Commit
·
558097a
1
Parent(s):
8220ddf
Upload 6 files
Browse files- config.json +147 -0
- preprocessor_config.json +9 -0
- rng_state.pth +3 -0
- scheduler.pt +3 -0
- trainer_state.json +1939 -0
- training_args.bin +3 -0
config.json
ADDED
|
@@ -0,0 +1,147 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "facebook/wav2vec2-base",
|
| 3 |
+
"activation_dropout": 0.0,
|
| 4 |
+
"adapter_kernel_size": 3,
|
| 5 |
+
"adapter_stride": 2,
|
| 6 |
+
"add_adapter": false,
|
| 7 |
+
"apply_spec_augment": true,
|
| 8 |
+
"architectures": [
|
| 9 |
+
"Wav2Vec2ForSequenceClassification"
|
| 10 |
+
],
|
| 11 |
+
"attention_dropout": 0.1,
|
| 12 |
+
"bos_token_id": 1,
|
| 13 |
+
"classifier_proj_size": 256,
|
| 14 |
+
"codevector_dim": 256,
|
| 15 |
+
"contrastive_logits_temperature": 0.1,
|
| 16 |
+
"conv_bias": false,
|
| 17 |
+
"conv_dim": [
|
| 18 |
+
512,
|
| 19 |
+
512,
|
| 20 |
+
512,
|
| 21 |
+
512,
|
| 22 |
+
512,
|
| 23 |
+
512,
|
| 24 |
+
512
|
| 25 |
+
],
|
| 26 |
+
"conv_kernel": [
|
| 27 |
+
10,
|
| 28 |
+
3,
|
| 29 |
+
3,
|
| 30 |
+
3,
|
| 31 |
+
3,
|
| 32 |
+
2,
|
| 33 |
+
2
|
| 34 |
+
],
|
| 35 |
+
"conv_stride": [
|
| 36 |
+
5,
|
| 37 |
+
2,
|
| 38 |
+
2,
|
| 39 |
+
2,
|
| 40 |
+
2,
|
| 41 |
+
2,
|
| 42 |
+
2
|
| 43 |
+
],
|
| 44 |
+
"ctc_loss_reduction": "sum",
|
| 45 |
+
"ctc_zero_infinity": false,
|
| 46 |
+
"diversity_loss_weight": 0.1,
|
| 47 |
+
"do_stable_layer_norm": false,
|
| 48 |
+
"eos_token_id": 2,
|
| 49 |
+
"feat_extract_activation": "gelu",
|
| 50 |
+
"feat_extract_norm": "group",
|
| 51 |
+
"feat_proj_dropout": 0.1,
|
| 52 |
+
"feat_quantizer_dropout": 0.0,
|
| 53 |
+
"final_dropout": 0.0,
|
| 54 |
+
"freeze_feat_extract_train": true,
|
| 55 |
+
"hidden_act": "gelu",
|
| 56 |
+
"hidden_dropout": 0.1,
|
| 57 |
+
"hidden_size": 768,
|
| 58 |
+
"id2label": {
|
| 59 |
+
"0": "L",
|
| 60 |
+
"1": "A",
|
| 61 |
+
"10": "M",
|
| 62 |
+
"11": "B",
|
| 63 |
+
"12": "I",
|
| 64 |
+
"2": "K",
|
| 65 |
+
"3": "D",
|
| 66 |
+
"4": "C",
|
| 67 |
+
"5": "H",
|
| 68 |
+
"6": "F",
|
| 69 |
+
"7": "E",
|
| 70 |
+
"8": "J",
|
| 71 |
+
"9": "G"
|
| 72 |
+
},
|
| 73 |
+
"initializer_range": 0.02,
|
| 74 |
+
"intermediate_size": 3072,
|
| 75 |
+
"label2id": {
|
| 76 |
+
"A": "1",
|
| 77 |
+
"B": "11",
|
| 78 |
+
"C": "4",
|
| 79 |
+
"D": "3",
|
| 80 |
+
"E": "7",
|
| 81 |
+
"F": "6",
|
| 82 |
+
"G": "9",
|
| 83 |
+
"H": "5",
|
| 84 |
+
"I": "12",
|
| 85 |
+
"J": "8",
|
| 86 |
+
"K": "2",
|
| 87 |
+
"L": "0",
|
| 88 |
+
"M": "10"
|
| 89 |
+
},
|
| 90 |
+
"layer_norm_eps": 1e-05,
|
| 91 |
+
"layerdrop": 0.0,
|
| 92 |
+
"mask_channel_length": 10,
|
| 93 |
+
"mask_channel_min_space": 1,
|
| 94 |
+
"mask_channel_other": 0.0,
|
| 95 |
+
"mask_channel_prob": 0.0,
|
| 96 |
+
"mask_channel_selection": "static",
|
| 97 |
+
"mask_feature_length": 10,
|
| 98 |
+
"mask_feature_min_masks": 0,
|
| 99 |
+
"mask_feature_prob": 0.0,
|
| 100 |
+
"mask_time_length": 10,
|
| 101 |
+
"mask_time_min_masks": 2,
|
| 102 |
+
"mask_time_min_space": 1,
|
| 103 |
+
"mask_time_other": 0.0,
|
| 104 |
+
"mask_time_prob": 0.05,
|
| 105 |
+
"mask_time_selection": "static",
|
| 106 |
+
"model_type": "wav2vec2",
|
| 107 |
+
"no_mask_channel_overlap": false,
|
| 108 |
+
"no_mask_time_overlap": false,
|
| 109 |
+
"num_adapter_layers": 3,
|
| 110 |
+
"num_attention_heads": 12,
|
| 111 |
+
"num_codevector_groups": 2,
|
| 112 |
+
"num_codevectors_per_group": 320,
|
| 113 |
+
"num_conv_pos_embedding_groups": 16,
|
| 114 |
+
"num_conv_pos_embeddings": 128,
|
| 115 |
+
"num_feat_extract_layers": 7,
|
| 116 |
+
"num_hidden_layers": 12,
|
| 117 |
+
"num_negatives": 100,
|
| 118 |
+
"output_hidden_size": 768,
|
| 119 |
+
"pad_token_id": 0,
|
| 120 |
+
"proj_codevector_dim": 256,
|
| 121 |
+
"tdnn_dilation": [
|
| 122 |
+
1,
|
| 123 |
+
2,
|
| 124 |
+
3,
|
| 125 |
+
1,
|
| 126 |
+
1
|
| 127 |
+
],
|
| 128 |
+
"tdnn_dim": [
|
| 129 |
+
512,
|
| 130 |
+
512,
|
| 131 |
+
512,
|
| 132 |
+
512,
|
| 133 |
+
1500
|
| 134 |
+
],
|
| 135 |
+
"tdnn_kernel": [
|
| 136 |
+
5,
|
| 137 |
+
3,
|
| 138 |
+
3,
|
| 139 |
+
1,
|
| 140 |
+
1
|
| 141 |
+
],
|
| 142 |
+
"torch_dtype": "float32",
|
| 143 |
+
"transformers_version": "4.28.1",
|
| 144 |
+
"use_weighted_layer_sum": false,
|
| 145 |
+
"vocab_size": 32,
|
| 146 |
+
"xvector_output_dim": 512
|
| 147 |
+
}
|
preprocessor_config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"do_normalize": true,
|
| 3 |
+
"feature_extractor_type": "Wav2Vec2FeatureExtractor",
|
| 4 |
+
"feature_size": 1,
|
| 5 |
+
"padding_side": "right",
|
| 6 |
+
"padding_value": 0.0,
|
| 7 |
+
"return_attention_mask": false,
|
| 8 |
+
"sampling_rate": 16000
|
| 9 |
+
}
|
rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0e57c638473963d7fc984357e01d81b3dc673f1b8ec64920cc283e979adf5ecc
|
| 3 |
+
size 14639
|
scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:36380f4fc084506b2062f11a6401c45dd68d3829a12f744d1a4fcda6f70767d0
|
| 3 |
+
size 627
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1939 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.45121951219512196,
|
| 3 |
+
"best_model_checkpoint": "wav2vec2-base-berkeley-emotions-decay-f1/checkpoint-2958",
|
| 4 |
+
"epoch": 17.0,
|
| 5 |
+
"global_step": 2958,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.06,
|
| 12 |
+
"learning_rate": 1.7241379310344828e-07,
|
| 13 |
+
"loss": 2.569,
|
| 14 |
+
"step": 10
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.11,
|
| 18 |
+
"learning_rate": 3.4482758620689656e-07,
|
| 19 |
+
"loss": 2.567,
|
| 20 |
+
"step": 20
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.17,
|
| 24 |
+
"learning_rate": 5.172413793103448e-07,
|
| 25 |
+
"loss": 2.568,
|
| 26 |
+
"step": 30
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.23,
|
| 30 |
+
"learning_rate": 6.896551724137931e-07,
|
| 31 |
+
"loss": 2.5676,
|
| 32 |
+
"step": 40
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.29,
|
| 36 |
+
"learning_rate": 8.620689655172414e-07,
|
| 37 |
+
"loss": 2.562,
|
| 38 |
+
"step": 50
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.34,
|
| 42 |
+
"learning_rate": 1.0344827586206896e-06,
|
| 43 |
+
"loss": 2.564,
|
| 44 |
+
"step": 60
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.4,
|
| 48 |
+
"learning_rate": 1.206896551724138e-06,
|
| 49 |
+
"loss": 2.5549,
|
| 50 |
+
"step": 70
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.46,
|
| 54 |
+
"learning_rate": 1.3793103448275862e-06,
|
| 55 |
+
"loss": 2.5454,
|
| 56 |
+
"step": 80
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.52,
|
| 60 |
+
"learning_rate": 1.5517241379310346e-06,
|
| 61 |
+
"loss": 2.5459,
|
| 62 |
+
"step": 90
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.57,
|
| 66 |
+
"learning_rate": 1.7241379310344827e-06,
|
| 67 |
+
"loss": 2.5321,
|
| 68 |
+
"step": 100
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.63,
|
| 72 |
+
"learning_rate": 1.8965517241379312e-06,
|
| 73 |
+
"loss": 2.51,
|
| 74 |
+
"step": 110
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.69,
|
| 78 |
+
"learning_rate": 2.068965517241379e-06,
|
| 79 |
+
"loss": 2.5365,
|
| 80 |
+
"step": 120
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.75,
|
| 84 |
+
"learning_rate": 2.241379310344828e-06,
|
| 85 |
+
"loss": 2.5137,
|
| 86 |
+
"step": 130
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.8,
|
| 90 |
+
"learning_rate": 2.413793103448276e-06,
|
| 91 |
+
"loss": 2.4893,
|
| 92 |
+
"step": 140
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 0.86,
