add model
Browse files- checkpoint-14000/config.json +88 -0
- checkpoint-14000/optimizer.pt +3 -0
- checkpoint-14000/preprocessor_config.json +9 -0
- checkpoint-14000/pytorch_model.bin +3 -0
- checkpoint-14000/rng_state.pth +3 -0
- checkpoint-14000/scaler.pt +3 -0
- checkpoint-14000/scheduler.pt +3 -0
- checkpoint-14000/trainer_state.json +310 -0
- checkpoint-14000/training_args.bin +3 -0
- checkpoint-3000/config.json +88 -0
- checkpoint-3000/optimizer.pt +3 -0
- checkpoint-3000/preprocessor_config.json +9 -0
- checkpoint-3000/pytorch_model.bin +3 -0
- checkpoint-3000/rng_state.pth +3 -0
- checkpoint-3000/scaler.pt +3 -0
- checkpoint-3000/scheduler.pt +3 -0
- checkpoint-3000/trainer_state.json +79 -0
- checkpoint-3000/training_args.bin +3 -0
- pytorch_model.bin +1 -1
checkpoint-14000/config.json
ADDED
|
@@ -0,0 +1,88 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "facebook/wav2vec2-base",
|
| 3 |
+
"activation_dropout": 0.0,
|
| 4 |
+
"apply_spec_augment": true,
|
| 5 |
+
"architectures": [
|
| 6 |
+
"Wav2Vec2ForCTC"
|
| 7 |
+
],
|
| 8 |
+
"attention_dropout": 0.1,
|
| 9 |
+
"bos_token_id": 1,
|
| 10 |
+
"classifier_proj_size": 256,
|
| 11 |
+
"codevector_dim": 256,
|
| 12 |
+
"contrastive_logits_temperature": 0.1,
|
| 13 |
+
"conv_bias": false,
|
| 14 |
+
"conv_dim": [
|
| 15 |
+
512,
|
| 16 |
+
512,
|
| 17 |
+
512,
|
| 18 |
+
512,
|
| 19 |
+
512,
|
| 20 |
+
512,
|
| 21 |
+
512
|
| 22 |
+
],
|
| 23 |
+
"conv_kernel": [
|
| 24 |
+
10,
|
| 25 |
+
3,
|
| 26 |
+
3,
|
| 27 |
+
3,
|
| 28 |
+
3,
|
| 29 |
+
2,
|
| 30 |
+
2
|
| 31 |
+
],
|
| 32 |
+
"conv_stride": [
|
| 33 |
+
5,
|
| 34 |
+
2,
|
| 35 |
+
2,
|
| 36 |
+
2,
|
| 37 |
+
2,
|
| 38 |
+
2,
|
| 39 |
+
2
|
| 40 |
+
],
|
| 41 |
+
"ctc_loss_reduction": "mean",
|
| 42 |
+
"ctc_zero_infinity": false,
|
| 43 |
+
"diversity_loss_weight": 0.1,
|
| 44 |
+
"do_stable_layer_norm": false,
|
| 45 |
+
"eos_token_id": 2,
|
| 46 |
+
"feat_extract_activation": "gelu",
|
| 47 |
+
"feat_extract_norm": "group",
|
| 48 |
+
"feat_proj_dropout": 0.1,
|
| 49 |
+
"feat_quantizer_dropout": 0.0,
|
| 50 |
+
"final_dropout": 0.0,
|
| 51 |
+
"freeze_feat_extract_train": true,
|
| 52 |
+
"hidden_act": "gelu",
|
| 53 |
+
"hidden_dropout": 0.1,
|
| 54 |
+
"hidden_size": 768,
|
| 55 |
+
"initializer_range": 0.02,
|
| 56 |
+
"intermediate_size": 3072,
|
| 57 |
+
"layer_norm_eps": 1e-05,
|
| 58 |
+
"layerdrop": 0.0,
|
| 59 |
+
"mask_channel_length": 10,
|
| 60 |
+
"mask_channel_min_space": 1,
|
| 61 |
+
"mask_channel_other": 0.0,
|
| 62 |
+
"mask_channel_prob": 0.0,
|
| 63 |
+
"mask_channel_selection": "static",
|
| 64 |
+
"mask_feature_length": 10,
|
| 65 |
+
"mask_feature_prob": 0.0,
|
| 66 |
+
"mask_time_length": 5,
|
| 67 |
+
"mask_time_min_space": 1,
|
| 68 |
+
"mask_time_other": 0.0,
|
| 69 |
+
"mask_time_prob": 0.05,
|
| 70 |
+
"mask_time_selection": "static",
|
| 71 |
+
"model_type": "wav2vec2",
|
| 72 |
+
