Commit
•
fd5771e
1
Parent(s):
8506a48
remove whispering
Browse files- .gitattributes +0 -50
- .gitignore +1 -0
- checkpoint-144-epoch-3/config.json +276 -0
- checkpoint-144-epoch-3/generation_config.json +12 -0
- checkpoint-216-epoch-5/config.json +276 -0
- checkpoint-216-epoch-5/generation_config.json +12 -0
- checkpoint-288-epoch-7/config.json +276 -0
- checkpoint-288-epoch-7/generation_config.json +12 -0
- checkpoint-312-epoch-7/config.json +276 -0
- checkpoint-312-epoch-7/generation_config.json +12 -0
- checkpoint-72-epoch-1/config.json +276 -0
- checkpoint-72-epoch-1/generation_config.json +12 -0
- run.sh +2 -1
- run_parler_tts_training.py +11 -0
.gitattributes
CHANGED
@@ -33,53 +33,3 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
|
|
33 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
36 |
-
wandb/run-20240513_205708-e3sq5zz5/files/media/audio/Speech[[:space:]]samples/eval_117_3f7a73b6eb939439853b.wav filter=lfs diff=lfs merge=lfs -text
|
37 |
-
wandb/run-20240513_205708-e3sq5zz5/files/media/audio/Speech[[:space:]]samples/eval_117_51c4feb7b25049501e37.wav filter=lfs diff=lfs merge=lfs -text
|
38 |
-
wandb/run-20240513_205708-e3sq5zz5/files/media/audio/Speech[[:space:]]samples/eval_117_6d818be44006fe311b25.wav filter=lfs diff=lfs merge=lfs -text
|
39 |
-
wandb/run-20240513_205708-e3sq5zz5/files/media/audio/Speech[[:space:]]samples/eval_117_7633a61e6ee80a024fb2.wav filter=lfs diff=lfs merge=lfs -text
|
40 |
-
wandb/run-20240513_205708-e3sq5zz5/files/media/audio/Speech[[:space:]]samples/eval_117_855fad46eee4554f15a7.wav filter=lfs diff=lfs merge=lfs -text
|
41 |
-
wandb/run-20240513_205708-e3sq5zz5/files/media/audio/Speech[[:space:]]samples/eval_117_dd8bdcb385062c83d9db.wav filter=lfs diff=lfs merge=lfs -text
|
42 |
-
wandb/run-20240513_205708-e3sq5zz5/files/media/audio/Speech[[:space:]]samples/eval_117_ea76ce7bbfab47610005.wav filter=lfs diff=lfs merge=lfs -text
|
43 |
-
wandb/run-20240513_205708-e3sq5zz5/files/media/audio/Speech[[:space:]]samples/eval_195_2abe63a219dd9980a311.wav filter=lfs diff=lfs merge=lfs -text
|
44 |
-
wandb/run-20240513_205708-e3sq5zz5/files/media/audio/Speech[[:space:]]samples/eval_195_69610571800405c8afc4.wav filter=lfs diff=lfs merge=lfs -text
|
45 |
-
wandb/run-20240513_205708-e3sq5zz5/files/media/audio/Speech[[:space:]]samples/eval_195_6cac4adea8fad0e75539.wav filter=lfs diff=lfs merge=lfs -text
|
46 |
-
wandb/run-20240513_205708-e3sq5zz5/files/media/audio/Speech[[:space:]]samples/eval_195_830d908cf01e69e53001.wav filter=lfs diff=lfs merge=lfs -text
|
47 |
-
wandb/run-20240513_205708-e3sq5zz5/files/media/audio/Speech[[:space:]]samples/eval_195_83726d791dfb671cadab.wav filter=lfs diff=lfs merge=lfs -text
|
48 |
-
wandb/run-20240513_205708-e3sq5zz5/files/media/audio/Speech[[:space:]]samples/eval_195_c0a3f3d7cd9b5e6f1ff3.wav filter=lfs diff=lfs merge=lfs -text
|
49 |
-
wandb/run-20240513_205708-e3sq5zz5/files/media/audio/Speech[[:space:]]samples/eval_195_c762fd4cdeb8dd603c93.wav filter=lfs diff=lfs merge=lfs -text
|
50 |
-