|
| 96 |
+
"learning_rate": 2.5862068965517246e-06,
|
| 97 |
+
"loss": 2.5016,
|
| 98 |
+
"step": 150
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.92,
|
| 102 |
+
"learning_rate": 2.7586206896551725e-06,
|
| 103 |
+
"loss": 2.4782,
|
| 104 |
+
"step": 160
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 0.98,
|
| 108 |
+
"learning_rate": 2.931034482758621e-06,
|
| 109 |
+
"loss": 2.4916,
|
| 110 |
+
"step": 170
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 1.0,
|
| 114 |
+
"eval_f1": 0.21138211382113825,
|
| 115 |
+
"eval_loss": 2.459113836288452,
|
| 116 |
+
"eval_runtime": 12.4194,
|
| 117 |
+
"eval_samples_per_second": 19.808,
|
| 118 |
+
"eval_steps_per_second": 9.904,
|
| 119 |
+
"step": 174
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 1.03,
|
| 123 |
+
"learning_rate": 3.103448275862069e-06,
|
| 124 |
+
"loss": 2.4868,
|
| 125 |
+
"step": 180
|
| 126 |
+
},
|
| 127 |
+
{
|
| 128 |
+
"epoch": 1.09,
|
| 129 |
+
"learning_rate": 3.2758620689655175e-06,
|
| 130 |
+
"loss": 2.3999,
|
| 131 |
+
"step": 190
|
| 132 |
+
},
|
| 133 |
+
{
|
| 134 |
+
"epoch": 1.15,
|
| 135 |
+
"learning_rate": 3.4482758620689654e-06,
|
| 136 |
+
"loss": 2.4493,
|
| 137 |
+
"step": 200
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 1.21,
|
| 141 |
+
"learning_rate": 3.620689655172414e-06,
|
| 142 |
+
"loss": 2.3774,
|
| 143 |
+
"step": 210
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 1.26,
|
| 147 |
+
"learning_rate": 3.7931034482758625e-06,
|
| 148 |
+
"loss": 2.3841,
|
| 149 |
+
"step": 220
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 1.32,
|
| 153 |
+
"learning_rate": 3.96551724137931e-06,
|
| 154 |
+
"loss": 2.3626,
|
| 155 |
+
"step": 230
|
| 156 |
+
},
|
| 157 |
+
{
|
| 158 |
+
"epoch": 1.38,
|
| 159 |
+
"learning_rate": 4.137931034482758e-06,
|
| 160 |
+
"loss": 2.3698,
|
| 161 |
+
"step": 240
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"epoch": 1.44,
|
| 165 |
+
"learning_rate": 4.310344827586207e-06,
|
| 166 |
+
"loss": 2.3962,
|
| 167 |
+
"step": 250
|
| 168 |
+
},
|
| 169 |
+
{
|
| 170 |
+
"epoch": 1.49,
|
| 171 |
+
"learning_rate": 4.482758620689656e-06,
|
| 172 |
+
"loss": 2.3698,
|
| 173 |
+
"step": 260
|
| 174 |
+
},
|
| 175 |
+
{
|
| 176 |
+
"epoch": 1.55,
|
| 177 |
+
"learning_rate": 4.655172413793104e-06,
|
| 178 |
+
"loss": 2.3315,
|
| 179 |
+
"step": 270
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 1.61,
|
| 183 |
+
"learning_rate": 4.827586206896552e-06,
|
| 184 |
+
"loss": 2.3425,
|
| 185 |
+
"step": 280
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 1.67,
|
| 189 |
+
"learning_rate": 4.9999999999999996e-06,
|
| 190 |
+
"loss": 2.4017,
|
| 191 |
+
"step": 290
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 1.72,
|
| 195 |
+
"learning_rate": 5.172413793103449e-06,
|
| 196 |
+
"loss": 2.2892,
|
| 197 |
+
"step": 300
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 1.78,
|
| 201 |
+
"learning_rate": 5.344827586206897e-06,
|
| 202 |
+
"loss": 2.3775,
|
| 203 |
+
"step": 310
|
| 204 |
+
},
|
| 205 |
+
{
|
| 206 |
+
"epoch": 1.84,
|
| 207 |
+
"learning_rate": 5.517241379310345e-06,
|
| 208 |
+
"loss": 2.2589,
|
| 209 |
+
"step": 320
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 1.9,
|
| 213 |
+
"learning_rate": 5.689655172413793e-06,
|
| 214 |
+
"loss": 2.2562,
|
| 215 |
+
"step": 330
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"epoch": 1.95,
|
| 219 |
+
"learning_rate": 5.862068965517242e-06,
|
| 220 |
+
"loss": 2.2591,
|
| 221 |
+
"step": 340
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 2.0,
|
| 225 |
+
"eval_f1": 0.3048780487804878,
|
| 226 |
+
"eval_loss": 2.270204782485962,
|
| 227 |
+
"eval_runtime": 11.4533,
|
| 228 |
+
"eval_samples_per_second": 21.479,
|
| 229 |
+
"eval_steps_per_second": 10.739,
|
| 230 |
+
"step": 348
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 2.01,
|
| 234 |
+
"learning_rate": 6.0344827586206896e-06,
|
| 235 |
+
"loss": 2.3039,
|
| 236 |
+
"step": 350
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 2.07,
|
| 240 |
+
"learning_rate": 6.206896551724138e-06,
|
| 241 |
+
"loss": 2.1748,
|
| 242 |
+
"step": 360
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 2.13,
|
| 246 |
+
"learning_rate": 6.379310344827586e-06,
|
| 247 |
+
"loss": 2.2222,
|
| 248 |
+
"step": 370
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 2.18,
|
| 252 |
+
"learning_rate": 6.551724137931035e-06,
|
| 253 |
+
"loss": 2.2989,
|
| 254 |
+
"step": 380
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 2.24,
|
| 258 |
+
"learning_rate": 6.724137931034483e-06,
|
| 259 |
+
"loss": 2.1511,
|
| 260 |
+
"step": 390
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 2.3,
|
| 264 |
+
"learning_rate": 6.896551724137931e-06,
|
| 265 |
+
"loss": 2.2883,
|
| 266 |
+
"step": 400
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 2.36,
|
| 270 |
+
"learning_rate": 7.0689655172413796e-06,
|
| 271 |
+
"loss": 2.163,
|
| 272 |
+
"step": 410
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 2.41,
|
| 276 |
+
"learning_rate": 7.241379310344828e-06,
|
| 277 |
+
"loss": 2.1922,
|
| 278 |
+
"step": 420
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 2.47,
|
| 282 |
+
"learning_rate": 7.413793103448276e-06,
|
| 283 |
+
"loss": 2.216,
|
| 284 |
+
"step": 430
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 2.53,
|
| 288 |
+
"learning_rate": 7.586206896551725e-06,
|
| 289 |
+
"loss": 2.2721,
|
| 290 |
+
"step": 440
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 2.59,
|
| 294 |
+
"learning_rate": 7.758620689655173e-06,
|
| 295 |
+
"loss": 2.2441,
|
| 296 |
+
"step": 450
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 2.64,
|
| 300 |
+
"learning_rate": 7.93103448275862e-06,
|
| 301 |
+
"loss": 2.2579,
|
| 302 |
+
"step": 460
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 2.7,
|
| 306 |
+
"learning_rate": 8.103448275862069e-06,
|
| 307 |
+
"loss": 2.1499,
|
| 308 |
+
"step": 470
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 2.76,
|
| 312 |
+
"learning_rate": 8.275862068965517e-06,
|
| 313 |
+
"loss": 2.3075,
|
| 314 |
+
"step": 480
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 2.82,
|
| 318 |
+
"learning_rate": 8.448275862068966e-06,
|
| 319 |
+
"loss": 2.1775,
|
| 320 |
+
"step": 490
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 2.87,
|
| 324 |
+
"learning_rate": 8.620689655172414e-06,
|
| 325 |
+
"loss": 2.2148,
|
| 326 |
+
"step": 500
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 2.93,
|
| 330 |
+
"learning_rate": 8.793103448275862e-06,
|
| 331 |
+
"loss": 2.1458,
|
| 332 |
+
"step": 510
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 2.99,
|
| 336 |
+
"learning_rate": 8.965517241379312e-06,
|
| 337 |
+
"loss": 2.1159,
|
| 338 |
+
"step": 520
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 3.0,
|
| 342 |
+
"eval_f1": 0.2845528455284553,
|
| 343 |
+
"eval_loss": 2.224947929382324,
|
| 344 |
+
"eval_runtime": 11.2398,
|
| 345 |
+
"eval_samples_per_second": 21.887,
|
| 346 |
+
"eval_steps_per_second": 10.943,
|
| 347 |
+
"step": 522
|
| 348 |
+
},
|
| 349 |
+
{
|
| 350 |
+
"epoch": 3.05,
|
| 351 |
+
"learning_rate": 9.13793103448276e-06,
|
| 352 |
+
"loss": 2.0612,
|
| 353 |
+
"step": 530
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 3.1,
|
| 357 |
+
"learning_rate": 9.310344827586207e-06,
|
| 358 |
+
"loss": 2.1366,
|
| 359 |
+
"step": 540
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 3.16,