"no_mask_channel_overlap": false,
|
| 73 |
+
"no_mask_time_overlap": false,
|
| 74 |
+
"num_attention_heads": 12,
|
| 75 |
+
"num_codevector_groups": 2,
|
| 76 |
+
"num_codevectors_per_group": 320,
|
| 77 |
+
"num_conv_pos_embedding_groups": 16,
|
| 78 |
+
"num_conv_pos_embeddings": 128,
|
| 79 |
+
"num_feat_extract_layers": 7,
|
| 80 |
+
"num_hidden_layers": 12,
|
| 81 |
+
"num_negatives": 100,
|
| 82 |
+
"pad_token_id": 29,
|
| 83 |
+
"proj_codevector_dim": 256,
|
| 84 |
+
"torch_dtype": "float32",
|
| 85 |
+
"transformers_version": "4.11.3",
|
| 86 |
+
"use_weighted_layer_sum": false,
|
| 87 |
+
"vocab_size": 32
|
| 88 |
+
}
|
checkpoint-14000/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2f50a431e15301d727e98a23578f39337b1db8f79d11e19f9b45d60b9760a462
|
| 3 |
+
size 721685265
|
checkpoint-14000/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 |
+
}
|
checkpoint-14000/pytorch_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:89caac3d689e0648d0f22781b687cfe200167410ea83146f8c429c12ce0f287c
|
| 3 |
+
size 377670039
|
checkpoint-14000/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7130112beb03f39dd80c175094607aa68c151619cf356c84f085e6e681f28deb
|
| 3 |
+
size 15587
|
checkpoint-14000/scaler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ee24a2d972e151c8045f79705300a28d9463c6be405a0666d2419f7389e40578
|
| 3 |
+
size 559
|
checkpoint-14000/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f2686397775ec7fcdb044e1e815e4df5d7a7a6d11ee121c08327c13f6d040a1d
|
| 3 |
+
size 623
|
checkpoint-14000/trainer_state.json
ADDED
|
@@ -0,0 +1,310 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.9328228831291199,
|
| 3 |
+
"best_model_checkpoint": "wav2vec2-base-swbd-turn-eos-long_short_utt_removed_3percent/checkpoint-3000",
|
| 4 |
+
"epoch": 19.525801952580196,
|
| 5 |
+
"global_step": 14000,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.7,
|
| 12 |
+
"learning_rate": 4.9800000000000004e-05,
|
| 13 |
+
"loss": 3.6678,
|
| 14 |
+
"step": 500
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 1.39,
|
| 18 |
+
"learning_rate": 9.970000000000001e-05,
|
| 19 |
+
"loss": 1.8593,
|
| 20 |
+
"step": 1000
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 1.39,
|
| 24 |
+
"eval_loss": 1.3255794048309326,
|
| 25 |
+
"eval_runtime": 303.7268,
|
| 26 |
+
"eval_samples_per_second": 4.488,
|
| 27 |
+
"eval_steps_per_second": 0.142,
|
| 28 |
+
"eval_wer": 0.5794877337079866,
|
| 29 |
+
"step": 1000
|
| 30 |
+
},
|
| 31 |
+
{
|
| 32 |
+
"epoch": 2.09,
|
| 33 |
+
"learning_rate": 9.627436281859072e-05,
|
| 34 |
+
"loss": 0.9996,
|
| 35 |
+
"step": 1500
|
| 36 |
+
},
|
| 37 |
+
{
|
| 38 |
+
"epoch": 2.79,
|
| 39 |
+
"learning_rate": 9.253373313343329e-05,
|
| 40 |
+
"loss": 0.7576,
|
| 41 |
+
"step": 2000
|
| 42 |
+
},
|
| 43 |
+
{
|
| 44 |
+
"epoch": 2.79,
|
| 45 |
+
"eval_loss": 1.0275607109069824,
|
| 46 |
+
"eval_runtime": 383.8668,
|