wandb/run-20240513_205708-e3sq5zz5/files/media/audio/Speech[[:space:]]samples/eval_234_07ddb3b15d19fc0d28b8.wav filter=lfs diff=lfs merge=lfs -text
|
51 |
-
wandb/run-20240513_205708-e3sq5zz5/files/media/audio/Speech[[:space:]]samples/eval_234_081798576741dcab4ed7.wav filter=lfs diff=lfs merge=lfs -text
|
52 |
-
wandb/run-20240513_205708-e3sq5zz5/files/media/audio/Speech[[:space:]]samples/eval_234_1c742d222224ecad309d.wav filter=lfs diff=lfs merge=lfs -text
|
53 |
-
wandb/run-20240513_205708-e3sq5zz5/files/media/audio/Speech[[:space:]]samples/eval_234_823c293a773a7747aae9.wav filter=lfs diff=lfs merge=lfs -text
|
54 |
-
wandb/run-20240513_205708-e3sq5zz5/files/media/audio/Speech[[:space:]]samples/eval_234_b9550b7eaa369565c70f.wav filter=lfs diff=lfs merge=lfs -text
|
55 |
-
wandb/run-20240513_205708-e3sq5zz5/files/media/audio/Speech[[:space:]]samples/eval_234_c4854e1ae7d92c3a58d5.wav filter=lfs diff=lfs merge=lfs -text
|
56 |
-
wandb/run-20240513_205708-e3sq5zz5/files/media/audio/Speech[[:space:]]samples/eval_234_e8462d5ee69db7355157.wav filter=lfs diff=lfs merge=lfs -text
|
57 |
-
wandb/run-20240513_205708-e3sq5zz5/files/media/audio/Speech[[:space:]]samples/eval_273_1b11908a3b16b03f6f28.wav filter=lfs diff=lfs merge=lfs -text
|
58 |
-
wandb/run-20240513_205708-e3sq5zz5/files/media/audio/Speech[[:space:]]samples/eval_273_623fa20cd98668542c5a.wav filter=lfs diff=lfs merge=lfs -text
|
59 |
-
wandb/run-20240513_205708-e3sq5zz5/files/media/audio/Speech[[:space:]]samples/eval_273_7e85e65244c43352d3ff.wav filter=lfs diff=lfs merge=lfs -text
|
60 |
-
wandb/run-20240513_205708-e3sq5zz5/files/media/audio/Speech[[:space:]]samples/eval_273_85b843d1a8a6ba22faee.wav filter=lfs diff=lfs merge=lfs -text
|
61 |
-
wandb/run-20240513_205708-e3sq5zz5/files/media/audio/Speech[[:space:]]samples/eval_273_89094caf461fb7b67ec7.wav filter=lfs diff=lfs merge=lfs -text
|
62 |
-
wandb/run-20240513_205708-e3sq5zz5/files/media/audio/Speech[[:space:]]samples/eval_273_8f6478a9918ad2b1ae55.wav filter=lfs diff=lfs merge=lfs -text
|
63 |
-
wandb/run-20240513_205708-e3sq5zz5/files/media/audio/Speech[[:space:]]samples/eval_273_9f4846c46a52d5e23c8d.wav filter=lfs diff=lfs merge=lfs -text
|
64 |
-
wandb/run-20240513_205708-e3sq5zz5/files/media/audio/Speech[[:space:]]samples/eval_312_01c26787b801562a3789.wav filter=lfs diff=lfs merge=lfs -text
|
65 |
-
wandb/run-20240513_205708-e3sq5zz5/files/media/audio/Speech[[:space:]]samples/eval_312_0ceabcd2840ac8898fe1.wav filter=lfs diff=lfs merge=lfs -text
|
66 |
-
wandb/run-20240513_205708-e3sq5zz5/files/media/audio/Speech[[:space:]]samples/eval_312_337071167ae9bad07b71.wav filter=lfs diff=lfs merge=lfs -text
|
67 |
-
wandb/run-20240513_205708-e3sq5zz5/files/media/audio/Speech[[:space:]]samples/eval_312_43fb61bde32a4b6742d4.wav filter=lfs diff=lfs merge=lfs -text
|
68 |
-
wandb/run-20240513_205708-e3sq5zz5/files/media/audio/Speech[[:space:]]samples/eval_312_47c9ee8dc54086885fe9.wav filter=lfs diff=lfs merge=lfs -text
|
69 |
-
wandb/run-20240513_205708-e3sq5zz5/files/media/audio/Speech[[:space:]]samples/eval_312_ac7b82097e1e55d1368c.wav filter=lfs diff=lfs merge=lfs -text
|
70 |
-