|
| 363 |
+
"learning_rate": 9.482758620689655e-06,
|
| 364 |
+
"loss": 2.0171,
|
| 365 |
+
"step": 550
|
| 366 |
+
},
|
| 367 |
+
{
|
| 368 |
+
"epoch": 3.22,
|
| 369 |
+
"learning_rate": 9.655172413793103e-06,
|
| 370 |
+
"loss": 2.1402,
|
| 371 |
+
"step": 560
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 3.28,
|
| 375 |
+
"learning_rate": 9.827586206896551e-06,
|
| 376 |
+
"loss": 2.1011,
|
| 377 |
+
"step": 570
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 3.33,
|
| 381 |
+
"learning_rate": 9.999999999999999e-06,
|
| 382 |
+
"loss": 2.1832,
|
| 383 |
+
"step": 580
|
| 384 |
+
},
|
| 385 |
+
{
|
| 386 |
+
"epoch": 3.39,
|
| 387 |
+
"learning_rate": 1.0172413793103447e-05,
|
| 388 |
+
"loss": 2.1226,
|
| 389 |
+
"step": 590
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 3.45,
|
| 393 |
+
"learning_rate": 1.0344827586206898e-05,
|
| 394 |
+
"loss": 2.0552,
|
| 395 |
+
"step": 600
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 3.51,
|
| 399 |
+
"learning_rate": 1.0517241379310346e-05,
|
| 400 |
+
"loss": 2.0428,
|
| 401 |
+
"step": 610
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 3.56,
|
| 405 |
+
"learning_rate": 1.0689655172413794e-05,
|
| 406 |
+
"loss": 2.119,
|
| 407 |
+
"step": 620
|
| 408 |
+
},
|
| 409 |
+
{
|
| 410 |
+
"epoch": 3.62,
|
| 411 |
+
"learning_rate": 1.0862068965517242e-05,
|
| 412 |
+
"loss": 2.1665,
|
| 413 |
+
"step": 630
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 3.68,
|
| 417 |
+
"learning_rate": 1.103448275862069e-05,
|
| 418 |
+
"loss": 2.0697,
|
| 419 |
+
"step": 640
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 3.74,
|
| 423 |
+
"learning_rate": 1.1206896551724138e-05,
|
| 424 |
+
"loss": 2.0581,
|
| 425 |
+
"step": 650
|
| 426 |
+
},
|
| 427 |
+
{
|
| 428 |
+
"epoch": 3.79,
|
| 429 |
+
"learning_rate": 1.1379310344827586e-05,
|
| 430 |
+
"loss": 2.2953,
|
| 431 |
+
"step": 660
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 3.85,
|
| 435 |
+
"learning_rate": 1.1551724137931034e-05,
|
| 436 |
+
"loss": 2.09,
|
| 437 |
+
"step": 670
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 3.91,
|
| 441 |
+
"learning_rate": 1.1724137931034483e-05,
|
| 442 |
+
"loss": 1.9762,
|
| 443 |
+
"step": 680
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 3.97,
|
| 447 |
+
"learning_rate": 1.1896551724137931e-05,
|
| 448 |
+
"loss": 2.1636,
|
| 449 |
+
"step": 690
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 4.0,
|
| 453 |
+
"eval_f1": 0.3170731707317073,
|
| 454 |
+
"eval_loss": 2.092029571533203,
|
| 455 |
+
"eval_runtime": 11.84,
|
| 456 |
+
"eval_samples_per_second": 20.777,
|
| 457 |
+
"eval_steps_per_second": 10.388,
|
| 458 |
+
"step": 696
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 4.02,
|
| 462 |
+
"learning_rate": 1.2068965517241379e-05,
|
| 463 |
+
"loss": 2.0153,
|
| 464 |
+
"step": 700
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 4.08,
|
| 468 |
+
"learning_rate": 1.2241379310344829e-05,
|
| 469 |
+
"loss": 2.0732,
|
| 470 |
+
"step": 710
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 4.14,
|
| 474 |
+
"learning_rate": 1.2413793103448277e-05,
|
| 475 |
+
"loss": 1.9913,
|
| 476 |
+
"step": 720
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 4.2,
|
| 480 |
+
"learning_rate": 1.2586206896551725e-05,
|
| 481 |
+
"loss": 2.0738,
|
| 482 |
+
"step": 730
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 4.25,
|
| 486 |
+
"learning_rate": 1.2758620689655172e-05,
|
| 487 |
+
"loss": 1.9731,
|
| 488 |
+
"step": 740
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 4.31,
|
| 492 |
+
"learning_rate": 1.293103448275862e-05,
|
| 493 |
+
"loss": 2.1501,
|
| 494 |
+
"step": 750
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 4.37,
|
| 498 |
+
"learning_rate": 1.310344827586207e-05,
|
| 499 |
+
"loss": 2.0999,
|
| 500 |
+
"step": 760
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 4.43,
|
| 504 |
+
"learning_rate": 1.3275862068965518e-05,
|
| 505 |
+
"loss": 2.0082,
|
| 506 |
+
"step": 770
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 4.48,
|
| 510 |
+
"learning_rate": 1.3448275862068966e-05,
|
| 511 |
+
"loss": 2.0015,
|
| 512 |
+
"step": 780
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 4.54,
|
| 516 |
+
"learning_rate": 1.3620689655172414e-05,
|
| 517 |
+
"loss": 2.0501,
|
| 518 |
+
"step": 790
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 4.6,
|
| 522 |
+
"learning_rate": 1.3793103448275862e-05,
|
| 523 |
+
"loss": 1.8933,
|
| 524 |
+
"step": 800
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 4.66,
|
| 528 |
+
"learning_rate": 1.396551724137931e-05,
|
| 529 |
+
"loss": 2.0852,
|
| 530 |
+
"step": 810
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 4.71,
|
| 534 |
+
"learning_rate": 1.4137931034482759e-05,
|
| 535 |
+
"loss": 2.1491,
|
| 536 |
+
"step": 820
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 4.77,
|
| 540 |
+
"learning_rate": 1.4310344827586207e-05,
|
| 541 |
+
"loss": 2.0248,
|
| 542 |
+
"step": 830
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 4.83,
|
| 546 |
+
"learning_rate": 1.4482758620689657e-05,
|
| 547 |
+
"loss": 2.0256,
|
| 548 |
+
"step": 840
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 4.89,
|
| 552 |
+
"learning_rate": 1.4655172413793105e-05,
|
| 553 |
+
"loss": 2.1742,
|
| 554 |
+
"step": 850
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 4.94,
|
| 558 |
+
"learning_rate": 1.4827586206896552e-05,
|
| 559 |
+
"loss": 1.9253,
|
| 560 |
+
"step": 860
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 5.0,
|
| 564 |
+
"learning_rate": 1.5e-05,
|
| 565 |
+
"loss": 2.0857,
|
| 566 |
+
"step": 870
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 5.0,
|
| 570 |
+
"eval_f1": 0.3089430894308943,
|
| 571 |
+
"eval_loss": 2.0949580669403076,
|
| 572 |
+
"eval_runtime": 12.0187,
|
| 573 |
+
"eval_samples_per_second": 20.468,
|
| 574 |
+
"eval_steps_per_second": 10.234,
|
| 575 |
+
"step": 870
|
| 576 |
+
},
|
| 577 |
+
{
|
| 578 |
+
"epoch": 5.06,
|
| 579 |
+
"learning_rate": 1.517241379310345e-05,
|
| 580 |
+
"loss": 1.9912,
|
| 581 |
+
"step": 880
|
| 582 |
+
},
|
| 583 |
+
{
|
| 584 |
+
"epoch": 5.11,
|
| 585 |
+
"learning_rate": 1.5344827586206898e-05,
|
| 586 |
+
"loss": 1.8998,
|
| 587 |
+
"step": 890
|
| 588 |
+
},
|
| 589 |
+
{
|
| 590 |
+
"epoch": 5.17,
|
| 591 |
+
"learning_rate": 1.5517241379310346e-05,
|
| 592 |
+
"loss": 1.8834,
|
| 593 |
+
"step": 900
|
| 594 |
+
},
|
| 595 |
+
{
|
| 596 |
+
"epoch": 5.23,
|
| 597 |
+
"learning_rate": 1.5689655172413794e-05,
|
| 598 |
+
"loss": 1.8255,
|
| 599 |
+
"step": 910
|
| 600 |
+
},
|
| 601 |
+
{
|
| 602 |
+
"epoch": 5.29,
|
| 603 |
+
"learning_rate": 1.586206896551724e-05,
|
| 604 |
+
"loss": 2.0895,
|
| 605 |
+
"step": 920
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 5.34,
|
| 609 |
+
"learning_rate": 1.603448275862069e-05,
|
| 610 |
+
"loss": 1.9981,
|
| 611 |
+
"step": 930
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 5.4,
|
| 615 |
+
"learning_rate": 1.6206896551724137e-05,
|
| 616 |
+
"loss": 1.9582,
|
| 617 |
+
"step": 940
|
| 618 |
+
},
|
| 619 |
+
{
|
| 620 |
+
"epoch": 5.46,
|
| 621 |
+
"learning_rate": 1.637931034482759e-05,
|
| 622 |
+
"loss": 2.05,
|
| 623 |
+
"step": 950
|
| 624 |
+
},
|
| 625 |
+
{
|
| 626 |
+
"epoch": 5.52,
|
| 627 |
+
"learning_rate": 1.6551724137931033e-05,
|
| 628 |
+
"loss": 1.9791,
|
| 629 |
+
"step": 960
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 5.57,
