999884903,
|
| 2666 |
-
"step": 26600
|
| 2667 |
-
},
|
| 2668 |
-
{
|
| 2669 |
-
"epoch": 0.8,
|
| 2670 |
-
"learning_rate": 5.805174872158762e-06,
|
| 2671 |
-
"loss": 0.0061,
|
| 2672 |
-
"r_loss": 0.004578050691634417,
|
| 2673 |
-
"s_loss": 0.0008361663785763085,
|
| 2674 |
-
"step": 26700
|
| 2675 |
-
},
|
| 2676 |
-
{
|
| 2677 |
-
"epoch": 0.8,
|
| 2678 |
-
"learning_rate": 5.647652972118998e-06,
|
| 2679 |
-
"loss": 0.0062,
|
| 2680 |
-
"r_loss": 0.0053869327530264854,
|
| 2681 |
-
"s_loss": 0.0008459505042992532,
|
| 2682 |
-
"step": 26800
|
| 2683 |
-
},
|
| 2684 |
-
{
|
| 2685 |
-
"epoch": 0.8,
|
| 2686 |
-
"learning_rate": 5.491800615852702e-06,
|
| 2687 |
-
"loss": 0.0061,
|
| 2688 |
-
"r_loss": 0.006079402752220631,
|
| 2689 |
-
"s_loss": 0.0008894907077774405,
|
| 2690 |
-
"step": 26900
|
| 2691 |
-
},
|
| 2692 |
-
{
|
| 2693 |
-
"epoch": 0.81,
|
| 2694 |
-
"learning_rate": 5.337645625502312e-06,
|
| 2695 |
-
"loss": 0.0061,
|
| 2696 |
-
"r_loss": 0.005296847317367792,
|
| 2697 |
-
"s_loss": 0.0007577301003038883,
|
| 2698 |
-
"step": 27000
|
| 2699 |
-
},
|
| 2700 |
-
{
|
| 2701 |
-
"epoch": 0.81,
|
| 2702 |
-
"eval_loss": 0.006496031768620014,
|
| 2703 |
-
"eval_r_loss": 0.00570162758231163,
|
| 2704 |
-
"eval_runtime": 8.61,
|
| 2705 |
-
"eval_s_loss": 0.0007944039534777403,
|
| 2706 |
-
"eval_samples_per_second": 78.048,
|
| 2707 |
-
"eval_steps_per_second": 78.048,
|
| 2708 |
-
"step": 27000
|
| 2709 |
-
},
|
| 2710 |
-
{
|
| 2711 |
-
"epoch": 0.81,
|
| 2712 |
-
"learning_rate": 5.18521552020322e-06,
|
| 2713 |
-
"loss": 0.0062,
|
| 2714 |
-
"r_loss": 0.0052690450102090836,
|
| 2715 |
-
"s_loss": 0.0009067388018593192,
|
| 2716 |
-
"step": 27100
|
| 2717 |
-
},
|
| 2718 |
-
{
|
| 2719 |
-
"epoch": 0.81,
|
| 2720 |
-
"learning_rate": 5.034537511171238e-06,
|
| 2721 |
-
"loss": 0.0061,
|
| 2722 |
-
"r_loss": 0.005185704678297043,
|
| 2723 |
-
"s_loss": 0.0008527652826160192,
|
| 2724 |
-
"step": 27200
|
| 2725 |
-
},
|
| 2726 |
-
{
|
| 2727 |
-
"epoch": 0.81,
|
| 2728 |
-
"learning_rate": 4.885638496844903e-06,
|
| 2729 |
-
"loss": 0.006,
|
| 2730 |
-
"r_loss": 0.005816085264086723,
|
| 2731 |
-
"s_loss": 0.0008826929260976613,
|
| 2732 |
-
"step": 27300
|
| 2733 |
-
},
|
| 2734 |
-
{
|
| 2735 |
-
"epoch": 0.82,
|
| 2736 |
-
"learning_rate": 4.738545058083715e-06,
|
| 2737 |
-
"loss": 0.006,
|
| 2738 |
-
"r_loss": 0.004605771973729134,
|
| 2739 |
-
"s_loss": 0.0008187288185581565,
|
| 2740 |
-
"step": 27400
|
| 2741 |
-
},
|
| 2742 |
-
{
|
| 2743 |
-
"epoch": 0.82,
|
| 2744 |
-
"learning_rate": 4.593283453422984e-06,
|
| 2745 |
-
"loss": 0.0061,
|
| 2746 |
-
"r_loss": 0.0052945734933018684,
|
| 2747 |
-
"s_loss": 0.0008752761059440672,
|
| 2748 |
-
"step": 27500
|
| 2749 |
-
},
|
| 2750 |
-
{
|
| 2751 |
-
"epoch": 0.82,
|
| 2752 |
-
"eval_loss": 0.006313642952591181,
|
| 2753 |
-
"eval_r_loss": 0.005525515414774418,
|
| 2754 |
-
"eval_runtime": 10.8866,
|
| 2755 |
-
"eval_s_loss": 0.0007881273631937802,
|