wandb/run-20240513_205708-e3sq5zz5/files/media/audio/Speech[[:space:]]samples/eval_312_f460b32c3554b8680b92.wav filter=lfs diff=lfs merge=lfs -text
|
71 |
-
wandb/run-20240513_205708-e3sq5zz5/files/media/audio/Speech[[:space:]]samples/eval_39_6481b4598446a5223728.wav filter=lfs diff=lfs merge=lfs -text
|
72 |
-
wandb/run-20240513_205708-e3sq5zz5/files/media/audio/Speech[[:space:]]samples/eval_39_65c5863263ecc0063e7c.wav filter=lfs diff=lfs merge=lfs -text
|
73 |
-
wandb/run-20240513_205708-e3sq5zz5/files/media/audio/Speech[[:space:]]samples/eval_39_75f77b4060ad41f6cd28.wav filter=lfs diff=lfs merge=lfs -text
|
74 |
-
wandb/run-20240513_205708-e3sq5zz5/files/media/audio/Speech[[:space:]]samples/eval_39_91ec273933f07274531b.wav filter=lfs diff=lfs merge=lfs -text
|
75 |
-
wandb/run-20240513_205708-e3sq5zz5/files/media/audio/Speech[[:space:]]samples/eval_39_a92c59803d84d120f9c8.wav filter=lfs diff=lfs merge=lfs -text
|
76 |
-
wandb/run-20240513_205708-e3sq5zz5/files/media/audio/Speech[[:space:]]samples/eval_39_e334e92b9d2007210460.wav filter=lfs diff=lfs merge=lfs -text
|
77 |
-
wandb/run-20240513_205708-e3sq5zz5/files/media/audio/Speech[[:space:]]samples/eval_39_e93fbf8d0314875d9a8a.wav filter=lfs diff=lfs merge=lfs -text
|
78 |
-
wandb/run-20240513_205708-e3sq5zz5/files/media/audio/Speech[[:space:]]samples/eval_78_2a1aa9c4aa9b28aeff3d.wav filter=lfs diff=lfs merge=lfs -text
|
79 |
-
wandb/run-20240513_205708-e3sq5zz5/files/media/audio/Speech[[:space:]]samples/eval_78_31d4a3506990aa644012.wav filter=lfs diff=lfs merge=lfs -text
|
80 |
-
wandb/run-20240513_205708-e3sq5zz5/files/media/audio/Speech[[:space:]]samples/eval_78_6f247990e446903b7c69.wav filter=lfs diff=lfs merge=lfs -text
|
81 |
-
wandb/run-20240513_205708-e3sq5zz5/files/media/audio/Speech[[:space:]]samples/eval_78_ccbbfb2e4ad764ed2d56.wav filter=lfs diff=lfs merge=lfs -text
|
82 |
-
wandb/run-20240513_205708-e3sq5zz5/files/media/audio/Speech[[:space:]]samples/eval_78_e0a9392514f4f0b15b68.wav filter=lfs diff=lfs merge=lfs -text
|
83 |
-
wandb/run-20240513_205708-e3sq5zz5/files/media/audio/Speech[[:space:]]samples/eval_78_e8936f475da43789ec82.wav filter=lfs diff=lfs merge=lfs -text
|
84 |
-
wandb/run-20240513_205708-e3sq5zz5/files/media/audio/Speech[[:space:]]samples/eval_78_ff389c0afad3d276441d.wav filter=lfs diff=lfs merge=lfs -text
|
85 |
-
wandb/run-20240513_205708-e3sq5zz5/run-e3sq5zz5.wandb filter=lfs diff=lfs merge=lfs -text
|
|
|
33 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
.gitignore
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
wandb
|
checkpoint-144-epoch-3/config.json
ADDED
@@ -0,0 +1,276 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_name_or_path": "/fsx/yoach/tmp/artefacts/training-400M-punctuated-v2/",
|
3 |
+
"architectures": [
|
4 |
+
"ParlerTTSForConditionalGeneration"
|
5 |
+
],
|
6 |
+
"audio_encoder": {
|
7 |
+
"_name_or_path": "ylacombe/dac_44khZ_8kbps",
|
8 |
+
"add_cross_attention": false,
|
9 |
+
"architectures": [
|
10 |
+
"DACModel"
|
11 |
+
],
|
12 |
+
"bad_words_ids": null,
|
13 |
+
"begin_suppress_tokens": null,
|
14 |
+
"bos_token_id": null,