|
| 633 |
+
"learning_rate": 1.6724137931034485e-05,
|
| 634 |
+
"loss": 2.0748,
|
| 635 |
+
"step": 970
|
| 636 |
+
},
|
| 637 |
+
{
|
| 638 |
+
"epoch": 5.63,
|
| 639 |
+
"learning_rate": 1.6896551724137932e-05,
|
| 640 |
+
"loss": 1.9022,
|
| 641 |
+
"step": 980
|
| 642 |
+
},
|
| 643 |
+
{
|
| 644 |
+
"epoch": 5.69,
|
| 645 |
+
"learning_rate": 1.706896551724138e-05,
|
| 646 |
+
"loss": 1.9849,
|
| 647 |
+
"step": 990
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 5.75,
|
| 651 |
+
"learning_rate": 1.7241379310344828e-05,
|
| 652 |
+
"loss": 1.9502,
|
| 653 |
+
"step": 1000
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 5.8,
|
| 657 |
+
"learning_rate": 1.7413793103448276e-05,
|
| 658 |
+
"loss": 1.8595,
|
| 659 |
+
"step": 1010
|
| 660 |
+
},
|
| 661 |
+
{
|
| 662 |
+
"epoch": 5.86,
|
| 663 |
+
"learning_rate": 1.7586206896551724e-05,
|
| 664 |
+
"loss": 1.8499,
|
| 665 |
+
"step": 1020
|
| 666 |
+
},
|
| 667 |
+
{
|
| 668 |
+
"epoch": 5.92,
|
| 669 |
+
"learning_rate": 1.7758620689655172e-05,
|
| 670 |
+
"loss": 1.8505,
|
| 671 |
+
"step": 1030
|
| 672 |
+
},
|
| 673 |
+
{
|
| 674 |
+
"epoch": 5.98,
|
| 675 |
+
"learning_rate": 1.7931034482758623e-05,
|
| 676 |
+
"loss": 2.0275,
|
| 677 |
+
"step": 1040
|
| 678 |
+
},
|
| 679 |
+
{
|
| 680 |
+
"epoch": 6.0,
|
| 681 |
+
"eval_f1": 0.3252032520325203,
|
| 682 |
+
"eval_loss": 2.106571912765503,
|
| 683 |
+
"eval_runtime": 11.9924,
|
| 684 |
+
"eval_samples_per_second": 20.513,
|
| 685 |
+
"eval_steps_per_second": 10.257,
|
| 686 |
+
"step": 1044
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 6.03,
|
| 690 |
+
"learning_rate": 1.8103448275862068e-05,
|
| 691 |
+
"loss": 2.0014,
|
| 692 |
+
"step": 1050
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 6.09,
|
| 696 |
+
"learning_rate": 1.827586206896552e-05,
|
| 697 |
+
"loss": 2.2037,
|
| 698 |
+
"step": 1060
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 6.15,
|
| 702 |
+
"learning_rate": 1.8448275862068964e-05,
|
| 703 |
+
"loss": 1.9573,
|
| 704 |
+
"step": 1070
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 6.21,
|
| 708 |
+
"learning_rate": 1.8620689655172415e-05,
|
| 709 |
+
"loss": 1.8635,
|
| 710 |
+
"step": 1080
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 6.26,
|
| 714 |
+
"learning_rate": 1.8793103448275863e-05,
|
| 715 |
+
"loss": 1.9216,
|
| 716 |
+
"step": 1090
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 6.32,
|
| 720 |
+
"learning_rate": 1.896551724137931e-05,
|
| 721 |
+
"loss": 1.7776,
|
| 722 |
+
"step": 1100
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 6.38,
|
| 726 |
+
"learning_rate": 1.9137931034482762e-05,
|
| 727 |
+
"loss": 2.0595,
|
| 728 |
+
"step": 1110
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 6.44,
|
| 732 |
+
"learning_rate": 1.9310344827586207e-05,
|
| 733 |
+
"loss": 1.8406,
|
| 734 |
+
"step": 1120
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 6.49,
|
| 738 |
+
"learning_rate": 1.9482758620689658e-05,
|
| 739 |
+
"loss": 1.8169,
|
| 740 |
+
"step": 1130
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 6.55,
|
| 744 |
+
"learning_rate": 1.9655172413793102e-05,
|
| 745 |
+
"loss": 1.8881,
|
| 746 |
+
"step": 1140
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 6.61,
|
| 750 |
+
"learning_rate": 1.9827586206896554e-05,
|
| 751 |
+
"loss": 1.64,
|
| 752 |
+
"step": 1150
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 6.67,
|
| 756 |
+
"learning_rate": 1.9999999999999998e-05,
|
| 757 |
+
"loss": 1.9238,
|
| 758 |
+
"step": 1160
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 6.72,
|
| 762 |
+
"learning_rate": 2.017241379310345e-05,
|
| 763 |
+
"loss": 2.0383,
|
| 764 |
+
"step": 1170
|
| 765 |
+
},
|
| 766 |
+
{
|
| 767 |
+
"epoch": 6.78,
|
| 768 |
+
"learning_rate": 2.0344827586206894e-05,
|
| 769 |
+
"loss": 1.9494,
|
| 770 |
+
"step": 1180
|
| 771 |
+
},
|
| 772 |
+
{
|
| 773 |
+
"epoch": 6.84,
|
| 774 |
+
"learning_rate": 2.0517241379310345e-05,
|
| 775 |
+
"loss": 1.9281,
|
| 776 |
+
"step": 1190
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 6.9,
|
| 780 |
+
"learning_rate": 2.0689655172413797e-05,
|
| 781 |
+
"loss": 1.9734,
|
| 782 |
+
"step": 1200
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 6.95,
|
| 786 |
+
"learning_rate": 2.086206896551724e-05,
|
| 787 |
+
"loss": 1.8332,
|
| 788 |
+
"step": 1210
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 7.0,
|
| 792 |
+
"eval_f1": 0.3821138211382114,
|
| 793 |
+
"eval_loss": 1.9587126970291138,
|
| 794 |
+
"eval_runtime": 11.5,
|
| 795 |
+
"eval_samples_per_second": 21.391,
|
| 796 |
+
"eval_steps_per_second": 10.696,
|
| 797 |
+
"step": 1218
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 7.01,
|
| 801 |
+
"learning_rate": 2.1034482758620692e-05,
|
| 802 |
+
"loss": 1.9479,
|
| 803 |
+
"step": 1220
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 7.07,
|
| 807 |
+
"learning_rate": 2.1206896551724137e-05,
|
| 808 |
+
"loss": 1.8237,
|
| 809 |
+
"step": 1230
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 7.13,
|
| 813 |
+
"learning_rate": 2.137931034482759e-05,
|
| 814 |
+
"loss": 1.8681,
|
| 815 |
+
"step": 1240
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 7.18,
|
| 819 |
+
"learning_rate": 2.1551724137931033e-05,
|
| 820 |
+
"loss": 1.8791,
|
| 821 |
+
"step": 1250
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 7.24,
|
| 825 |
+
"learning_rate": 2.1724137931034484e-05,
|
| 826 |
+
"loss": 1.8906,
|
| 827 |
+
"step": 1260
|
| 828 |
+
},
|
| 829 |
+
{
|
| 830 |
+
"epoch": 7.3,
|
| 831 |
+
"learning_rate": 2.1896551724137932e-05,
|
| 832 |
+
"loss": 2.0289,
|
| 833 |
+
"step": 1270
|
| 834 |
+
},
|
| 835 |
+
{
|
| 836 |
+
"epoch": 7.36,
|
| 837 |
+
"learning_rate": 2.206896551724138e-05,
|
| 838 |
+
"loss": 1.8865,
|
| 839 |
+
"step": 1280
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 7.41,
|
| 843 |
+
"learning_rate": 2.2241379310344828e-05,
|
| 844 |
+
"loss": 1.8217,
|
| 845 |
+
"step": 1290
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 7.47,
|
| 849 |
+
"learning_rate": 2.2413793103448276e-05,
|
| 850 |
+
"loss": 1.7755,
|
| 851 |
+
"step": 1300
|
| 852 |
+
},
|
| 853 |
+
{
|
| 854 |
+
"epoch": 7.53,
|
| 855 |
+
"learning_rate": 2.2586206896551727e-05,
|
| 856 |
+
"loss": 1.6638,
|
| 857 |
+
"step": 1310
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 7.59,
|
| 861 |
+
"learning_rate": 2.275862068965517e-05,
|
| 862 |
+
"loss": 1.7748,
|
| 863 |
+
"step": 1320
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 7.64,
|
| 867 |
+
"learning_rate": 2.2931034482758623e-05,
|
| 868 |
+
"loss": 1.8823,
|
| 869 |
+
"step": 1330
|
| 870 |
+
},
|
| 871 |
+
{
|
| 872 |
+
"epoch": 7.7,
|
| 873 |
+
"learning_rate": 2.3103448275862067e-05,
|
| 874 |
+
"loss": 1.6694,
|
| 875 |
+
"step": 1340
|
| 876 |
+
},
|
| 877 |
+
{
|
| 878 |
+
"epoch": 7.76,
|
| 879 |
+
"learning_rate": 2.327586206896552e-05,
|
| 880 |
+
"loss": 1.9103,
|
| 881 |
+
"step": 1350
|
| 882 |
+
},
|
| 883 |
+
{
|
| 884 |
+
"epoch": 7.82,
|
| 885 |
+
"learning_rate": 2.3448275862068967e-05,
|
| 886 |
+
"loss": 2.0666,
|
| 887 |
+
"step": 1360
|
| 888 |
+
},
|
| 889 |
+
{
|
| 890 |
+
"epoch": 7.87,
|
| 891 |
+
"learning_rate": 2.3620689655172415e-05,
|
| 892 |
+
"loss": 1.6608,
|
| 893 |
+
"step": 1370
|
| 894 |
+
},
|
| 895 |
+
{
|
| 896 |
+
"epoch": 7.93,
|
| 897 |
+
"learning_rate": 2.3793103448275862e-05,
|
| 898 |
+
"loss": 1.8385,
|
| 899 |
+
"step": 1380
