mikhail-panzo commited on
Commit
2e93ac4
1 Parent(s): d80ca25

Training in progress, step 1000, checkpoint

Browse files
checkpoint-1000/added_tokens.json ADDED
@@ -0,0 +1,4 @@
 
 
 
 
 
1
+ {
2
+ "<ctc_blank>": 80,
3
+ "<mask>": 79
4
+ }
checkpoint-1000/config.json ADDED
@@ -0,0 +1,92 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "mikhail-panzo/zlm_b64_le4_s12000",
3
+ "activation_dropout": 0.1,
4
+ "apply_spec_augment": true,
5
+ "architectures": [
6
+ "SpeechT5ForTextToSpeech"
7
+ ],
8
+ "attention_dropout": 0.1,
9
+ "bos_token_id": 0,
10
+ "conv_bias": false,
11
+ "conv_dim": [
12
+ 512,
13
+ 512,
14
+ 512,
15
+ 512,
16
+ 512,
17
+ 512,
18
+ 512
19
+ ],
20
+ "conv_kernel": [
21
+ 10,
22
+ 3,
23
+ 3,
24
+ 3,
25
+ 3,
26
+ 2,
27
+ 2
28
+ ],
29
+ "conv_stride": [
30
+ 5,
31
+ 2,
32
+ 2,
33
+ 2,
34
+ 2,
35
+ 2,
36
+ 2
37
+ ],
38
+ "decoder_attention_heads": 12,
39
+ "decoder_ffn_dim": 3072,
40
+ "decoder_layerdrop": 0.1,
41
+ "decoder_layers": 6,
42
+ "decoder_start_token_id": 2,
43
+ "encoder_attention_heads": 12,
44
+ "encoder_ffn_dim": 3072,
45
+ "encoder_layerdrop": 0.1,
46
+ "encoder_layers": 12,
47
+ "encoder_max_relative_position": 160,
48
+ "eos_token_id": 2,
49
+ "feat_extract_activation": "gelu",
50
+ "feat_extract_norm": "group",
51
+ "feat_proj_dropout": 0.0,
52
+ "guided_attention_loss_num_heads": 2,
53
+ "guided_attention_loss_scale": 10.0,
54
+ "guided_attention_loss_sigma": 0.4,
55
+ "hidden_act": "gelu",
56
+ "hidden_dropout": 0.1,
57
+ "hidden_size": 768,
58
+ "initializer_range": 0.02,
59
+ "is_encoder_decoder": true,
60
+ "layer_norm_eps": 1e-05,
61
+ "mask_feature_length": 10,
62
+ "mask_feature_min_masks": 0,
63
+ "mask_feature_prob": 0.0,
64
+ "mask_time_length": 10,
65
+ "mask_time_min_masks": 2,
66
+ "mask_time_prob": 0.05,
67
+ "max_length": 1876,
68
+ "max_speech_positions": 1876,
69
+ "max_text_positions": 600,
70
+ "model_type": "speecht5",
71
+ "num_conv_pos_embedding_groups": 16,
72
+ "num_conv_pos_embeddings": 128,
73
+ "num_feat_extract_layers": 7,
74
+ "num_mel_bins": 80,
75
+ "pad_token_id": 1,
76
+ "positional_dropout": 0.1,
77
+ "reduction_factor": 2,
78
+ "scale_embedding": false,
79
+ "speaker_embedding_dim": 512,
80
+ "speech_decoder_postnet_dropout": 0.5,
81
+ "speech_decoder_postnet_kernel": 5,
82
+ "speech_decoder_postnet_layers": 5,
83
+ "speech_decoder_postnet_units": 256,
84
+ "speech_decoder_prenet_dropout": 0.5,
85
+ "speech_decoder_prenet_layers": 2,
86
+ "speech_decoder_prenet_units": 256,
87
+ "torch_dtype": "float32",
88
+ "transformers_version": "4.41.0.dev0",
89
+ "use_cache": false,
90
+ "use_guided_attention_loss": true,
91
+ "vocab_size": 81
92
+ }
checkpoint-1000/generation_config.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 0,
4
+ "decoder_start_token_id": 2,
5
+ "eos_token_id": 2,
6