|
15 |
+
"chunk_size_feed_forward": 0,
|
16 |
+
"codebook_size": 1024,
|
17 |
+
"cross_attention_hidden_size": null,
|
18 |
+
"decoder_start_token_id": null,
|
19 |
+
"diversity_penalty": 0.0,
|
20 |
+
"do_sample": false,
|
21 |
+
"early_stopping": false,
|
22 |
+
"encoder_no_repeat_ngram_size": 0,
|
23 |
+
"eos_token_id": null,
|
24 |
+
"exponential_decay_length_penalty": null,
|
25 |
+
"finetuning_task": null,
|
26 |
+
"forced_bos_token_id": null,
|
27 |
+
"forced_eos_token_id": null,
|
28 |
+
"frame_rate": 86,
|
29 |
+
"id2label": {
|
30 |
+
"0": "LABEL_0",
|
31 |
+
"1": "LABEL_1"
|
32 |
+
},
|
33 |
+
"is_decoder": false,
|
34 |
+
"is_encoder_decoder": false,
|
35 |
+
"label2id": {
|
36 |
+
"LABEL_0": 0,
|
37 |
+
"LABEL_1": 1
|
38 |
+
},
|
39 |
+
"latent_dim": 1024,
|
40 |
+
"length_penalty": 1.0,
|
41 |
+
"max_length": 20,
|
42 |
+
"min_length": 0,
|
43 |
+
"model_bitrate": 8,
|
44 |
+
"model_type": "dac",
|
45 |
+
"no_repeat_ngram_size": 0,
|
46 |
+
"num_beam_groups": 1,
|
47 |
+
"num_beams": 1,
|
48 |
+
"num_codebooks": 9,
|
49 |
+
"num_return_sequences": 1,
|
50 |
+
"output_attentions": false,
|
51 |
+
"output_hidden_states": false,
|
52 |
+
"output_scores": false,
|
53 |
+
"pad_token_id": null,
|
54 |
+
"prefix": null,
|
55 |
+
"problem_type": null,
|
56 |
+
"pruned_heads": {},
|
57 |
+
"remove_invalid_values": false,
|
58 |
+
"repetition_penalty": 1.0,
|
59 |
+
"return_dict": true,
|
60 |
+
"return_dict_in_generate": false,
|
61 |
+
"sampling_rate": 44100,
|
62 |
+
"sep_token_id": null,
|
63 |
+
"suppress_tokens": null,
|
64 |
+
"task_specific_params": null,
|
65 |
+
"temperature": 1.0,
|
66 |
+
"tf_legacy_loss": false,
|
67 |
+
"tie_encoder_decoder": false,
|
68 |
+
"tie_word_embeddings": true,
|
69 |
+
"tokenizer_class": null,
|
70 |
+
"top_k": 50,
|
71 |
+
"top_p": 1.0,
|
72 |
+
"torch_dtype": "float32",
|
73 |
+
"torchscript": false,
|
74 |
+
"typical_p": 1.0,
|
75 |
+
"use_bfloat16": false
|
76 |
+
},
|
77 |
+
"decoder": {
|
78 |
+
"_name_or_path": "/fsx/yoach/tmp/artefacts/decoder_400M/",
|
79 |
+
"activation_dropout": 0.0,
|
80 |
+
"activation_function": "gelu",
|
81 |
+
"add_cross_attention": true,
|
82 |
+
"architectures": [
|
83 |
+
"ParlerTTSForCausalLM"
|
84 |
+
],
|
85 |
+
"attention_dropout": 0.0,
|
86 |
+
"bad_words_ids": null,
|
87 |
+
"begin_suppress_tokens": null,
|
88 |
+
"bos_token_id": 1025,
|
89 |
+
"chunk_size_feed_forward": 0,
|
90 |
+
"cross_attention_hidden_size": null,
|
91 |
+
"decoder_start_token_id": null,
|
92 |
+
"diversity_penalty": 0.0,
|
93 |
+
"do_sample": false,
|
94 |
+
"dropout": 0.1,
|
95 |
+
"early_stopping": false,
|
96 |
+
"encoder_no_repeat_ngram_size": 0,
|
97 |
+
"eos_token_id": 1024,
|
98 |
+
"exponential_decay_length_penalty": null,
|
99 |
+
"ffn_dim": 4096,
|
100 |
+
"finetuning_task": null,
|
101 |
+
"forced_bos_token_id": null,
|
102 |
+
"forced_eos_token_id": null,
|
103 |
+
"hidden_size": 1024,
|
104 |
+
"id2label": {
|
105 |
+
"0": "LABEL_0",
|
106 |
+
"1": "LABEL_1"
|
107 |
+
},
|
108 |
+
"initializer_factor": 0.02,
|
109 |
+
"is_decoder": true,
|
110 |
+
"is_encoder_decoder": false,