| 47 |
+
"eval_samples_per_second": 3.551,
|
| 48 |
+
"eval_steps_per_second": 0.112,
|
| 49 |
+
"eval_wer": 0.4334810331784286,
|
| 50 |
+
"step": 2000
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 3.49,
|
| 54 |
+
"learning_rate": 8.87856071964018e-05,
|
| 55 |
+
"loss": 0.6464,
|
| 56 |
+
"step": 2500
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 4.18,
|
| 60 |
+
"learning_rate": 8.503748125937032e-05,
|
| 61 |
+
"loss": 0.556,
|
| 62 |
+
"step": 3000
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 4.18,
|
| 66 |
+
"eval_loss": 0.9328228831291199,
|
| 67 |
+
"eval_runtime": 295.659,
|
| 68 |
+
"eval_samples_per_second": 4.61,
|
| 69 |
+
"eval_steps_per_second": 0.145,
|
| 70 |
+
"eval_wer": 0.3811376490507583,
|
| 71 |
+
"step": 3000
|
| 72 |
+
},
|
| 73 |
+
{
|
| 74 |
+
"epoch": 4.88,
|
| 75 |
+
"learning_rate": 8.128935532233884e-05,
|
| 76 |
+
"loss": 0.4897,
|
| 77 |
+
"step": 3500
|
| 78 |
+
},
|
| 79 |
+
{
|
| 80 |
+
"epoch": 5.58,
|
| 81 |
+
"learning_rate": 7.75487256371814e-05,
|
| 82 |
+
"loss": 0.4301,
|
| 83 |
+
"step": 4000
|
| 84 |
+
},
|
| 85 |
+
{
|
| 86 |
+
"epoch": 5.58,
|
| 87 |
+
"eval_loss": 0.9558691382408142,
|
| 88 |
+
"eval_runtime": 391.7361,
|
| 89 |
+
"eval_samples_per_second": 3.479,
|
| 90 |
+
"eval_steps_per_second": 0.11,
|
| 91 |
+
"eval_wer": 0.3691775640332865,
|
| 92 |
+
"step": 4000
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 6.28,
|
| 96 |
+
"learning_rate": 7.380059970014994e-05,
|
| 97 |
+
"loss": 0.3938,
|
| 98 |
+
"step": 4500
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 6.97,
|
| 102 |
+
"learning_rate": 7.005247376311844e-05,
|
| 103 |
+
"loss": 0.3649,
|
| 104 |
+
"step": 5000
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 6.97,
|
| 108 |
+
"eval_loss": 1.070428729057312,
|
| 109 |
+
"eval_runtime": 346.5542,
|
| 110 |
+
"eval_samples_per_second": 3.933,
|
| 111 |
+
"eval_steps_per_second": 0.124,
|
| 112 |
+
"eval_wer": 0.38769408119889043,
|
| 113 |
+
"step": 5000
|
| 114 |
+
},
|
| 115 |
+
{
|
| 116 |
+
"epoch": 7.67,
|
| 117 |
+
"learning_rate": 6.630434782608695e-05,
|
| 118 |
+
"loss": 0.3179,
|
| 119 |
+
"step": 5500
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 8.37,
|
| 123 |
+
"learning_rate": 6.255622188905547e-05,
|
| 124 |
+
"loss": 0.3064,
|
| 125 |
+
"step": 6000
|
| 126 |
+
},
|
| 127 |
+
{
|
| 128 |
+
"epoch": 8.37,
|
| 129 |
+
"eval_loss": 1.0000288486480713,
|
| 130 |
+
"eval_runtime": 352.1995,
|
| 131 |
+
"eval_samples_per_second": 3.87,
|
| 132 |
+
"eval_steps_per_second": 0.122,
|
| 133 |
+
"eval_wer": 0.33495442919413526,
|
| 134 |
+
"step": 6000
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 9.07,
|
| 138 |
+
"learning_rate": 5.8815592203898054e-05,
|
| 139 |
+
"loss": 0.2794,
|
| 140 |
+
"step": 6500
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 9.76,
|
| 144 |
+
"learning_rate": 5.5067466266866574e-05,