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 7.99,
|
| 903 |
+
"learning_rate": 2.396551724137931e-05,
|
| 904 |
+
"loss": 2.0224,
|
| 905 |
+
"step": 1390
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 8.0,
|
| 909 |
+
"eval_f1": 0.32926829268292684,
|
| 910 |
+
"eval_loss": 2.060194730758667,
|
| 911 |
+
"eval_runtime": 12.5242,
|
| 912 |
+
"eval_samples_per_second": 19.642,
|
| 913 |
+
"eval_steps_per_second": 9.821,
|
| 914 |
+
"step": 1392
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"epoch": 8.05,
|
| 918 |
+
"learning_rate": 2.4137931034482758e-05,
|
| 919 |
+
"loss": 1.831,
|
| 920 |
+
"step": 1400
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 8.1,
|
| 924 |
+
"learning_rate": 2.4310344827586206e-05,
|
| 925 |
+
"loss": 1.7572,
|
| 926 |
+
"step": 1410
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 8.16,
|
| 930 |
+
"learning_rate": 2.4482758620689657e-05,
|
| 931 |
+
"loss": 1.8719,
|
| 932 |
+
"step": 1420
|
| 933 |
+
},
|
| 934 |
+
{
|
| 935 |
+
"epoch": 8.22,
|
| 936 |
+
"learning_rate": 2.4655172413793105e-05,
|
| 937 |
+
"loss": 1.8331,
|
| 938 |
+
"step": 1430
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 8.28,
|
| 942 |
+
"learning_rate": 2.4827586206896553e-05,
|
| 943 |
+
"loss": 1.9726,
|
| 944 |
+
"step": 1440
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 8.33,
|
| 948 |
+
"learning_rate": 2.5e-05,
|
| 949 |
+
"loss": 1.7639,
|
| 950 |
+
"step": 1450
|
| 951 |
+
},
|
| 952 |
+
{
|
| 953 |
+
"epoch": 8.39,
|
| 954 |
+
"learning_rate": 2.517241379310345e-05,
|
| 955 |
+
"loss": 1.9096,
|
| 956 |
+
"step": 1460
|
| 957 |
+
},
|
| 958 |
+
{
|
| 959 |
+
"epoch": 8.45,
|
| 960 |
+
"learning_rate": 2.5344827586206897e-05,
|
| 961 |
+
"loss": 1.8732,
|
| 962 |
+
"step": 1470
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 8.51,
|
| 966 |
+
"learning_rate": 2.5517241379310345e-05,
|
| 967 |
+
"loss": 1.5993,
|
| 968 |
+
"step": 1480
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 8.56,
|
| 972 |
+
"learning_rate": 2.5689655172413793e-05,
|
| 973 |
+
"loss": 1.7585,
|
| 974 |
+
"step": 1490
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 8.62,
|
| 978 |
+
"learning_rate": 2.586206896551724e-05,
|
| 979 |
+
"loss": 1.8841,
|
| 980 |
+
"step": 1500
|
| 981 |
+
},
|
| 982 |
+
{
|
| 983 |
+
"epoch": 8.68,
|
| 984 |
+
"learning_rate": 2.603448275862069e-05,
|
| 985 |
+
"loss": 1.5775,
|
| 986 |
+
"step": 1510
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 8.74,
|
| 990 |
+
"learning_rate": 2.620689655172414e-05,
|
| 991 |
+
"loss": 1.7454,
|
| 992 |
+
"step": 1520
|
| 993 |
+
},
|
| 994 |
+
{
|
| 995 |
+
"epoch": 8.79,
|
| 996 |
+
"learning_rate": 2.6379310344827588e-05,
|
| 997 |
+
"loss": 1.9249,
|
| 998 |
+
"step": 1530
|
| 999 |
+
},
|
| 1000 |
+
{
|
| 1001 |
+
"epoch": 8.85,
|
| 1002 |
+
"learning_rate": 2.6551724137931036e-05,
|
| 1003 |
+
"loss": 1.8351,
|
| 1004 |
+
"step": 1540
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 8.91,
|
| 1008 |
+
"learning_rate": 2.6724137931034484e-05,
|
| 1009 |
+
"loss": 1.6522,
|
| 1010 |
+
"step": 1550
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 8.97,
|
| 1014 |
+
"learning_rate": 2.689655172413793e-05,
|
| 1015 |
+
"loss": 1.8086,
|
| 1016 |
+
"step": 1560
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"epoch": 9.0,
|
| 1020 |
+
"eval_f1": 0.32926829268292684,
|
| 1021 |
+
"eval_loss": 2.068312406539917,
|
| 1022 |
+
"eval_runtime": 12.1374,
|
| 1023 |
+
"eval_samples_per_second": 20.268,
|
| 1024 |
+
"eval_steps_per_second": 10.134,
|
| 1025 |
+
"step": 1566
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 9.02,
|
| 1029 |
+
"learning_rate": 2.706896551724138e-05,
|
| 1030 |
+
"loss": 1.7357,
|
| 1031 |
+
"step": 1570
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 9.08,
|
| 1035 |
+
"learning_rate": 2.7241379310344827e-05,
|
| 1036 |
+
"loss": 1.6286,
|
| 1037 |
+
"step": 1580
|
| 1038 |
+
},
|
| 1039 |
+
{
|
| 1040 |
+
"epoch": 9.14,
|
| 1041 |
+
"learning_rate": 2.741379310344828e-05,
|
| 1042 |
+
"loss": 1.6207,
|
| 1043 |
+
"step": 1590
|
| 1044 |
+
},
|
| 1045 |
+
{
|
| 1046 |
+
"epoch": 9.2,
|
| 1047 |
+
"learning_rate": 2.7586206896551723e-05,
|
| 1048 |
+
"loss": 1.6797,
|
| 1049 |
+
"step": 1600
|
| 1050 |
+
},
|
| 1051 |
+
{
|
| 1052 |
+
"epoch": 9.25,
|
| 1053 |
+
"learning_rate": 2.7758620689655175e-05,
|
| 1054 |
+
"loss": 1.6136,
|
| 1055 |
+
"step": 1610
|
| 1056 |
+
},
|
| 1057 |
+
{
|
| 1058 |
+
"epoch": 9.31,
|
| 1059 |
+
"learning_rate": 2.793103448275862e-05,
|
| 1060 |
+
"loss": 1.5962,
|
| 1061 |
+
"step": 1620
|
| 1062 |
+
},
|
| 1063 |
+
{
|
| 1064 |
+
"epoch": 9.37,
|
| 1065 |
+
"learning_rate": 2.810344827586207e-05,
|
| 1066 |
+
"loss": 1.8679,
|
| 1067 |
+
"step": 1630
|
| 1068 |
+
},
|
| 1069 |
+
{
|
| 1070 |
+
"epoch": 9.43,
|
| 1071 |
+
"learning_rate": 2.8275862068965518e-05,
|
| 1072 |
+
"loss": 1.9907,
|
| 1073 |
+
"step": 1640
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 9.48,
|
| 1077 |
+
"learning_rate": 2.8448275862068966e-05,
|
| 1078 |
+
"loss": 1.8782,
|
| 1079 |
+
"step": 1650
|
| 1080 |
+
},
|
| 1081 |
+
{
|
| 1082 |
+
"epoch": 9.54,
|
| 1083 |
+
"learning_rate": 2.8620689655172414e-05,
|
| 1084 |
+
"loss": 1.6307,
|
| 1085 |
+
"step": 1660
|
| 1086 |
+
},
|
| 1087 |
+
{
|
| 1088 |
+
"epoch": 9.6,
|
| 1089 |
+
"learning_rate": 2.8793103448275862e-05,
|
| 1090 |
+
"loss": 1.7819,
|
| 1091 |
+
"step": 1670
|
| 1092 |
+
},
|
| 1093 |
+
{
|
| 1094 |
+
"epoch": 9.66,
|
| 1095 |
+
"learning_rate": 2.8965517241379313e-05,
|
| 1096 |
+
"loss": 1.9863,
|
| 1097 |
+
"step": 1680
|
| 1098 |
+
},
|
| 1099 |
+
{
|
| 1100 |
+
"epoch": 9.71,
|
| 1101 |
+
"learning_rate": 2.9137931034482758e-05,
|
| 1102 |
+
"loss": 1.8648,
|
| 1103 |
+
"step": 1690
|
| 1104 |
+
},
|
| 1105 |
+
{
|
| 1106 |
+
"epoch": 9.77,
|
| 1107 |
+
"learning_rate": 2.931034482758621e-05,
|
| 1108 |
+
"loss": 1.8474,
|
| 1109 |
+
"step": 1700
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 9.83,
|
| 1113 |
+
"learning_rate": 2.9482758620689654e-05,
|
| 1114 |
+
"loss": 1.9404,
|
| 1115 |
+
"step": 1710
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 9.89,
|
| 1119 |
+
"learning_rate": 2.9655172413793105e-05,
|
| 1120 |
+
"loss": 1.9395,
|
| 1121 |
+
"step": 1720
|
| 1122 |
+
},
|
| 1123 |
+
{
|
| 1124 |
+
"epoch": 9.94,
|
| 1125 |
+
"learning_rate": 2.982758620689655e-05,
|
| 1126 |
+
"loss": 1.6639,
|
| 1127 |
+
"step": 1730
|
| 1128 |
+
},
|
| 1129 |
+
{
|
| 1130 |
+
"epoch": 10.0,
|
| 1131 |
+
"learning_rate": 3e-05,
|
| 1132 |
+
"loss": 1.8941,
|
| 1133 |
+
"step": 1740
|
| 1134 |
+
},
|
| 1135 |
+
{
|
| 1136 |
+
"epoch": 10.0,
|
| 1137 |
+
"eval_f1": 0.34552845528455284,
|
| 1138 |
+
"eval_loss": 2.014744997024536,
|
| 1139 |
+
"eval_runtime": 12.1182,
|
| 1140 |
+
"eval_samples_per_second": 20.3,
|
| 1141 |
+
"eval_steps_per_second": 10.15,
|
| 1142 |
+
"step": 1740
|
| 1143 |
+
},
|
| 1144 |
+
{
|
| 1145 |
+
"epoch": 10.06,
|
| 1146 |
+
"learning_rate": 2.9980842911877396e-05,
|
| 1147 |
+
"loss": 1.6203,
|
| 1148 |
+
"step": 1750
|
| 1149 |
+
},
|
| 1150 |
+
{
|
| 1151 |
+
"epoch": 10.11,
|
| 1152 |
+
"learning_rate": 2.9961685823754788e-05,
|
| 1153 |
+
"loss": 1.6125,
|
| 1154 |
+
"step": 1760
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 10.17,
|
| 1158 |
+
"learning_rate": 2.9942528735632184e-05,
|
| 1159 |
+
"loss": 1.5883,
|
| 1160 |
+
"step": 1770
|
| 1161 |
+
},
|
| 1162 |
+
{
|
| 1163 |
+
"epoch": 10.23,
|
| 1164 |
+