| 2756 |
-
"eval_samples_per_second": 61.727,
|
| 2757 |
-
"eval_steps_per_second": 61.727,
|
| 2758 |
-
"step": 27500
|
| 2759 |
-
},
|
| 2760 |
-
{
|
| 2761 |
-
"epoch": 0.82,
|
| 2762 |
-
"learning_rate": 4.449879614386313e-06,
|
| 2763 |
-
"loss": 0.0061,
|
| 2764 |
-
"r_loss": 0.005148299969732761,
|
| 2765 |
-
"s_loss": 0.0007778692524880171,
|
| 2766 |
-
"step": 27600
|
| 2767 |
-
},
|
| 2768 |
-
{
|
| 2769 |
-
"epoch": 0.83,
|
| 2770 |
-
"learning_rate": 4.308359140856364e-06,
|
| 2771 |
-
"loss": 0.006,
|
| 2772 |
-
"r_loss": 0.004953067749738693,
|
| 2773 |
-
"s_loss": 0.0008753555594012141,
|
| 2774 |
-
"step": 27700
|
| 2775 |
-
},
|
| 2776 |
-
{
|
| 2777 |
-
"epoch": 0.83,
|
| 2778 |
-
"learning_rate": 4.168747296504881e-06,
|
| 2779 |
-
"loss": 0.0061,
|
| 2780 |
-
"r_loss": 0.005346274934709072,
|
| 2781 |
-
"s_loss": 0.0008310099365189672,
|
| 2782 |
-
"step": 27800
|
| 2783 |
-
},
|
| 2784 |
-
{
|
| 2785 |
-
"epoch": 0.83,
|
| 2786 |
-
"learning_rate": 4.031069004282739e-06,
|
| 2787 |
-
"loss": 0.006,
|
| 2788 |
-
"r_loss": 0.00487141078338027,
|
| 2789 |
-
"s_loss": 0.0008442049147561193,
|
| 2790 |
-
"step": 27900
|
| 2791 |
-
},
|
| 2792 |
-
{
|
| 2793 |
-
"epoch": 0.84,
|
| 2794 |
-
"learning_rate": 3.895348841970758e-06,
|
| 2795 |
-
"loss": 0.0059,
|
| 2796 |
-
"r_loss": 0.004993779119104147,
|
| 2797 |
-
"s_loss": 0.0007509638089686632,
|
| 2798 |
-
"step": 28000
|
| 2799 |
-
},
|
| 2800 |
-
{
|
| 2801 |
-
"epoch": 0.84,
|
| 2802 |
-
"eval_loss": 0.006438162177801132,
|
| 2803 |
-
"eval_r_loss": 0.005650770850479603,
|
| 2804 |
-
"eval_runtime": 8.473,
|
| 2805 |
-
"eval_s_loss": 0.0007873910944908857,
|
| 2806 |
-
"eval_samples_per_second": 79.311,
|
| 2807 |
-
"eval_steps_per_second": 79.311,
|
| 2808 |
-
"step": 28000
|
| 2809 |
-
},
|
| 2810 |
-
{
|
| 2811 |
-
"epoch": 0.84,
|
| 2812 |
-
"learning_rate": 3.7616110377922263e-06,
|
| 2813 |
-
"loss": 0.0061,
|
| 2814 |
-
"r_loss": 0.006000261753797531,
|
| 2815 |
-
"s_loss": 0.0009082874748855829,
|
| 2816 |
-
"step": 28100
|
| 2817 |
-
},
|
| 2818 |
-
{
|
| 2819 |
-
"epoch": 0.84,
|
| 2820 |
-
"learning_rate": 3.6298794660877154e-06,
|
| 2821 |
-
"loss": 0.0061,
|
| 2822 |
-
"r_loss": 0.006097930949181318,
|
| 2823 |
-
"s_loss": 0.0009037306881509721,
|
| 2824 |
-
"step": 28200
|
| 2825 |
-
},
|
| 2826 |
-
{
|
| 2827 |
-
"epoch": 0.84,
|
| 2828 |
-
"learning_rate": 3.50017764305319e-06,
|
| 2829 |
-
"loss": 0.0061,
|
| 2830 |
-
"r_loss": 0.006982284598052502,
|
| 2831 |
-
"s_loss": 0.0008789664716459811,
|
| 2832 |
-
"step": 28300
|
| 2833 |
-
},
|
| 2834 |
-
{
|
| 2835 |
-
"epoch": 0.85,
|
| 2836 |
-
"learning_rate": 3.3725287225419365e-06,
|
| 2837 |
-
"loss": 0.0061,
|
| 2838 |
-
"r_loss": 0.004647083580493927,
|
| 2839 |
-
"s_loss": 0.0008192590321414173,
|
| 2840 |
-
"step": 28400
|
| 2841 |
-
},
|
| 2842 |
-
{
|
| 2843 |
-
"epoch": 0.85,
|
| 2844 |
-
"learning_rate": 3.2469554919312733e-06,
|
| 2845 |
-
"loss": 0.006,
|
| 2846 |
-