|
| 145 |
+
"loss": 0.259,
|
| 146 |
+
"step": 7000
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 9.76,
|
| 150 |
+
"eval_loss": 1.0542327165603638,
|
| 151 |
+
"eval_runtime": 347.0396,
|
| 152 |
+
"eval_samples_per_second": 3.928,
|
| 153 |
+
"eval_steps_per_second": 0.124,
|
| 154 |
+
"eval_wer": 0.3227421737094276,
|
| 155 |
+
"step": 7000
|
| 156 |
+
},
|
| 157 |
+
{
|
| 158 |
+
"epoch": 10.46,
|
| 159 |
+
"learning_rate": 5.131934032983509e-05,
|
| 160 |
+
"loss": 0.2359,
|
| 161 |
+
"step": 7500
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"epoch": 11.16,
|
| 165 |
+
"learning_rate": 4.7571214392803595e-05,
|
| 166 |
+
"loss": 0.2288,
|
| 167 |
+
"step": 8000
|
| 168 |
+
},
|
| 169 |
+
{
|
| 170 |
+
"epoch": 11.16,
|
| 171 |
+
"eval_loss": 1.0856573581695557,
|
| 172 |
+
"eval_runtime": 411.1438,
|
| 173 |
+
"eval_samples_per_second": 3.315,
|
| 174 |
+
"eval_steps_per_second": 0.105,
|
| 175 |
+
"eval_wer": 0.3141683778234086,
|
| 176 |
+
"step": 8000
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 11.85,
|
| 180 |
+
"learning_rate": 4.3823088455772116e-05,
|
| 181 |
+
"loss": 0.2083,
|
| 182 |
+
"step": 8500
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 12.55,
|
| 186 |
+
"learning_rate": 4.007496251874063e-05,
|
| 187 |
+
"loss": 0.1915,
|
| 188 |
+
"step": 9000
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 12.55,
|
| 192 |
+
"eval_loss": 1.0417132377624512,
|
| 193 |
+
"eval_runtime": 346.3488,
|
| 194 |
+
"eval_samples_per_second": 3.935,
|
| 195 |
+
"eval_steps_per_second": 0.124,
|
| 196 |
+
"eval_wer": 0.3155012788645124,
|
| 197 |
+
"step": 9000
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 13.25,
|
| 201 |
+
"learning_rate": 3.6326836581709144e-05,
|
| 202 |
+
"loss": 0.1805,
|
| 203 |
+
"step": 9500
|
| 204 |
+
},
|
| 205 |
+
{
|
| 206 |
+
"epoch": 13.95,
|
| 207 |
+
"learning_rate": 3.2578710644677665e-05,
|
| 208 |
+
"loss": 0.1706,
|
| 209 |
+
"step": 10000
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 13.95,
|
| 213 |
+
"eval_loss": 1.1116188764572144,
|
| 214 |
+
"eval_runtime": 345.9351,
|
| 215 |
+
"eval_samples_per_second": 3.94,
|
| 216 |
+
"eval_steps_per_second": 0.124,
|
| 217 |
+
"eval_wer": 0.31132245397888975,
|
| 218 |
+
"step": 10000
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 14.64,
|
| 222 |
+
"learning_rate": 2.883058470764618e-05,
|
| 223 |
+
"loss": 0.1577,
|
| 224 |
+
"step": 10500
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 15.34,
|
| 228 |
+
"learning_rate": 2.5089955022488758e-05,
|
| 229 |
+
"loss": 0.1444,
|
| 230 |
+
"step": 11000
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 15.34,
|
| 234 |
+
"eval_loss": 1.156752586364746,
|
| 235 |
+
"eval_runtime": 344.3787,
|
| 236 |
+
"eval_samples_per_second": 3.958,
|
| 237 |
+
"eval_steps_per_second": 0.125,
|
| 238 |
+
"eval_wer": 0.30908894412622934,
|
| 239 |
+