+ "max_length": 1876,
7
+ "pad_token_id": 1,
8
+ "transformers_version": "4.41.0.dev0"
9
+ }
checkpoint-1000/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c5b760fdc3f45bdd8e928ab49e2a52d2fcc9affd98085f0059f2464204baeb20
3
+ size 577789320
checkpoint-1000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a59688858dd9375730df3eb2faac33a8d53a58edab121390bd4aa60cf6097be5
3
+ size 1155772233
checkpoint-1000/preprocessor_config.json ADDED
@@ -0,0 +1,19 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "do_normalize": false,
3
+ "feature_extractor_type": "SpeechT5FeatureExtractor",
4
+ "feature_size": 1,
5
+ "fmax": 7600,
6
+ "fmin": 80,
7
+ "frame_signal_scale": 1.0,
8
+ "hop_length": 16,
9
+ "mel_floor": 1e-10,
10
+ "num_mel_bins": 80,
11
+ "padding_side": "right",
12
+ "padding_value": 0.0,
13
+ "processor_class": "SpeechT5Processor",
14
+ "reduction_factor": 2,
15
+ "return_attention_mask": true,
16
+ "sampling_rate": 16000,
17
+ "win_function": "hann_window",
18
+ "win_length": 64
19
+ }
checkpoint-1000/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f3f52d7cae16c0621ee61e9c1e33fb9bcbf430a70031aecae43c8c3e688099d1
3
+ size 14244
checkpoint-1000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:45070bc91d144eb7f163f351456cac0db81ee5e61490751953caec092c3e6eb7
3
+ size 1064
checkpoint-1000/special_tokens_map.json ADDED
@@ -0,0 +1,37 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<s>",
4
+ "lstrip": false,
5
+ "normalized": false,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "</s>",
11
+ "lstrip": false,
12
+ "normalized": false,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "mask_token": {
17
+ "content": "<mask>",
18
+ "lstrip": true,
19
+ "normalized": true,
20
+ "rstrip": false,
21
+ "single_word": false
22
+ },
23
+ "pad_token": {
24
+ "content": "<pad>",
25
+ "lstrip": false,
26
+ "normalized": false,
27
+ "rstrip": false,
28
+ "single_word": false
29
+ },
30
+ "unk_token": {
31
+ "content": "<unk>",
32
+ "lstrip": false,
33
+ "normalized": false,
34
+ "rstrip": false,
35
+ "single_word": false
36
+ }
37
+ }
checkpoint-1000/spm_char.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7fcc48f3e225f627b1641db410ceb0c8649bd2b0c982e150b03f8be3728ab560
3
+ size 238473
checkpoint-1000/tokenizer_config.json ADDED
@@ -0,0 +1,63 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "added_tokens_decoder": {
3
+ "0": {
4
+ "content": "<s>",
5
+ "lstrip": false,
6
+ "normalized": false,
7
+ "rstrip": false,
8
+ "single_word": false,
9
+ "special": true
10
+ },
11
+ "1": {
12
+ "content": "<pad>",
13
+ "lstrip": false,
14
+ "normalized": false,
15
+ "rstrip": false,
16
+ "single_word": false,
17
+ "special": true
18
+ },
19
+ "2": {
20
+ "content": "</s>",
21
+ "lstrip": false,
22
+ "normalized": false,
23
+ "rstrip": false,
24