"r_loss": 0.00463445670902729,
|
| 2847 |
-
"s_loss": 0.0008258245070464909,
|
| 2848 |
-
"step": 28500
|
| 2849 |
-
},
|
| 2850 |
-
{
|
| 2851 |
-
"epoch": 0.85,
|
| 2852 |
-
"eval_loss": 0.006398391909897327,
|
| 2853 |
-
"eval_r_loss": 0.005606560967862606,
|
| 2854 |
-
"eval_runtime": 8.6124,
|
| 2855 |
-
"eval_s_loss": 0.0007918307092040777,
|
| 2856 |
-
"eval_samples_per_second": 78.027,
|
| 2857 |
-
"eval_steps_per_second": 78.027,
|
| 2858 |
-
"step": 28500
|
| 2859 |
-
},
|
| 2860 |
-
{
|
| 2861 |
-
"epoch": 0.85,
|
| 2862 |
-
"learning_rate": 3.12470466097668e-06,
|
| 2863 |
-
"loss": 0.0059,
|
| 2864 |
-
"r_loss": 0.004881501197814941,
|
| 2865 |
-
"s_loss": 0.0008381219813600183,
|
| 2866 |
-
"step": 28600
|
| 2867 |
-
},
|
| 2868 |
-
{
|
| 2869 |
-
"epoch": 0.86,
|
| 2870 |
-
"learning_rate": 3.0033283767617047e-06,
|
| 2871 |
-
"loss": 0.006,
|
| 2872 |
-
"r_loss": 0.004456360824406147,
|
| 2873 |
-
"s_loss": 0.0007237752433866262,
|
| 2874 |
-
"step": 28700
|
| 2875 |
-
},
|
| 2876 |
-
{
|
| 2877 |
-
"epoch": 0.86,
|
| 2878 |
-
"learning_rate": 2.8840936906237913e-06,
|
| 2879 |
-
"loss": 0.006,
|
| 2880 |
-
"r_loss": 0.005578363314270973,
|
| 2881 |
-
"s_loss": 0.0008311424753628671,
|
| 2882 |
-
"step": 28800
|
| 2883 |
-
},
|
| 2884 |
-
{
|
| 2885 |
-
"epoch": 0.86,
|
| 2886 |
-
"learning_rate": 2.7670218878646273e-06,
|
| 2887 |
-
"loss": 0.0059,
|
| 2888 |
-
"r_loss": 0.005354044958949089,
|
| 2889 |
-
"s_loss": 0.0008516995585523546,
|
| 2890 |
-
"step": 28900
|
| 2891 |
-
},
|
| 2892 |
-
{
|
| 2893 |
-
"epoch": 0.87,
|
| 2894 |
-
"learning_rate": 2.6521338676765317e-06,
|
| 2895 |
-
"loss": 0.006,
|
| 2896 |
-
"r_loss": 0.0043981922790408134,
|
| 2897 |
-
"s_loss": 0.0007206485024653375,
|
| 2898 |
-
"step": 29000
|
| 2899 |
-
},
|
| 2900 |
-
{
|
| 2901 |
-
"epoch": 0.87,
|
| 2902 |
-
"eval_loss": 0.006455121096223593,
|
| 2903 |
-
"eval_r_loss": 0.00566498190164566,
|
| 2904 |
-
"eval_runtime": 10.8485,
|
| 2905 |
-
"eval_s_loss": 0.0007901391945779324,
|
| 2906 |
-
"eval_samples_per_second": 61.944,
|
| 2907 |
-
"eval_steps_per_second": 61.944,
|
| 2908 |
-
"step": 29000
|
| 2909 |
-
},
|
| 2910 |
-
{
|
| 2911 |
-
"epoch": 0.87,
|
| 2912 |
-
"learning_rate": 2.5394501394116805e-06,
|
| 2913 |
-
"loss": 0.006,
|
| 2914 |
-
"r_loss": 0.0046515436843037605,
|
| 2915 |
-
"s_loss": 0.0008327921386808157,
|
| 2916 |
-
"step": 29100
|
| 2917 |
-
},
|
| 2918 |
-
{
|
| 2919 |
-
"epoch": 0.87,
|
| 2920 |
-
"learning_rate": 2.4289908189208086e-06,
|
| 2921 |
-
"loss": 0.006,
|
| 2922 |
-
"r_loss": 0.005409231409430504,
|
| 2923 |
-
"s_loss": 0.0008956742822192609,
|
| 2924 |
-
"step": 29200
|
| 2925 |
-
},
|
| 2926 |
-
{
|
| 2927 |
-
"epoch": 0.87,
|
| 2928 |
-
"learning_rate": 2.320775624962243e-06,
|
| 2929 |
-
"loss": 0.0059,
|
| 2930 |
-
"r_loss": 0.005478174425661564,
|
| 2931 |
-
"s_loss": 0.0007779388688504696,
|
| 2932 |
-
"step": 29300
|
| 2933 |
-
},
|
| 2934 |
-
{
|
| 2935 |
-
"epoch": 0.88,
|
| 2936 |
-