"learning_rate": 2.992337164750958e-05,
|
| 1165 |
+
"loss": 1.658,
|
| 1166 |
+
"step": 1780
|
| 1167 |
+
},
|
| 1168 |
+
{
|
| 1169 |
+
"epoch": 10.29,
|
| 1170 |
+
"learning_rate": 2.9904214559386975e-05,
|
| 1171 |
+
"loss": 1.5954,
|
| 1172 |
+
"step": 1790
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 10.34,
|
| 1176 |
+
"learning_rate": 2.9885057471264367e-05,
|
| 1177 |
+
"loss": 1.7201,
|
| 1178 |
+
"step": 1800
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 10.4,
|
| 1182 |
+
"learning_rate": 2.9865900383141762e-05,
|
| 1183 |
+
"loss": 1.5687,
|
| 1184 |
+
"step": 1810
|
| 1185 |
+
},
|
| 1186 |
+
{
|
| 1187 |
+
"epoch": 10.46,
|
| 1188 |
+
"learning_rate": 2.9846743295019157e-05,
|
| 1189 |
+
"loss": 1.7798,
|
| 1190 |
+
"step": 1820
|
| 1191 |
+
},
|
| 1192 |
+
{
|
| 1193 |
+
"epoch": 10.52,
|
| 1194 |
+
"learning_rate": 2.982758620689655e-05,
|
| 1195 |
+
"loss": 1.684,
|
| 1196 |
+
"step": 1830
|
| 1197 |
+
},
|
| 1198 |
+
{
|
| 1199 |
+
"epoch": 10.57,
|
| 1200 |
+
"learning_rate": 2.9808429118773948e-05,
|
| 1201 |
+
"loss": 1.6633,
|
| 1202 |
+
"step": 1840
|
| 1203 |
+
},
|
| 1204 |
+
{
|
| 1205 |
+
"epoch": 10.63,
|
| 1206 |
+
"learning_rate": 2.9789272030651344e-05,
|
| 1207 |
+
"loss": 1.8764,
|
| 1208 |
+
"step": 1850
|
| 1209 |
+
},
|
| 1210 |
+
{
|
| 1211 |
+
"epoch": 10.69,
|
| 1212 |
+
"learning_rate": 2.977011494252874e-05,
|
| 1213 |
+
"loss": 1.7563,
|
| 1214 |
+
"step": 1860
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 10.75,
|
| 1218 |
+
"learning_rate": 2.975095785440613e-05,
|
| 1219 |
+
"loss": 1.7403,
|
| 1220 |
+
"step": 1870
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 10.8,
|
| 1224 |
+
"learning_rate": 2.9731800766283527e-05,
|
| 1225 |
+
"loss": 1.5366,
|
| 1226 |
+
"step": 1880
|
| 1227 |
+
},
|
| 1228 |
+
{
|
| 1229 |
+
"epoch": 10.86,
|
| 1230 |
+
"learning_rate": 2.9712643678160922e-05,
|
| 1231 |
+
"loss": 1.6187,
|
| 1232 |
+
"step": 1890
|
| 1233 |
+
},
|
| 1234 |
+
{
|
| 1235 |
+
"epoch": 10.92,
|
| 1236 |
+
"learning_rate": 2.9693486590038314e-05,
|
| 1237 |
+
"loss": 1.8297,
|
| 1238 |
+
"step": 1900
|
| 1239 |
+
},
|
| 1240 |
+
{
|
| 1241 |
+
"epoch": 10.98,
|
| 1242 |
+
"learning_rate": 2.967432950191571e-05,
|
| 1243 |
+
"loss": 1.7848,
|
| 1244 |
+
"step": 1910
|
| 1245 |
+
},
|
| 1246 |
+
{
|
| 1247 |
+
"epoch": 11.0,
|
| 1248 |
+
"eval_f1": 0.4268292682926829,
|
| 1249 |
+
"eval_loss": 1.944414734840393,
|
| 1250 |
+
"eval_runtime": 12.1026,
|
| 1251 |
+
"eval_samples_per_second": 20.326,
|
| 1252 |
+
"eval_steps_per_second": 10.163,
|
| 1253 |
+
"step": 1914
|
| 1254 |
+
},
|
| 1255 |
+
{
|
| 1256 |
+
"epoch": 11.03,
|
| 1257 |
+
"learning_rate": 2.9655172413793105e-05,
|
| 1258 |
+
"loss": 1.6721,
|
| 1259 |
+
"step": 1920
|
| 1260 |
+
},
|
| 1261 |
+
{
|
| 1262 |
+
"epoch": 11.09,
|
| 1263 |
+
"learning_rate": 2.96360153256705e-05,
|
| 1264 |
+
"loss": 1.3034,
|
| 1265 |
+
"step": 1930
|
| 1266 |
+
},
|
| 1267 |
+
{
|
| 1268 |
+
"epoch": 11.15,
|
| 1269 |
+
"learning_rate": 2.9616858237547892e-05,
|
| 1270 |
+
"loss": 1.6881,
|
| 1271 |
+
"step": 1940
|
| 1272 |
+
},
|
| 1273 |
+
{
|
| 1274 |
+
"epoch": 11.21,
|
| 1275 |
+
"learning_rate": 2.9597701149425288e-05,
|
| 1276 |
+
"loss": 1.4882,
|
| 1277 |
+
"step": 1950
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 11.26,
|
| 1281 |
+
"learning_rate": 2.9578544061302683e-05,
|
| 1282 |
+
"loss": 1.4722,
|
| 1283 |
+
"step": 1960
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 11.32,
|
| 1287 |
+
"learning_rate": 2.9559386973180075e-05,
|
| 1288 |
+
"loss": 1.5784,
|
| 1289 |
+
"step": 1970
|
| 1290 |
+
},
|
| 1291 |
+
{
|
| 1292 |
+
"epoch": 11.38,
|
| 1293 |
+
"learning_rate": 2.954022988505747e-05,
|
| 1294 |
+
"loss": 1.5223,
|
| 1295 |
+
"step": 1980
|
| 1296 |
+
},
|
| 1297 |
+
{
|
| 1298 |
+
"epoch": 11.44,
|
| 1299 |
+
"learning_rate": 2.9521072796934866e-05,
|
| 1300 |
+
"loss": 1.6843,
|
| 1301 |
+
"step": 1990
|
| 1302 |
+
},
|
| 1303 |
+
{
|
| 1304 |
+
"epoch": 11.49,
|
| 1305 |
+
"learning_rate": 2.950191570881226e-05,
|
| 1306 |
+
"loss": 1.6447,
|
| 1307 |
+
"step": 2000
|
| 1308 |
+
},
|
| 1309 |
+
{
|
| 1310 |
+
"epoch": 11.55,
|
| 1311 |
+
"learning_rate": 2.9482758620689654e-05,
|
| 1312 |
+
"loss": 1.6076,
|
| 1313 |
+
"step": 2010
|
| 1314 |
+
},
|
| 1315 |
+
{
|
| 1316 |
+
"epoch": 11.61,
|
| 1317 |
+
"learning_rate": 2.946360153256705e-05,
|
| 1318 |
+
"loss": 1.6644,
|
| 1319 |
+
"step": 2020
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 11.67,
|
| 1323 |
+
"learning_rate": 2.9444444444444445e-05,
|
| 1324 |
+
"loss": 1.7464,
|
| 1325 |
+
"step": 2030
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 11.72,
|
| 1329 |
+
"learning_rate": 2.942528735632184e-05,
|
| 1330 |
+
"loss": 1.5289,
|
| 1331 |
+
"step": 2040
|
| 1332 |
+
},
|
| 1333 |
+
{
|
| 1334 |
+
"epoch": 11.78,
|
| 1335 |
+
"learning_rate": 2.9406130268199235e-05,
|
| 1336 |
+
"loss": 1.6279,
|
| 1337 |
+
"step": 2050
|
| 1338 |
+
},
|
| 1339 |
+
{
|
| 1340 |
+
"epoch": 11.84,
|
| 1341 |
+
"learning_rate": 2.938697318007663e-05,
|
| 1342 |
+
"loss": 1.5911,
|
| 1343 |
+
"step": 2060
|
| 1344 |
+
},
|
| 1345 |
+
{
|
| 1346 |
+
"epoch": 11.9,
|
| 1347 |
+
"learning_rate": 2.9367816091954026e-05,
|
| 1348 |
+
"loss": 1.6629,
|
| 1349 |
+
"step": 2070
|
| 1350 |
+
},
|
| 1351 |
+
{
|
| 1352 |
+
"epoch": 11.95,
|
| 1353 |
+
"learning_rate": 2.9348659003831418e-05,
|
| 1354 |
+
"loss": 1.4013,
|
| 1355 |
+
"step": 2080
|
| 1356 |
+
},
|
| 1357 |
+
{
|
| 1358 |
+
"epoch": 12.0,
|
| 1359 |
+
"eval_f1": 0.36585365853658536,
|
| 1360 |
+
"eval_loss": 1.9624298810958862,
|
| 1361 |
+
"eval_runtime": 11.265,
|
| 1362 |
+
"eval_samples_per_second": 21.838,
|
| 1363 |
+
"eval_steps_per_second": 10.919,
|
| 1364 |
+
"step": 2088
|
| 1365 |
+
},
|
| 1366 |
+
{
|
| 1367 |
+
"epoch": 12.01,
|
| 1368 |
+
"learning_rate": 2.9329501915708814e-05,
|
| 1369 |
+
"loss": 1.4498,
|
| 1370 |
+
"step": 2090
|
| 1371 |
+
},
|
| 1372 |
+
{
|
| 1373 |
+
"epoch": 12.07,
|
| 1374 |
+
"learning_rate": 2.931034482758621e-05,
|
| 1375 |
+
"loss": 1.3999,
|
| 1376 |
+
"step": 2100
|
| 1377 |
+
},
|
| 1378 |
+
{
|
| 1379 |
+
"epoch": 12.13,
|
| 1380 |
+
"learning_rate": 2.92911877394636e-05,
|
| 1381 |
+
"loss": 1.3161,
|
| 1382 |
+
"step": 2110
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 12.18,
|
| 1386 |
+
"learning_rate": 2.9272030651340997e-05,
|
| 1387 |
+
"loss": 1.3258,
|
| 1388 |
+
"step": 2120
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 12.24,
|
| 1392 |
+
"learning_rate": 2.9252873563218392e-05,
|
| 1393 |
+
"loss": 1.3642,
|
| 1394 |
+
"step": 2130
|
| 1395 |
+
},
|
| 1396 |
+
{
|
| 1397 |
+
"epoch": 12.3,
|
| 1398 |
+
"learning_rate": 2.9233716475095787e-05,
|
| 1399 |
+
"loss": 1.5188,
|
| 1400 |
+
"step": 2140
|
| 1401 |
+
},
|
| 1402 |
+
{
|
| 1403 |
+
"epoch": 12.36,
|
| 1404 |
+
"learning_rate": 2.921455938697318e-05,
|
| 1405 |
+
"loss": 1.6076,
|
| 1406 |
+
"step": 2150
|
| 1407 |
+
},
|
| 1408 |
+
{
|
| 1409 |
+
"epoch": 12.41,
|
| 1410 |
+
"learning_rate": 2.9195402298850575e-05,
|
| 1411 |
+
"loss": 1.6371,
|
| 1412 |
+
"step": 2160
|
| 1413 |
+
},
|
| 1414 |
+
{
|
| 1415 |
+
"epoch": 12.47,
|
| 1416 |
+
"learning_rate": 2.917624521072797e-05,
|
| 1417 |
+
"loss": 1.4424,
|
| 1418 |
+
"step": 2170
|
| 1419 |
+
},
|
| 1420 |
+
{
|
| 1421 |
+
"epoch": 12.53,
|
| 1422 |
+
"learning_rate": 2.9157088122605362e-05,
|
| 1423 |
+
"loss": 1.3859,
|
| 1424 |
+
"step": 2180