"step": 11000
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 16.04,
|
| 243 |
+
"learning_rate": 2.1349325337331336e-05,
|
| 244 |
+
"loss": 0.1396,
|
| 245 |
+
"step": 11500
|
| 246 |
+
},
|
| 247 |
+
{
|
| 248 |
+
"epoch": 16.74,
|
| 249 |
+
"learning_rate": 1.760119940029985e-05,
|
| 250 |
+
"loss": 0.1247,
|
| 251 |
+
"step": 12000
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 16.74,
|
| 255 |
+
"eval_loss": 1.1672117710113525,
|
| 256 |
+
"eval_runtime": 361.7016,
|
| 257 |
+
"eval_samples_per_second": 3.768,
|
| 258 |
+
"eval_steps_per_second": 0.119,
|
| 259 |
+
"eval_wer": 0.30195612233870095,
|
| 260 |
+
"step": 12000
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 17.43,
|
| 264 |
+
"learning_rate": 1.3853073463268368e-05,
|
| 265 |
+
"loss": 0.1239,
|
| 266 |
+
"step": 12500
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 18.13,
|
| 270 |
+
"learning_rate": 1.0104947526236883e-05,
|
| 271 |
+
"loss": 0.116,
|
| 272 |
+
"step": 13000
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 18.13,
|
| 276 |
+
"eval_loss": 1.1757261753082275,
|
| 277 |
+
"eval_runtime": 358.9537,
|
| 278 |
+
"eval_samples_per_second": 3.797,
|
| 279 |
+
"eval_steps_per_second": 0.12,
|
| 280 |
+
"eval_wer": 0.30591880110955005,
|
| 281 |
+
"step": 13000
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 18.83,
|
| 285 |
+
"learning_rate": 6.36431784107946e-06,
|
| 286 |
+
"loss": 0.1093,
|
| 287 |
+
"step": 13500
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"epoch": 19.53,
|
| 291 |
+
"learning_rate": 2.6161919040479763e-06,
|
| 292 |
+
"loss": 0.1057,
|
| 293 |
+
"step": 14000
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 19.53,
|
| 297 |
+
"eval_loss": 1.1997692584991455,
|
| 298 |
+
"eval_runtime": 297.9761,
|
| 299 |
+
"eval_samples_per_second": 4.574,
|
| 300 |
+
"eval_steps_per_second": 0.144,
|
| 301 |
+
"eval_wer": 0.30609892287186136,
|
| 302 |
+
"step": 14000
|
| 303 |
+
}
|
| 304 |
+
],
|
| 305 |
+
"max_steps": 14340,
|
| 306 |
+
"num_train_epochs": 20,
|
| 307 |
+
"total_flos": 2.5513525084170887e+19,
|
| 308 |
+
"trial_name": null,
|
| 309 |
+
"trial_params": null
|
| 310 |
+
}
|
checkpoint-14000/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:99ee4afa7ba3f16a4e5857ce3b6a0ab4232820185e063b7bb3ea94486802c2a2
|
| 3 |
+
size 2927
|
checkpoint-3000/config.json
ADDED
|
@@ -0,0 +1,88 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "facebook/wav2vec2-base",
|
| 3 |
+
"activation_dropout": 0.0,
|
| 4 |
+
"apply_spec_augment": true,
|
| 5 |
+
"architectures": [
|
| 6 |
+
"Wav2Vec2ForCTC"
|
| 7 |
+
],
|
| 8 |
+
"attention_dropout": 0.1,
|
| 9 |
+
"bos_token_id": 1,
|
| 10 |
+
"classifier_proj_size": 256,
|
| 11 |
+
"codevector_dim": 256,
|
| 12 |
+
"contrastive_logits_temperature": 0.1,
|
| 13 |
+
"conv_bias": false,
|
| 14 |
+
"conv_dim": [
|
| 15 |
+
512,
|
| 16 |
+
512,
|
| 17 |
+
512,
|
| 18 |
+