+ "single_word": false,
25
+ "special": true
26
+ },
27
+ "3": {
28
+ "content": "<unk>",
29
+ "lstrip": false,
30
+ "normalized": false,
31
+ "rstrip": false,
32
+ "single_word": false,
33
+ "special": true
34
+ },
35
+ "79": {
36
+ "content": "<mask>",
37
+ "lstrip": true,
38
+ "normalized": true,
39
+ "rstrip": false,
40
+ "single_word": false,
41
+ "special": true
42
+ },
43
+ "80": {
44
+ "content": "<ctc_blank>",
45
+ "lstrip": false,
46
+ "normalized": true,
47
+ "rstrip": false,
48
+ "single_word": false,
49
+ "special": false
50
+ }
51
+ },
52
+ "bos_token": "<s>",
53
+ "clean_up_tokenization_spaces": true,
54
+ "eos_token": "</s>",
55
+ "mask_token": "<mask>",
56
+ "model_max_length": 600,
57
+ "normalize": false,
58
+ "pad_token": "<pad>",
59
+ "processor_class": "SpeechT5Processor",
60
+ "sp_model_kwargs": {},
61
+ "tokenizer_class": "SpeechT5Tokenizer",
62
+ "unk_token": "<unk>"
63
+ }
checkpoint-1000/trainer_state.json ADDED
@@ -0,0 +1,189 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 0.4557439386844635,
3
+ "best_model_checkpoint": "mikhail-panzo/zlm-fil_b64_le5_s8000/checkpoint-1000",
4
+ "epoch": 44.44444444444444,
5
+ "eval_steps": 500,
6
+ "global_step": 1000,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 2.2222222222222223,
13
+ "grad_norm": 4.803515434265137,
14
+ "learning_rate": 2.5000000000000004e-07,
15
+ "loss": 0.8968,
16
+ "step": 50
17
+ },
18
+ {
19
+ "epoch": 4.444444444444445,
20
+ "grad_norm": 2.6444506645202637,
21
+ "learning_rate": 5.000000000000001e-07,
22
+ "loss": 0.8539,
23
+ "step": 100
24
+ },
25
+ {
26
+ "epoch": 6.666666666666667,
27
+ "grad_norm": 1.9820537567138672,
28
+ "learning_rate": 7.5e-07,
29
+ "loss": 0.7856,
30
+ "step": 150
31
+ },
32
+ {
33
+ "epoch": 8.88888888888889,
34
+ "grad_norm": 1.2031773328781128,
35
+ "learning_rate": 1.0000000000000002e-06,
36
+ "loss": 0.7306,
37
+ "step": 200
38
+ },
39
+ {
40
+ "epoch": 11.11111111111111,
41
+ "grad_norm": 1.728398084640503,
42
+ "learning_rate": 1.25e-06,
43
+ "loss": 0.6821,
44
+ "step": 250
45
+ },
46
+ {
47
+ "epoch": 13.333333333333334,
48
+ "grad_norm": 1.232366681098938,
49
+ "learning_rate": 1.5e-06,
50
+ "loss": 0.6348,
51
+ "step": 300
52
+ },
53
+ {
54
+ "epoch": 15.555555555555555,
55
+ "grad_norm": 1.4039260149002075,
56
+ "learning_rate": 1.75e-06,
57
+ "loss": 0.5962,
58
+ "step": 350
59
+ },
60
+ {
61
+ "epoch": 17.77777777777778,
62
+ "grad_norm": 1.1848849058151245,
63
+ "learning_rate": 2.0000000000000003e-06,
64
+ "loss": 0.5758,
65
+ "step": 400
66
+ },
67
+ {
68
+ "epoch": 20.0,
69
+ "grad_norm": 0.9365352392196655,
70
+ "learning_rate": 2.25e-06,
71
+ "loss": 0.5636,
72
+ "step": 450
73
+ },
74
+ {
75
+ "epoch": 22.22222222222222,
76
+ "grad_norm": 0.9893051981925964,
77
+ "learning_rate": 2.5e-06,
78
+ "loss": 0.5529,
79