"learning_rate": 2.214823875681753e-06,
|
| 2937 |
-
"loss": 0.006,
|
| 2938 |
-
"r_loss": 0.005899087525904179,
|
| 2939 |
-
"s_loss": 0.0008006141288205981,
|
| 2940 |
-
"step": 29400
|
| 2941 |
-
},
|
| 2942 |
-
{
|
| 2943 |
-
"epoch": 0.88,
|
| 2944 |
-
"learning_rate": 2.1111544851639887e-06,
|
| 2945 |
-
"loss": 0.006,
|
| 2946 |
-
"r_loss": 0.004816343542188406,
|
| 2947 |
-
"s_loss": 0.0008681662729941308,
|
| 2948 |
-
"step": 29500
|
| 2949 |
-
},
|
| 2950 |
-
{
|
| 2951 |
-
"epoch": 0.88,
|
| 2952 |
-
"eval_loss": 0.006466238759458065,
|
| 2953 |
-
"eval_r_loss": 0.005683359690010548,
|
| 2954 |
-
"eval_runtime": 8.6417,
|
| 2955 |
-
"eval_s_loss": 0.0007828791276551783,
|
| 2956 |
-
"eval_samples_per_second": 77.763,
|
| 2957 |
-
"eval_steps_per_second": 77.763,
|
| 2958 |
-
"step": 29500
|
| 2959 |
-
},
|
| 2960 |
-
{
|
| 2961 |
-
"epoch": 0.88,
|
| 2962 |
-
"learning_rate": 2.009785960055983e-06,
|
| 2963 |
-
"loss": 0.0059,
|
| 2964 |
-
"r_loss": 0.005228263325989246,
|
| 2965 |
-
"s_loss": 0.0008743102662265301,
|
| 2966 |
-
"step": 29600
|
| 2967 |
-
},
|
| 2968 |
-
{
|
| 2969 |
-
"epoch": 0.89,
|
| 2970 |
-
"learning_rate": 1.910736396263449e-06,
|
| 2971 |
-
"loss": 0.006,
|
| 2972 |
-
"r_loss": 0.005140687804669142,
|
| 2973 |
-
"s_loss": 0.0007702410221099854,
|
| 2974 |
-
"step": 29700
|
| 2975 |
-
},
|
| 2976 |
-
{
|
| 2977 |
-
"epoch": 0.89,
|
| 2978 |
-
"learning_rate": 1.8140234757203395e-06,
|
| 2979 |
-
"loss": 0.006,
|
| 2980 |
-
"r_loss": 0.004959780722856522,
|
| 2981 |
-
"s_loss": 0.0007955725886859,
|
| 2982 |
-
"step": 29800
|
| 2983 |
-
},
|
| 2984 |
-
{
|
| 2985 |
-
"epoch": 0.89,
|
| 2986 |
-
"learning_rate": 1.7196644632323532e-06,
|
| 2987 |
-
"loss": 0.0059,
|
| 2988 |
-
"r_loss": 0.00630287267267704,
|
| 2989 |
-
"s_loss": 0.0008392567397095263,
|
| 2990 |
-
"step": 29900
|
| 2991 |
-
},
|
| 2992 |
-
{
|
| 2993 |
-
"epoch": 0.9,
|
| 2994 |
-
"learning_rate": 1.6276762033949055e-06,
|
| 2995 |
-
"loss": 0.006,
|
| 2996 |
-
"r_loss": 0.00496282521635294,
|
| 2997 |
-
"s_loss": 0.0008307393291033804,
|
| 2998 |
-
"step": 30000
|
| 2999 |
-
},
|
| 3000 |
-
{
|
| 3001 |
-
"epoch": 0.9,
|
| 3002 |
-
"eval_loss": 0.006491221487522125,
|
| 3003 |
-
"eval_r_loss": 0.005704117473214865,
|
| 3004 |
-
"eval_runtime": 13.2979,
|
| 3005 |
-
"eval_s_loss": 0.0007871038978919387,
|
| 3006 |
-
"eval_samples_per_second": 50.534,
|
| 3007 |
-
"eval_steps_per_second": 50.534,
|
| 3008 |
-
"step": 30000
|
| 3009 |
-
},
|
| 3010 |
-
{
|
| 3011 |
-
"epoch": 0.9,
|
| 3012 |
-
"learning_rate": 1.5380751175860619e-06,
|
| 3013 |
-
"loss": 0.006,
|
| 3014 |
-
"r_loss": 0.004992773290723562,
|
| 3015 |
-
"s_loss": 0.0007617148803547025,
|
| 3016 |
-
"step": 30100
|
| 3017 |
-
},
|
| 3018 |
-
{
|
| 3019 |
-
"epoch": 0.9,
|
| 3020 |
-
"learning_rate": 1.4508772010350967e-06,
|
| 3021 |
-
"loss": 0.0059,
|
| 3022 |
-
"r_loss": 0.005444124806672335,
|
| 3023 |
-
"s_loss": 0.000857758685015142,
|
| 3024 |
-
"step": 30200
|