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 12.59,
|
| 1428 |
+
"learning_rate": 2.9137931034482758e-05,
|
| 1429 |
+
"loss": 1.4307,
|
| 1430 |
+
"step": 2190
|
| 1431 |
+
},
|
| 1432 |
+
{
|
| 1433 |
+
"epoch": 12.64,
|
| 1434 |
+
"learning_rate": 2.9118773946360153e-05,
|
| 1435 |
+
"loss": 1.5026,
|
| 1436 |
+
"step": 2200
|
| 1437 |
+
},
|
| 1438 |
+
{
|
| 1439 |
+
"epoch": 12.7,
|
| 1440 |
+
"learning_rate": 2.909961685823755e-05,
|
| 1441 |
+
"loss": 1.6566,
|
| 1442 |
+
"step": 2210
|
| 1443 |
+
},
|
| 1444 |
+
{
|
| 1445 |
+
"epoch": 12.76,
|
| 1446 |
+
"learning_rate": 2.908045977011494e-05,
|
| 1447 |
+
"loss": 1.6411,
|
| 1448 |
+
"step": 2220
|
| 1449 |
+
},
|
| 1450 |
+
{
|
| 1451 |
+
"epoch": 12.82,
|
| 1452 |
+
"learning_rate": 2.9061302681992336e-05,
|
| 1453 |
+
"loss": 1.574,
|
| 1454 |
+
"step": 2230
|
| 1455 |
+
},
|
| 1456 |
+
{
|
| 1457 |
+
"epoch": 12.87,
|
| 1458 |
+
"learning_rate": 2.9042145593869735e-05,
|
| 1459 |
+
"loss": 1.486,
|
| 1460 |
+
"step": 2240
|
| 1461 |
+
},
|
| 1462 |
+
{
|
| 1463 |
+
"epoch": 12.93,
|
| 1464 |
+
"learning_rate": 2.9022988505747127e-05,
|
| 1465 |
+
"loss": 1.6049,
|
| 1466 |
+
"step": 2250
|
| 1467 |
+
},
|
| 1468 |
+
{
|
| 1469 |
+
"epoch": 12.99,
|
| 1470 |
+
"learning_rate": 2.9003831417624522e-05,
|
| 1471 |
+
"loss": 1.3337,
|
| 1472 |
+
"step": 2260
|
| 1473 |
+
},
|
| 1474 |
+
{
|
| 1475 |
+
"epoch": 13.0,
|
| 1476 |
+
"eval_f1": 0.4227642276422765,
|
| 1477 |
+
"eval_loss": 1.7803982496261597,
|
| 1478 |
+
"eval_runtime": 11.4307,
|
| 1479 |
+
"eval_samples_per_second": 21.521,
|
| 1480 |
+
"eval_steps_per_second": 10.76,
|
| 1481 |
+
"step": 2262
|
| 1482 |
+
},
|
| 1483 |
+
{
|
| 1484 |
+
"epoch": 13.05,
|
| 1485 |
+
"learning_rate": 2.8984674329501918e-05,
|
| 1486 |
+
"loss": 1.4821,
|
| 1487 |
+
"step": 2270
|
| 1488 |
+
},
|
| 1489 |
+
{
|
| 1490 |
+
"epoch": 13.1,
|
| 1491 |
+
"learning_rate": 2.8965517241379313e-05,
|
| 1492 |
+
"loss": 1.4081,
|
| 1493 |
+
"step": 2280
|
| 1494 |
+
},
|
| 1495 |
+
{
|
| 1496 |
+
"epoch": 13.16,
|
| 1497 |
+
"learning_rate": 2.8946360153256705e-05,
|
| 1498 |
+
"loss": 1.3393,
|
| 1499 |
+
"step": 2290
|
| 1500 |
+
},
|
| 1501 |
+
{
|
| 1502 |
+
"epoch": 13.22,
|
| 1503 |
+
"learning_rate": 2.89272030651341e-05,
|
| 1504 |
+
"loss": 1.3755,
|
| 1505 |
+
"step": 2300
|
| 1506 |
+
},
|
| 1507 |
+
{
|
| 1508 |
+
"epoch": 13.28,
|
| 1509 |
+
"learning_rate": 2.8908045977011496e-05,
|
| 1510 |
+
"loss": 1.3658,
|
| 1511 |
+
"step": 2310
|
| 1512 |
+
},
|
| 1513 |
+
{
|
| 1514 |
+
"epoch": 13.33,
|
| 1515 |
+
"learning_rate": 2.8888888888888888e-05,
|
| 1516 |
+
"loss": 1.0842,
|
| 1517 |
+
"step": 2320
|
| 1518 |
+
},
|
| 1519 |
+
{
|
| 1520 |
+
"epoch": 13.39,
|
| 1521 |
+
"learning_rate": 2.8869731800766284e-05,
|
| 1522 |
+
"loss": 1.3182,
|
| 1523 |
+
"step": 2330
|
| 1524 |
+
},
|
| 1525 |
+
{
|
| 1526 |
+
"epoch": 13.45,
|
| 1527 |
+
"learning_rate": 2.885057471264368e-05,
|
| 1528 |
+
"loss": 1.4022,
|
| 1529 |
+
"step": 2340
|
| 1530 |
+
},
|
| 1531 |
+
{
|
| 1532 |
+
"epoch": 13.51,
|
| 1533 |
+
"learning_rate": 2.8831417624521075e-05,
|
| 1534 |
+
"loss": 1.3685,
|
| 1535 |
+
"step": 2350
|
| 1536 |
+
},
|
| 1537 |
+
{
|
| 1538 |
+
"epoch": 13.56,
|
| 1539 |
+
"learning_rate": 2.8812260536398467e-05,
|
| 1540 |
+
"loss": 1.488,
|
| 1541 |
+
"step": 2360
|
| 1542 |
+
},
|
| 1543 |
+
{
|
| 1544 |
+
"epoch": 13.62,
|
| 1545 |
+
"learning_rate": 2.8793103448275862e-05,
|
| 1546 |
+
"loss": 1.224,
|
| 1547 |
+
"step": 2370
|
| 1548 |
+
},
|
| 1549 |
+
{
|
| 1550 |
+
"epoch": 13.68,
|
| 1551 |
+
"learning_rate": 2.8773946360153257e-05,
|
| 1552 |
+
"loss": 1.3765,
|
| 1553 |
+
"step": 2380
|
| 1554 |
+
},
|
| 1555 |
+
{
|
| 1556 |
+
"epoch": 13.74,
|
| 1557 |
+
"learning_rate": 2.875478927203065e-05,
|
| 1558 |
+
"loss": 1.3204,
|
| 1559 |
+
"step": 2390
|
| 1560 |
+
},
|
| 1561 |
+
{
|
| 1562 |
+
"epoch": 13.79,
|
| 1563 |
+
"learning_rate": 2.8735632183908045e-05,
|
| 1564 |
+
"loss": 1.2625,
|
| 1565 |
+
"step": 2400
|
| 1566 |
+
},
|
| 1567 |
+
{
|
| 1568 |
+
"epoch": 13.85,
|
| 1569 |
+
"learning_rate": 2.871647509578544e-05,
|
| 1570 |
+
"loss": 1.2118,
|
| 1571 |
+
"step": 2410
|
| 1572 |
+
},
|
| 1573 |
+
{
|
| 1574 |
+
"epoch": 13.91,
|
| 1575 |
+
"learning_rate": 2.8697318007662836e-05,
|
| 1576 |
+
"loss": 1.4499,
|
| 1577 |
+
"step": 2420
|
| 1578 |
+
},
|
| 1579 |
+
{
|
| 1580 |
+
"epoch": 13.97,
|
| 1581 |
+
"learning_rate": 2.8678160919540228e-05,
|
| 1582 |
+
"loss": 1.3977,
|
| 1583 |
+
"step": 2430
|
| 1584 |
+
},
|
| 1585 |
+
{
|
| 1586 |
+
"epoch": 14.0,
|
| 1587 |
+
"eval_f1": 0.3699186991869919,
|
| 1588 |
+
"eval_loss": 1.8904528617858887,
|
| 1589 |
+
"eval_runtime": 11.8691,
|
| 1590 |
+
"eval_samples_per_second": 20.726,
|
| 1591 |
+
"eval_steps_per_second": 10.363,
|
| 1592 |
+
"step": 2436
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 14.02,
|
| 1596 |
+
"learning_rate": 2.8659003831417627e-05,
|
| 1597 |
+
"loss": 1.6674,
|
| 1598 |
+
"step": 2440
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 14.08,
|
| 1602 |
+
"learning_rate": 2.8639846743295022e-05,
|
| 1603 |
+
"loss": 1.2464,
|
| 1604 |
+
"step": 2450
|
| 1605 |
+
},
|
| 1606 |
+
{
|
| 1607 |
+
"epoch": 14.14,
|
| 1608 |
+
"learning_rate": 2.8620689655172414e-05,
|
| 1609 |
+
"loss": 1.2784,
|
| 1610 |
+
"step": 2460
|
| 1611 |
+
},
|
| 1612 |
+
{
|
| 1613 |
+
"epoch": 14.2,
|
| 1614 |
+
"learning_rate": 2.860153256704981e-05,
|
| 1615 |
+
"loss": 1.4698,
|
| 1616 |
+
"step": 2470
|
| 1617 |
+
},
|
| 1618 |
+
{
|
| 1619 |
+
"epoch": 14.25,
|
| 1620 |
+
"learning_rate": 2.8582375478927205e-05,
|
| 1621 |
+
"loss": 1.4046,
|
| 1622 |
+
"step": 2480
|
| 1623 |
+
},
|
| 1624 |
+
{
|
| 1625 |
+
"epoch": 14.31,
|
| 1626 |
+
"learning_rate": 2.85632183908046e-05,
|
| 1627 |
+
"loss": 1.0906,
|
| 1628 |
+
"step": 2490
|
| 1629 |
+
},
|
| 1630 |
+
{
|
| 1631 |
+
"epoch": 14.37,
|
| 1632 |
+
"learning_rate": 2.8544061302681992e-05,
|
| 1633 |
+
"loss": 1.3381,
|
| 1634 |
+
"step": 2500
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 14.43,
|
| 1638 |
+
"learning_rate": 2.8524904214559388e-05,
|
| 1639 |
+
"loss": 1.1792,
|
| 1640 |
+
"step": 2510
|
| 1641 |
+
},
|
| 1642 |
+
{
|
| 1643 |
+
"epoch": 14.48,
|
| 1644 |
+
"learning_rate": 2.8505747126436783e-05,
|
| 1645 |
+
"loss": 1.2152,
|
| 1646 |
+
"step": 2520
|
| 1647 |
+
},
|
| 1648 |
+
{
|
| 1649 |
+
"epoch": 14.54,
|
| 1650 |
+
"learning_rate": 2.848659003831418e-05,
|
| 1651 |
+
"loss": 1.4155,
|
| 1652 |
+
"step": 2530
|
| 1653 |
+
},
|
| 1654 |
+
{
|
| 1655 |
+
"epoch": 14.6,
|
| 1656 |
+
"learning_rate": 2.846743295019157e-05,
|
| 1657 |
+
"loss": 1.1157,
|
| 1658 |
+
"step": 2540
|
| 1659 |
+
},
|
| 1660 |
+
{
|
| 1661 |
+
"epoch": 14.66,
|
| 1662 |
+
"learning_rate": 2.8448275862068966e-05,
|
| 1663 |
+
"loss": 1.2993,
|
| 1664 |
+
"step": 2550
|
| 1665 |
+
},
|
| 1666 |
+
{
|
| 1667 |
+
"epoch": 14.71,
|
| 1668 |
+
"learning_rate": 2.842911877394636e-05,
|
| 1669 |
+
"loss": 1.2198,
|
| 1670 |
+
"step": 2560
|
| 1671 |
+
},
|
| 1672 |
+
{
|
| 1673 |
+
"epoch": 14.77,
|
| 1674 |
+
"learning_rate": 2.8409961685823754e-05,
|
| 1675 |
+
"loss": 1.1076,
|
| 1676 |
+
"step": 2570
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"epoch": 14.83,
|
| 1680 |
+
"learning_rate": 2.839080459770115e-05,
|
| 1681 |
+
"loss": 1.1955,
|
| 1682 |
+
"step": 2580
|
| 1683 |
+
},
|
| 1684 |
+
{
|
| 1685 |
+
"epoch": 14.89,
|
| 1686 |
+