+ "step": 500
80
+ },
81
+ {
82
+ "epoch": 22.22222222222222,
83
+ "eval_loss": 0.4999743402004242,
84
+ "eval_runtime": 6.8701,
85
+ "eval_samples_per_second": 23.144,
86
+ "eval_steps_per_second": 2.911,
87
+ "step": 500
88
+ },
89
+ {
90
+ "epoch": 24.444444444444443,
91
+ "grad_norm": 0.7535696029663086,
92
+ "learning_rate": 2.7500000000000004e-06,
93
+ "loss": 0.5476,
94
+ "step": 550
95
+ },
96
+ {
97
+ "epoch": 26.666666666666668,
98
+ "grad_norm": 0.7487667202949524,
99
+ "learning_rate": 3e-06,
100
+ "loss": 0.536,
101
+ "step": 600
102
+ },
103
+ {
104
+ "epoch": 28.88888888888889,
105
+ "grad_norm": 0.9480589628219604,
106
+ "learning_rate": 3.2500000000000002e-06,
107
+ "loss": 0.5271,
108
+ "step": 650
109
+ },
110
+ {
111
+ "epoch": 31.11111111111111,
112
+ "grad_norm": 1.0027987957000732,
113
+ "learning_rate": 3.5e-06,
114
+ "loss": 0.5202,
115
+ "step": 700
116
+ },
117
+ {
118
+ "epoch": 33.333333333333336,
119
+ "grad_norm": 0.7353075742721558,
120
+ "learning_rate": 3.7500000000000005e-06,
121
+ "loss": 0.5128,
122
+ "step": 750
123
+ },
124
+ {
125
+ "epoch": 35.55555555555556,
126
+ "grad_norm": 1.2223567962646484,
127
+ "learning_rate": 4.000000000000001e-06,
128
+ "loss": 0.5159,
129
+ "step": 800
130
+ },
131
+ {
132
+ "epoch": 37.77777777777778,
133
+ "grad_norm": 0.835096001625061,
134
+ "learning_rate": 4.25e-06,
135
+ "loss": 0.5051,
136
+ "step": 850
137
+ },
138
+ {
139
+ "epoch": 40.0,
140
+ "grad_norm": 0.8878389000892639,
141
+ "learning_rate": 4.5e-06,
142
+ "loss": 0.5044,
143
+ "step": 900
144
+ },
145
+ {
146
+ "epoch": 42.22222222222222,
147
+ "grad_norm": 0.8464574813842773,
148
+ "learning_rate": 4.75e-06,
149
+ "loss": 0.4984,
150
+ "step": 950
151
+ },
152
+ {
153
+ "epoch": 44.44444444444444,
154
+ "grad_norm": 0.8119956254959106,
155
+ "learning_rate": 5e-06,
156
+ "loss": 0.4974,
157
+ "step": 1000
158
+ },
159
+ {
160
+ "epoch": 44.44444444444444,
161
+ "eval_loss": 0.4557439386844635,
162
+ "eval_runtime": 6.6345,
163
+ "eval_samples_per_second": 23.965,
164
+ "eval_steps_per_second": 3.015,
165
+ "step": 1000
166
+ }
167
+ ],
168
+ "logging_steps": 50,
169
+ "max_steps": 8000,
170
+ "num_input_tokens_seen": 0,
171
+ "num_train_epochs": 364,
172
+ "save_steps": 500,
173
+ "stateful_callbacks": {
174
+ "TrainerControl": {
175
+ "args": {
176
+ "should_epoch_stop": false,
177
+ "should_evaluate": false,
178
+ "should_log": false,
179
+ "should_save": true,
180
+ "should_training_stop": false
181
+ },
182
+ "attributes": {}
183
+ }
184
+ },
185
+ "total_flos": 1.4552547496115904e+16,
186
+ "train_batch_size": 32,
187
+ "trial_name": null,
188
+ "trial_params": null
189
+ }
checkpoint-1000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ad4496fb86a719121d50f59f5a3033b349102dc8db9bac2fc6b307af92943070
3
+ size 5304