| 3025 |
-
},
|
| 3026 |
-
{
|
| 3027 |
-
"epoch": 0.9,
|
| 3028 |
-
"learning_rate": 1.3660980199670624e-06,
|
| 3029 |
-
"loss": 0.006,
|
| 3030 |
-
"r_loss": 0.004532184451818466,
|
| 3031 |
-
"s_loss": 0.0007677034009248018,
|
| 3032 |
-
"step": 30300
|
| 3033 |
-
},
|
| 3034 |
-
{
|
| 3035 |
-
"epoch": 0.91,
|
| 3036 |
-
"learning_rate": 1.2837527088239886e-06,
|
| 3037 |
-
"loss": 0.0058,
|
| 3038 |
-
"r_loss": 0.003988460171967745,
|
| 3039 |
-
"s_loss": 0.0007943719392642379,
|
| 3040 |
-
"step": 30400
|
| 3041 |
-
},
|
| 3042 |
-
{
|
| 3043 |
-
"epoch": 0.91,
|
| 3044 |
-
"learning_rate": 1.2038559675631167e-06,
|
| 3045 |
-
"loss": 0.006,
|
| 3046 |
-
"r_loss": 0.005079349968582392,
|
| 3047 |
-
"s_loss": 0.0008392308373004198,
|
| 3048 |
-
"step": 30500
|
| 3049 |
-
},
|
| 3050 |
-
{
|
| 3051 |
-
"epoch": 0.91,
|
| 3052 |
-
"eval_loss": 0.0063784122467041016,
|
| 3053 |
-
"eval_r_loss": 0.005593848414719105,
|
| 3054 |
-
"eval_runtime": 11.2008,
|
| 3055 |
-
"eval_s_loss": 0.0007845640648156404,
|
| 3056 |
-
"eval_samples_per_second": 59.996,
|
| 3057 |
-
"eval_steps_per_second": 59.996,
|
| 3058 |
-
"step": 30500
|
| 3059 |
-
},
|
| 3060 |
-
{
|
| 3061 |
-
"epoch": 0.91,
|
| 3062 |
-
"learning_rate": 1.1264220590327507e-06,
|
| 3063 |
-
"loss": 0.0061,
|
| 3064 |
-
"r_loss": 0.004688158631324768,
|
| 3065 |
-
"s_loss": 0.0008526835008524358,
|
| 3066 |
-
"step": 30600
|
| 3067 |
-
},
|
| 3068 |
-
{
|
| 3069 |
-
"epoch": 0.92,
|
| 3070 |
-
"learning_rate": 1.0514648064260858e-06,
|
| 3071 |
-
"loss": 0.0059,
|
| 3072 |
-
"r_loss": 0.005976270884275436,
|
| 3073 |
-
"s_loss": 0.0009033031528815627,
|
| 3074 |
-
"step": 30700
|
| 3075 |
-
},
|
| 3076 |
-
{
|
| 3077 |
-
"epoch": 0.92,
|
| 3078 |
-
"learning_rate": 9.789975908135684e-07,
|
| 3079 |
-
"loss": 0.006,
|
| 3080 |
-
"r_loss": 0.004994697868824005,
|
| 3081 |
-
"s_loss": 0.0007998015498742461,
|
| 3082 |
-
"step": 30800
|
| 3083 |
-
},
|
| 3084 |
-
{
|
| 3085 |
-
"epoch": 0.92,
|
| 3086 |
-
"learning_rate": 9.09033348754143e-07,
|
| 3087 |
-
"loss": 0.0059,
|
| 3088 |
-
"r_loss": 0.004837275482714176,
|
| 3089 |
-
"s_loss": 0.0008055656799115241,
|
| 3090 |
-
"step": 30900
|
| 3091 |
-
},
|
| 3092 |
-
{
|
| 3093 |
-
"epoch": 0.93,
|
| 3094 |
-
"learning_rate": 8.415845699858748e-07,
|
| 3095 |
-
"loss": 0.0059,
|
| 3096 |
-
"r_loss": 0.00475228950381279,
|
| 3097 |
-
"s_loss": 0.0007835312280803919,
|
| 3098 |
-
"step": 31000
|
| 3099 |
-
},
|
| 3100 |
-
{
|
| 3101 |
-
"epoch": 0.93,
|
| 3102 |
-
"eval_loss": 0.006395082455128431,
|
| 3103 |
-
"eval_r_loss": 0.005607670173048973,
|
| 3104 |
-
"eval_runtime": 9.003,
|
| 3105 |
-
"eval_s_loss": 0.0007874123984947801,
|
| 3106 |
-
"eval_samples_per_second": 74.642,
|
| 3107 |
-
"eval_steps_per_second": 74.642,
|
| 3108 |
-
"step": 31000
|
| 3109 |
-
},
|
| 3110 |
-
{
|
| 3111 |
-
"epoch": 0.93,
|
| 3112 |
-
"learning_rate": 7.766632951963343e-07,
|
| 3113 |
-
"loss": 0.0059,
|
| 3114 |
-
"r_loss": 0.005764458328485489,