"learning_rate": 2.8371647509578545e-05,
|
| 1687 |
+
"loss": 1.2308,
|
| 1688 |
+
"step": 2590
|
| 1689 |
+
},
|
| 1690 |
+
{
|
| 1691 |
+
"epoch": 14.94,
|
| 1692 |
+
"learning_rate": 2.835249042145594e-05,
|
| 1693 |
+
"loss": 1.3558,
|
| 1694 |
+
"step": 2600
|
| 1695 |
+
},
|
| 1696 |
+
{
|
| 1697 |
+
"epoch": 15.0,
|
| 1698 |
+
"learning_rate": 2.8333333333333332e-05,
|
| 1699 |
+
"loss": 1.3117,
|
| 1700 |
+
"step": 2610
|
| 1701 |
+
},
|
| 1702 |
+
{
|
| 1703 |
+
"epoch": 15.0,
|
| 1704 |
+
"eval_f1": 0.4146341463414634,
|
| 1705 |
+
"eval_loss": 1.8130720853805542,
|
| 1706 |
+
"eval_runtime": 12.5653,
|
| 1707 |
+
"eval_samples_per_second": 19.578,
|
| 1708 |
+
"eval_steps_per_second": 9.789,
|
| 1709 |
+
"step": 2610
|
| 1710 |
+
},
|
| 1711 |
+
{
|
| 1712 |
+
"epoch": 15.06,
|
| 1713 |
+
"learning_rate": 2.8314176245210727e-05,
|
| 1714 |
+
"loss": 1.1162,
|
| 1715 |
+
"step": 2620
|
| 1716 |
+
},
|
| 1717 |
+
{
|
| 1718 |
+
"epoch": 15.11,
|
| 1719 |
+
"learning_rate": 2.8295019157088123e-05,
|
| 1720 |
+
"loss": 1.1019,
|
| 1721 |
+
"step": 2630
|
| 1722 |
+
},
|
| 1723 |
+
{
|
| 1724 |
+
"epoch": 15.17,
|
| 1725 |
+
"learning_rate": 2.8275862068965518e-05,
|
| 1726 |
+
"loss": 1.1809,
|
| 1727 |
+
"step": 2640
|
| 1728 |
+
},
|
| 1729 |
+
{
|
| 1730 |
+
"epoch": 15.23,
|
| 1731 |
+
"learning_rate": 2.8256704980842914e-05,
|
| 1732 |
+
"loss": 1.2734,
|
| 1733 |
+
"step": 2650
|
| 1734 |
+
},
|
| 1735 |
+
{
|
| 1736 |
+
"epoch": 15.29,
|
| 1737 |
+
"learning_rate": 2.823754789272031e-05,
|
| 1738 |
+
"loss": 1.0834,
|
| 1739 |
+
"step": 2660
|
| 1740 |
+
},
|
| 1741 |
+
{
|
| 1742 |
+
"epoch": 15.34,
|
| 1743 |
+
"learning_rate": 2.8218390804597705e-05,
|
| 1744 |
+
"loss": 1.173,
|
| 1745 |
+
"step": 2670
|
| 1746 |
+
},
|
| 1747 |
+
{
|
| 1748 |
+
"epoch": 15.4,
|
| 1749 |
+
"learning_rate": 2.8199233716475097e-05,
|
| 1750 |
+
"loss": 1.2238,
|
| 1751 |
+
"step": 2680
|
| 1752 |
+
},
|
| 1753 |
+
{
|
| 1754 |
+
"epoch": 15.46,
|
| 1755 |
+
"learning_rate": 2.8180076628352492e-05,
|
| 1756 |
+
"loss": 1.2348,
|
| 1757 |
+
"step": 2690
|
| 1758 |
+
},
|
| 1759 |
+
{
|
| 1760 |
+
"epoch": 15.52,
|
| 1761 |
+
"learning_rate": 2.8160919540229887e-05,
|
| 1762 |
+
"loss": 0.9774,
|
| 1763 |
+
"step": 2700
|
| 1764 |
+
},
|
| 1765 |
+
{
|
| 1766 |
+
"epoch": 15.57,
|
| 1767 |
+
"learning_rate": 2.814176245210728e-05,
|
| 1768 |
+
"loss": 1.1478,
|
| 1769 |
+
"step": 2710
|
| 1770 |
+
},
|
| 1771 |
+
{
|
| 1772 |
+
"epoch": 15.63,
|
| 1773 |
+
"learning_rate": 2.8122605363984675e-05,
|
| 1774 |
+
"loss": 1.2591,
|
| 1775 |
+
"step": 2720
|
| 1776 |
+
},
|
| 1777 |
+
{
|
| 1778 |
+
"epoch": 15.69,
|
| 1779 |
+
"learning_rate": 2.810344827586207e-05,
|
| 1780 |
+
"loss": 1.1669,
|
| 1781 |
+
"step": 2730
|
| 1782 |
+
},
|
| 1783 |
+
{
|
| 1784 |
+
"epoch": 15.75,
|
| 1785 |
+
"learning_rate": 2.8084291187739466e-05,
|
| 1786 |
+
"loss": 1.2834,
|
| 1787 |
+
"step": 2740
|
| 1788 |
+
},
|
| 1789 |
+
{
|
| 1790 |
+
"epoch": 15.8,
|
| 1791 |
+
"learning_rate": 2.8065134099616858e-05,
|
| 1792 |
+
"loss": 1.2256,
|
| 1793 |
+
"step": 2750
|
| 1794 |
+
},
|
| 1795 |
+
{
|
| 1796 |
+
"epoch": 15.86,
|
| 1797 |
+
"learning_rate": 2.8045977011494253e-05,
|
| 1798 |
+
"loss": 1.3441,
|
| 1799 |
+
"step": 2760
|
| 1800 |
+
},
|
| 1801 |
+
{
|
| 1802 |
+
"epoch": 15.92,
|
| 1803 |
+
"learning_rate": 2.802681992337165e-05,
|
| 1804 |
+
"loss": 1.1368,
|
| 1805 |
+
"step": 2770
|
| 1806 |
+
},
|
| 1807 |
+
{
|
| 1808 |
+
"epoch": 15.98,
|
| 1809 |
+
"learning_rate": 2.800766283524904e-05,
|
| 1810 |
+
"loss": 1.0474,
|
| 1811 |
+
"step": 2780
|
| 1812 |
+
},
|
| 1813 |
+
{
|
| 1814 |
+
"epoch": 16.0,
|
| 1815 |
+
"eval_f1": 0.44308943089430897,
|
| 1816 |
+
"eval_loss": 1.9097241163253784,
|
| 1817 |
+
"eval_runtime": 11.5235,
|
| 1818 |
+
"eval_samples_per_second": 21.348,
|
| 1819 |
+
"eval_steps_per_second": 10.674,
|
| 1820 |
+
"step": 2784
|
| 1821 |
+
},
|
| 1822 |
+
{
|
| 1823 |
+
"epoch": 16.03,
|
| 1824 |
+
"learning_rate": 2.7988505747126436e-05,
|
| 1825 |
+
"loss": 1.2468,
|
| 1826 |
+
"step": 2790
|
| 1827 |
+
},
|
| 1828 |
+
{
|
| 1829 |
+
"epoch": 16.09,
|
| 1830 |
+
"learning_rate": 2.796934865900383e-05,
|
| 1831 |
+
"loss": 1.2014,
|
| 1832 |
+
"step": 2800
|
| 1833 |
+
},
|
| 1834 |
+
{
|
| 1835 |
+
"epoch": 16.15,
|
| 1836 |
+
"learning_rate": 2.7950191570881227e-05,
|
| 1837 |
+
"loss": 1.0133,
|
| 1838 |
+
"step": 2810
|
| 1839 |
+
},
|
| 1840 |
+
{
|
| 1841 |
+
"epoch": 16.21,
|
| 1842 |
+
"learning_rate": 2.793103448275862e-05,
|
| 1843 |
+
"loss": 1.0917,
|
| 1844 |
+
"step": 2820
|
| 1845 |
+
},
|
| 1846 |
+
{
|
| 1847 |
+
"epoch": 16.26,
|
| 1848 |
+
"learning_rate": 2.7911877394636014e-05,
|
| 1849 |
+
"loss": 1.2575,
|
| 1850 |
+
"step": 2830
|
| 1851 |
+
},
|
| 1852 |
+
{
|
| 1853 |
+
"epoch": 16.32,
|
| 1854 |
+
"learning_rate": 2.7892720306513413e-05,
|
| 1855 |
+
"loss": 1.1548,
|
| 1856 |
+
"step": 2840
|
| 1857 |
+
},
|
| 1858 |
+
{
|
| 1859 |
+
"epoch": 16.38,
|
| 1860 |
+
"learning_rate": 2.7873563218390805e-05,
|
| 1861 |
+
"loss": 0.9297,
|
| 1862 |
+
"step": 2850
|
| 1863 |
+
},
|
| 1864 |
+
{
|
| 1865 |
+
"epoch": 16.44,
|
| 1866 |
+
"learning_rate": 2.78544061302682e-05,
|
| 1867 |
+
"loss": 1.0547,
|
| 1868 |
+
"step": 2860
|
| 1869 |
+
},
|
| 1870 |
+
{
|
| 1871 |
+
"epoch": 16.49,
|
| 1872 |
+
"learning_rate": 2.7835249042145596e-05,
|
| 1873 |
+
"loss": 0.924,
|
| 1874 |
+
"step": 2870
|
| 1875 |
+
},
|
| 1876 |
+
{
|
| 1877 |
+
"epoch": 16.55,
|
| 1878 |
+
"learning_rate": 2.781609195402299e-05,
|
| 1879 |
+
"loss": 1.0336,
|
| 1880 |
+
"step": 2880
|
| 1881 |
+
},
|
| 1882 |
+
{
|
| 1883 |
+
"epoch": 16.61,
|
| 1884 |
+
"learning_rate": 2.7796934865900384e-05,
|
| 1885 |
+
"loss": 0.9839,
|
| 1886 |
+
"step": 2890
|
| 1887 |
+
},
|
| 1888 |
+
{
|
| 1889 |
+
"epoch": 16.67,
|
| 1890 |
+
"learning_rate": 2.777777777777778e-05,
|
| 1891 |
+
"loss": 0.8042,
|
| 1892 |
+
"step": 2900
|
| 1893 |
+
},
|
| 1894 |
+
{
|
| 1895 |
+
"epoch": 16.72,
|
| 1896 |
+
"learning_rate": 2.7758620689655175e-05,
|
| 1897 |
+
"loss": 1.0443,
|
| 1898 |
+
"step": 2910
|
| 1899 |
+
},
|
| 1900 |
+
{
|
| 1901 |
+
"epoch": 16.78,
|
| 1902 |
+
"learning_rate": 2.7739463601532567e-05,
|
| 1903 |
+
"loss": 1.0883,
|
| 1904 |
+
"step": 2920
|
| 1905 |
+
},
|
| 1906 |
+
{
|
| 1907 |
+
"epoch": 16.84,
|
| 1908 |
+
"learning_rate": 2.7720306513409962e-05,
|
| 1909 |
+
"loss": 1.0444,
|
| 1910 |
+
"step": 2930
|
| 1911 |
+
},
|
| 1912 |
+
{
|
| 1913 |
+
"epoch": 16.9,
|
| 1914 |
+
"learning_rate": 2.7701149425287357e-05,
|
| 1915 |
+
"loss": 1.1755,
|
| 1916 |
+
"step": 2940
|
| 1917 |
+
},
|
| 1918 |
+
{
|
| 1919 |
+
"epoch": 16.95,
|
| 1920 |
+
"learning_rate": 2.7681992337164753e-05,
|
| 1921 |
+
"loss": 0.8636,
|
| 1922 |
+
"step": 2950
|
| 1923 |
+
},
|
| 1924 |
+
{
|
| 1925 |
+
"epoch": 17.0,
|
| 1926 |
+
"eval_f1": 0.45121951219512196,
|
| 1927 |
+
"eval_loss": 1.9556288719177246,
|
| 1928 |
+
"eval_runtime": 11.4652,
|
| 1929 |
+
"eval_samples_per_second": 21.456,
|
| 1930 |
+
"eval_steps_per_second": 10.728,
|
| 1931 |
+
"step": 2958
|
| 1932 |
+
}
|
| 1933 |
+
],
|
| 1934 |
+
"max_steps": 17400,
|
| 1935 |
+
"num_train_epochs": 100,
|
| 1936 |
+
"total_flos": 1.0921876643966929e+18,
|
| 1937 |
+
"trial_name": null,
|
| 1938 |
+
"trial_params": null
|
| 1939 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2ffc2e5758e23c41e4a7e8c2d4977aae6e0601990e6d5c1e7d561dfee2b0d0ed
|
| 3 |
+
size 3579
|