512,
|
| 19 |
+
512,
|
| 20 |
+
512,
|
| 21 |
+
512
|
| 22 |
+
],
|
| 23 |
+
"conv_kernel": [
|
| 24 |
+
10,
|
| 25 |
+
3,
|
| 26 |
+
3,
|
| 27 |
+
3,
|
| 28 |
+
3,
|
| 29 |
+
2,
|
| 30 |
+
2
|
| 31 |
+
],
|
| 32 |
+
"conv_stride": [
|
| 33 |
+
5,
|
| 34 |
+
2,
|
| 35 |
+
2,
|
| 36 |
+
2,
|
| 37 |
+
2,
|
| 38 |
+
2,
|
| 39 |
+
2
|
| 40 |
+
],
|
| 41 |
+
"ctc_loss_reduction": "mean",
|
| 42 |
+
"ctc_zero_infinity": false,
|
| 43 |
+
"diversity_loss_weight": 0.1,
|
| 44 |
+
"do_stable_layer_norm": false,
|
| 45 |
+
"eos_token_id": 2,
|
| 46 |
+
"feat_extract_activation": "gelu",
|
| 47 |
+
"feat_extract_norm": "group",
|
| 48 |
+
"feat_proj_dropout": 0.1,
|
| 49 |
+
"feat_quantizer_dropout": 0.0,
|
| 50 |
+
"final_dropout": 0.0,
|
| 51 |
+
"freeze_feat_extract_train": true,
|
| 52 |
+
"hidden_act": "gelu",
|
| 53 |
+
"hidden_dropout": 0.1,
|
| 54 |
+
"hidden_size": 768,
|
| 55 |
+
"initializer_range": 0.02,
|
| 56 |
+
"intermediate_size": 3072,
|
| 57 |
+
"layer_norm_eps": 1e-05,
|
| 58 |
+
"layerdrop": 0.0,
|
| 59 |
+
"mask_channel_length": 10,
|
| 60 |
+
"mask_channel_min_space": 1,
|
| 61 |
+
"mask_channel_other": 0.0,
|
| 62 |
+
"mask_channel_prob": 0.0,
|
| 63 |
+
"mask_channel_selection": "static",
|
| 64 |
+
"mask_feature_length": 10,
|
| 65 |
+
"mask_feature_prob": 0.0,
|
| 66 |
+
"mask_time_length": 5,
|
| 67 |
+
"mask_time_min_space": 1,
|
| 68 |
+
"mask_time_other": 0.0,
|
| 69 |
+
"mask_time_prob": 0.05,
|
| 70 |
+
"mask_time_selection": "static",
|
| 71 |
+
"model_type": "wav2vec2",
|
| 72 |
+
"no_mask_channel_overlap": false,
|
| 73 |
+
"no_mask_time_overlap": false,
|
| 74 |
+
"num_attention_heads": 12,
|
| 75 |
+
"num_codevector_groups": 2,
|
| 76 |
+
"num_codevectors_per_group": 320,
|
| 77 |
+
"num_conv_pos_embedding_groups": 16,
|
| 78 |
+
"num_conv_pos_embeddings": 128,
|
| 79 |
+
"num_feat_extract_layers": 7,
|
| 80 |
+
"num_hidden_layers": 12,
|
| 81 |
+
"num_negatives": 100,
|
| 82 |
+
"pad_token_id": 29,
|
| 83 |
+
"proj_codevector_dim": 256,
|
| 84 |
+
"torch_dtype": "float32",
|
| 85 |
+
"transformers_version": "4.11.3",
|
| 86 |
+
"use_weighted_layer_sum": false,
|
| 87 |
+
"vocab_size": 32
|
| 88 |
+
}
|
checkpoint-3000/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:332cfe6359068e4a5bff1a1b55b5f385f70b64da8f2e4dbc9fdf50348078f435
|
| 3 |
+
size 721685265
|
checkpoint-3000/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 |
+
}
|
checkpoint-3000/pytorch_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f2bded6ea77f057eaa95495f6efc4859e88b57b6a96a722dcaf6477de052c46b
|
| 3 |
+
size 377670039
|
checkpoint-3000/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:60f02c821b5f0735b1cbfdf897cc84a8179f0aa2c3c140fbd69901a6d2f008b8
|
| 3 |
+
size 15523
|
checkpoint-3000/scaler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1caea6d43c13149ecea564ac3bc6fae6e827b7d3c34dc7f777a288dbb3b3c5e3
|
| 3 |
+
size 559
|