|
| 3115 |
-
"s_loss": 0.0008497489034198225,
|
| 3116 |
-
"step": 31100
|
| 3117 |
-
},
|
| 3118 |
-
{
|
| 3119 |
-
"epoch": 0.93,
|
| 3120 |
-
"learning_rate": 7.142811138731459e-07,
|
| 3121 |
-
"loss": 0.006,
|
| 3122 |
-
"r_loss": 0.005635739304125309,
|
| 3123 |
-
"s_loss": 0.0007945407414808869,
|
| 3124 |
-
"step": 31200
|
| 3125 |
-
},
|
| 3126 |
-
{
|
| 3127 |
-
"epoch": 0.93,
|
| 3128 |
-
"learning_rate": 6.544491622350635e-07,
|
| 3129 |
-
"loss": 0.006,
|
| 3130 |
-
"r_loss": 0.005129328928887844,
|
| 3131 |
-
"s_loss": 0.0008712293347343802,
|
| 3132 |
-
"step": 31300
|
| 3133 |
-
},
|
| 3134 |
-
{
|
| 3135 |
-
"epoch": 0.94,
|
| 3136 |
-
"learning_rate": 5.971781212439903e-07,
|
| 3137 |
-
"loss": 0.0059,
|
| 3138 |
-
"r_loss": 0.00499305035918951,
|
| 3139 |
-
"s_loss": 0.0008739501936361194,
|
| 3140 |
-
"step": 31400
|
| 3141 |
-
},
|
| 3142 |
-
{
|
| 3143 |
-
"epoch": 0.94,
|
| 3144 |
-
"learning_rate": 5.42478214698256e-07,
|
| 3145 |
-
"loss": 0.006,
|
| 3146 |
-
"r_loss": 0.005520367994904518,
|
| 3147 |
-
"s_loss": 0.0008639748557470739,
|
| 3148 |
-
"step": 31500
|
| 3149 |
-
},
|
| 3150 |
-
{
|
| 3151 |
-
"epoch": 0.94,
|
| 3152 |
-
"eval_loss": 0.006423806771636009,
|
| 3153 |
-
"eval_r_loss": 0.005639917217195034,
|
| 3154 |
-
"eval_runtime": 9.5481,
|
| 3155 |
-
"eval_s_loss": 0.0007838893216103315,
|
| 3156 |
-
"eval_samples_per_second": 70.381,
|
| 3157 |
-
"eval_steps_per_second": 70.381,
|
| 3158 |
-
"step": 31500
|
| 3159 |
-
},
|
| 3160 |
-
{
|
| 3161 |
-
"epoch": 0.94,
|
| 3162 |
-
"learning_rate": 4.903592074074914e-07,
|
| 3163 |
-
"loss": 0.0059,
|
| 3164 |
-
"r_loss": 0.004892979748547077,
|
| 3165 |
-
"s_loss": 0.0007891397108323872,
|
| 3166 |
-
"step": 31600
|
| 3167 |
-
},
|
| 3168 |
-
{
|
| 3169 |
-
"epoch": 0.95,
|
| 3170 |
-
"learning_rate": 4.408304034494748e-07,
|
| 3171 |
-
"loss": 0.006,
|
| 3172 |
-
"r_loss": 0.00487134512513876,
|
| 3173 |
-
"s_loss": 0.0007864002254791558,
|
| 3174 |
-
"step": 31700
|
| 3175 |
-
},
|
| 3176 |
-
{
|
| 3177 |
-
"epoch": 0.95,
|
| 3178 |
-
"learning_rate": 3.9390064450918195e-07,
|
| 3179 |
-
"loss": 0.0059,
|
| 3180 |
-
"r_loss": 0.004203725606203079,
|
| 3181 |
-
"s_loss": 0.0007678180118091404,
|
| 3182 |
-
"step": 31800
|
| 3183 |
-
},
|
| 3184 |
-
{
|
| 3185 |
-
"epoch": 0.95,
|
| 3186 |
-
"learning_rate": 3.495783083004273e-07,
|
| 3187 |
-
"loss": 0.0059,
|
| 3188 |
-
"r_loss": 0.005060626659542322,
|
| 3189 |
-
"s_loss": 0.0008796448237262666,
|
| 3190 |
-
"step": 31900
|
| 3191 |
-
},
|
| 3192 |
-
{
|
| 3193 |
-
"epoch": 0.95,
|
| 3194 |
-
"learning_rate": 3.0787130707028155e-07,
|
| 3195 |
-
"loss": 0.0059,
|
| 3196 |
-
"r_loss": 0.00450053671374917,
|
| 3197 |
-
"s_loss": 0.0007761311717331409,
|
| 3198 |
-
"step": 32000
|
| 3199 |
-
},
|
| 3200 |
-
{
|
| 3201 |
-
"epoch": 0.95,
|
| 3202 |
-
"eval_loss": 0.006359361112117767,
|
| 3203 |
-
"eval_r_loss": 0.005576182622462511,
|
| 3204 |
-
"eval_runtime": 8.5745,
|
| 3205 |
-
"eval_s_loss": 0.0007831782568246126,
|
| 3206 |
-
"eval_samples_per_second": 78.372,
|
| 3207 |
-
"eval_steps_per_second": 78.372,
|
| 3208 |
-
"step": 32000
|
| 3209 |
-
},
|
| 3210 |
-
{
|
| 3211 |
-
"epoch": 0.96,
|
| 3212 |
-
"learning_rate": 2.687870861866354e-07,
|
| 3213 |
-
"loss": 0.0058,
|
| 3214 |
-
"r_loss": 0.004816841334104538,
|
| 3215 |
-
"s_loss": 0.0008092151256278157,
|
| 3216 |
-
"step": 32100
|
| 3217 |
-
},
|
| 3218 |
-
{
|
| 3219 |
-
"epoch": 0.96,
|
| 3220 |
-
"learning_rate": 2.3233262280905887e-07,
|
| 3221 |
-
"loss": 0.0058,
|
| 3222 |
-
"r_loss": 0.005017046816647053,
|
| 3223 |
-
"s_loss": 0.0008538334514014423,
|
| 3224 |
-
"step": 32200
|
| 3225 |
-
},
|
| 3226 |
-
{
|
| 3227 |
-
"epoch": 0.96,
|
| 3228 |
-
"learning_rate": 1.985144246432896e-07,
|
| 3229 |
-
"loss": 0.0059,
|
| 3230 |
-
"r_loss": 0.004773393739014864,
|
| 3231 |
-
"s_loss": 0.0008774587768130004,
|
| 3232 |
-
"step": 32300
|
| 3233 |
-
},
|
| 3234 |
-
{
|
| 3235 |
-
"epoch": 0.97,
|
| 3236 |
-
"learning_rate": 1.673385287794771e-07,
|
| 3237 |
-
"loss": 0.0059,
|
| 3238 |
-
"r_loss": 0.004963357001543045,
|
| 3239 |
-
"s_loss": 0.0008022256079129875,
|
| 3240 |
-
"step": 32400
|
| 3241 |
-
},
|
| 3242 |
-
{
|
| 3243 |
-
"epoch": 0.97,
|
| 3244 |
-
"learning_rate": 1.3881050061448963e-07,
|
| 3245 |
-
"loss": 0.0058,
|
| 3246 |
-
"r_loss": 0.005114908795803785,
|
| 3247 |
-
"s_loss": 0.0008205736521631479,
|
| 3248 |
-
"step": 32500
|
| 3249 |
-
},
|
| 3250 |
-
{
|
| 3251 |
-
"epoch": 0.97,
|
| 3252 |
-
"eval_loss": 0.006366991437971592,
|
| 3253 |
-
"eval_r_loss": 0.005584825295954943,
|
| 3254 |
-
"eval_runtime": 8.9973,
|
| 3255 |
-
"eval_s_loss": 0.0007821662584319711,
|
| 3256 |
-
"eval_samples_per_second": 74.689,
|
| 3257 |
-
"eval_steps_per_second": 74.689,
|
| 3258 |
-
"step": 32500
|
| 3259 |
-
},
|
| 3260 |
-
{
|
| 3261 |
-
"epoch": 0.97,
|
| 3262 |
-
"learning_rate": 1.1342690454311188e-07,
|
| 3263 |
-
"loss": 0.006,
|
| 3264 |
-
"r_loss": 0.0049808090552687645,
|
| 3265 |
-
"s_loss": 0.0008341091452166438,
|
| 3266 |
-
"step": 32600
|
| 3267 |
-
},
|
| 3268 |
-
{
|
| 3269 |
-
"epoch": 0.98,
|
| 3270 |
-
"learning_rate": 9.015622251951239e-08,
|
| 3271 |
-
"loss": 0.006,
|
| 3272 |
-
"r_loss": 0.005653849337249994,
|
| 3273 |
-
"s_loss": 0.0008855736814439297,
|
| 3274 |
-
"step": 32700
|
| 3275 |
-
},
|
| 3276 |
-
{
|
| 3277 |
-
"epoch": 0.98,
|
| 3278 |
-
"learning_rate": 6.95471864728997e-08,
|
| 3279 |
-
"loss": 0.0059,
|
| 3280 |
-
"r_loss": 0.006156204268336296,
|
| 3281 |
-
"s_loss": 0.0008836152264848351,
|
| 3282 |
-
"step": 32800
|
| 3283 |
-
},
|
| 3284 |
-
{
|
| 3285 |
-
"epoch": 0.98,
|
| 3286 |
-
"learning_rate": 5.1603475446377335e-08,
|
| 3287 |
-
"loss": 0.0059,
|
| 3288 |
-
"r_loss": 0.005304061807692051,
|
| 3289 |
-
"s_loss": 0.0009092881809920073,
|
| 3290 |
-
"step": 32900
|
| 3291 |
-
},
|
| 3292 |
-
{
|
| 3293 |
-
"epoch": 0.98,
|
| 3294 |
-
"learning_rate": 3.6328292679828624e-08,
|