checkpoint-3000/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cc6ad5795ee101c9e093b7d3192f0732dbe33d4d16ba88e400c4511f02a7403f
|
| 3 |
+
size 623
|
checkpoint-3000/trainer_state.json
ADDED
|
@@ -0,0 +1,79 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.9328228831291199,
|
| 3 |
+
"best_model_checkpoint": "wav2vec2-base-swbd-turn-eos-long_short_utt_removed_3percent/checkpoint-3000",
|
| 4 |
+
"epoch": 4.184100418410042,
|
| 5 |
+
"global_step": 3000,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.7,
|
| 12 |
+
"learning_rate": 4.9800000000000004e-05,
|
| 13 |
+
"loss": 3.6678,
|
| 14 |
+
"step": 500
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 1.39,
|
| 18 |
+
"learning_rate": 9.970000000000001e-05,
|
| 19 |
+
"loss": 1.8593,
|
| 20 |
+
"step": 1000
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 1.39,
|
| 24 |
+
"eval_loss": 1.3255794048309326,
|
| 25 |
+
"eval_runtime": 303.7268,
|
| 26 |
+
"eval_samples_per_second": 4.488,
|
| 27 |
+
"eval_steps_per_second": 0.142,
|
| 28 |
+
"eval_wer": 0.5794877337079866,
|
| 29 |
+
"step": 1000
|
| 30 |
+
},
|
| 31 |
+
{
|
| 32 |
+
"epoch": 2.09,
|
| 33 |
+
"learning_rate": 9.627436281859072e-05,
|
| 34 |
+
"loss": 0.9996,
|
| 35 |
+
"step": 1500
|
| 36 |
+
},
|
| 37 |
+
{
|
| 38 |
+
"epoch": 2.79,
|
| 39 |
+
"learning_rate": 9.253373313343329e-05,
|
| 40 |
+
"loss": 0.7576,
|
| 41 |
+
"step": 2000
|
| 42 |
+
},
|
| 43 |
+
{
|
| 44 |
+
"epoch": 2.79,
|
| 45 |
+
"eval_loss": 1.0275607109069824,
|
| 46 |
+
"eval_runtime": 383.8668,
|
| 47 |
+
"eval_samples_per_second": 3.551,
|
| 48 |
+
"eval_steps_per_second": 0.112,
|
| 49 |
+
"eval_wer": 0.4334810331784286,
|
| 50 |
+
"step": 2000
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 3.49,
|
| 54 |
+
"learning_rate": 8.87856071964018e-05,
|
| 55 |
+
"loss": 0.6464,
|
| 56 |
+
"step": 2500
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 4.18,
|
| 60 |
+
"learning_rate": 8.503748125937032e-05,
|
| 61 |
+
"loss": 0.556,
|
| 62 |
+
"step": 3000
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 4.18,
|
| 66 |
+
"eval_loss": 0.9328228831291199,
|
| 67 |
+
"eval_runtime": 295.659,
|
| 68 |
+
"eval_samples_per_second": 4.61,
|
| 69 |
+
"eval_steps_per_second": 0.145,
|
| 70 |
+
"eval_wer": 0.3811376490507583,
|
| 71 |
+
"step": 3000
|
| 72 |
+
}
|
| 73 |
+
],
|
| 74 |
+
"max_steps": 14340,
|
| 75 |
+
"num_train_epochs": 20,
|
| 76 |
+
"total_flos": 5.481447643746225e+18,
|
| 77 |
+
"trial_name": null,
|
| 78 |
+
"trial_params": null
|
| 79 |
+
}
|
checkpoint-3000/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:99ee4afa7ba3f16a4e5857ce3b6a0ab4232820185e063b7bb3ea94486802c2a2
|
| 3 |
+
size 2927
|
pytorch_model.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 377670039
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f2bded6ea77f057eaa95495f6efc4859e88b57b6a96a722dcaf6477de052c46b
|
| 3 |
size 377670039
|