| 3295 |
-
"loss": 0.0059,
|
| 3296 |
-
"r_loss": 0.004687966778874397,
|
| 3297 |
-
"s_loss": 0.0007698750705458224,
|
| 3298 |
-
"step": 33000
|
| 3299 |
-
},
|
| 3300 |
-
{
|
| 3301 |
-
"epoch": 0.98,
|
| 3302 |
-
"eval_loss": 0.006356844212859869,
|
| 3303 |
-
"eval_r_loss": 0.005574433133006096,
|
| 3304 |
-
"eval_runtime": 8.7324,
|
| 3305 |
-
"eval_s_loss": 0.0007824110798537731,
|
| 3306 |
-
"eval_samples_per_second": 76.955,
|
| 3307 |
-
"eval_steps_per_second": 76.955,
|
| 3308 |
-
"step": 33000
|
| 3309 |
-
},
|
| 3310 |
-
{
|
| 3311 |
-
"epoch": 0.99,
|
| 3312 |
-
"learning_rate": 2.37243650380925e-08,
|
| 3313 |
-
"loss": 0.0058,
|
| 3314 |
-
"r_loss": 0.0053665488958358765,
|
| 3315 |
-
"s_loss": 0.0008012793259695172,
|
| 3316 |
-
"step": 33100
|
| 3317 |
-
},
|
| 3318 |
-
{
|
| 3319 |
-
"epoch": 0.99,
|
| 3320 |
-
"learning_rate": 1.3793942524174541e-08,
|
| 3321 |
-
"loss": 0.0059,
|
| 3322 |
-
"r_loss": 0.004292100202292204,
|
| 3323 |
-
"s_loss": 0.0007570346933789551,
|
| 3324 |
-
"step": 33200
|
| 3325 |
-
},
|
| 3326 |
-
{
|
| 3327 |
-
"epoch": 0.99,
|
| 3328 |
-
"learning_rate": 6.538797877583269e-09,
|
| 3329 |
-
"loss": 0.0059,
|
| 3330 |
-
"r_loss": 0.005391569808125496,
|
| 3331 |
-
"s_loss": 0.0008714336436241865,
|
| 3332 |
-
"step": 33300
|
| 3333 |
-
},
|
| 3334 |
-
{
|
| 3335 |
-
"epoch": 1.0,
|
| 3336 |
-
"learning_rate": 1.9602262578682917e-09,
|
| 3337 |
-
"loss": 0.0059,
|
| 3338 |
-
"r_loss": 0.004852307494729757,
|
| 3339 |
-
"s_loss": 0.0007645284058526158,
|
| 3340 |
-
"step": 33400
|
| 3341 |
-
},
|
| 3342 |
-
{
|
| 3343 |
-
"epoch": 1.0,
|
| 3344 |
-
"learning_rate": 5.90450134152487e-11,
|
| 3345 |
-
"loss": 0.0059,
|
| 3346 |
-
"r_loss": 0.0045965323224663734,
|
| 3347 |
-
"s_loss": 0.0008654047851450741,
|
| 3348 |
-
"step": 33500
|
| 3349 |
-
},
|
| 3350 |
-
{
|
| 3351 |
-
"epoch": 1.0,
|
| 3352 |
-
"eval_loss": 0.006353782489895821,
|
| 3353 |
-
"eval_r_loss": 0.0055714258924126625,
|
| 3354 |
-
"eval_runtime": 9.2592,
|
| 3355 |
-
"eval_s_loss": 0.0007823564810678363,
|
| 3356 |
-
"eval_samples_per_second": 72.576,
|
| 3357 |
-
"eval_steps_per_second": 72.576,
|
| 3358 |
-
"step": 33500
|
| 3359 |
-
}
|
| 3360 |
-
],
|
| 3361 |
-
"max_steps": 33513,
|
| 3362 |
-
"num_train_epochs": 9223372036854775807,
|
| 3363 |
-
"total_flos": 2.119174456955371e+16,
|
| 3364 |
-
"trial_name": null,
|
| 3365 |
-
"trial_params": null
|
| 3366 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
last-checkpoint/training_args.bin
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:308f6c0d3916cfaa1801ef8138ac88dd659462c3200361fad5cfa39ff8dd4faf
|
| 3 |
-
size 3899
|
|
|
|
|
|
|
|
|
|
|
|
last-checkpoint/vocab.json
DELETED
|
The diff for this file is too large to render.
See raw diff
|
|
|
model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 264627500
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:34e13ead3f3ca792bed6d415823f1aaf158e7fa1020ff23991dc7bd3acab5151
|
| 3 |
size 264627500
|