|
111 |
+
"label2id": {
|
112 |
+
"LABEL_0": 0,
|
113 |
+
"LABEL_1": 1
|
114 |
+
},
|
115 |
+
"layerdrop": 0.0,
|
116 |
+
"length_penalty": 1.0,
|
117 |
+
"max_length": 20,
|
118 |
+
"max_position_embeddings": 4096,
|
119 |
+
"min_length": 0,
|
120 |
+
"model_type": "parler_tts_decoder",
|
121 |
+
"no_repeat_ngram_size": 0,
|
122 |
+
"num_attention_heads": 16,
|
123 |
+
"num_beam_groups": 1,
|
124 |
+
"num_beams": 1,
|
125 |
+
"num_codebooks": 9,
|
126 |
+
"num_hidden_layers": 24,
|
127 |
+
"num_return_sequences": 1,
|
128 |
+
"output_attentions": false,
|
129 |
+
"output_hidden_states": false,
|
130 |
+
"output_scores": false,
|
131 |
+
"pad_token_id": 1024,
|
132 |
+
"prefix": null,
|
133 |
+
"problem_type": null,
|
134 |
+
"pruned_heads": {},
|
135 |
+
"remove_invalid_values": false,
|
136 |
+
"repetition_penalty": 1.0,
|
137 |
+
"return_dict": true,
|
138 |
+
"return_dict_in_generate": false,
|
139 |
+
"scale_embedding": false,
|
140 |
+
"sep_token_id": null,
|
141 |
+
"suppress_tokens": null,
|
142 |
+
"task_specific_params": null,
|
143 |
+
"temperature": 1.0,
|
144 |
+
"tf_legacy_loss": false,
|
145 |
+
"tie_encoder_decoder": false,
|
146 |
+
"tie_word_embeddings": false,
|
147 |
+
"tokenizer_class": null,
|
148 |
+
"top_k": 50,
|
149 |
+
"top_p": 1.0,
|
150 |
+
"torch_dtype": "float32",
|
151 |
+
"torchscript": false,
|
152 |
+
"typical_p": 1.0,
|
153 |
+
"use_bfloat16": false,
|
154 |
+
"use_cache": true,
|
155 |
+
"vocab_size": 1088
|
156 |
+
},
|
157 |
+
"decoder_start_token_id": 1025,
|
158 |
+
"is_encoder_decoder": true,
|
159 |
+
"model_type": "parler_tts",
|
160 |
+
"pad_token_id": 1024,
|
161 |
+
"text_encoder": {
|
162 |
+
"_name_or_path": "google/flan-t5-base",
|
163 |
+
"add_cross_attention": false,
|
164 |
+
"architectures": [
|
165 |
+
"T5ForConditionalGeneration"
|
166 |
+
],
|
167 |
+
"bad_words_ids": null,
|
168 |
+
"begin_suppress_tokens": null,
|
169 |
+
"bos_token_id": null,
|
170 |
+
"chunk_size_feed_forward": 0,
|
171 |
+
"classifier_dropout": 0.0,
|
172 |
+
"cross_attention_hidden_size": null,
|
173 |
+
"d_ff": 2048,
|
174 |
+
"d_kv": 64,
|
175 |
+
"d_model": 768,
|
176 |
+
"decoder_start_token_id": 0,
|
177 |
+
"dense_act_fn": "gelu_new",
|
178 |
+
"diversity_penalty": 0.0,
|
179 |
+
"do_sample": false,
|
180 |
+
"dropout_rate": 0.1,
|
181 |
+
"early_stopping": false,
|
182 |
+
"encoder_no_repeat_ngram_size": 0,
|
183 |
+
"eos_token_id": 1,
|
184 |
+
"exponential_decay_length_penalty": null,
|
185 |
+
"feed_forward_proj": "gated-gelu",
|
186 |
+
"finetuning_task": null,
|
187 |
+
"forced_bos_token_id": null,
|
188 |
+
"forced_eos_token_id": null,
|
189 |
+
"id2label": {
|
190 |
+
"0": "LABEL_0",
|
191 |
+
"1": "LABEL_1"
|
192 |
+
},
|
193 |
+
"initializer_factor": 1.0,
|
194 |
+
"is_decoder": false,
|
195 |
+
"is_encoder_decoder": true,
|
196 |
+
"is_gated_act": true,
|
197 |
+
"label2id": {
|
198 |
+
"LABEL_0": 0,
|
199 |
+
"LABEL_1": 1
|
200 |
+
},
|
201 |
+
"layer_norm_epsilon": 1e-06,
|
202 |
+
"length_penalty": 1.0,
|
203 |
+
"max_length": 20,
|
204 |
+
"min_length": 0,
|
205 |
+
"model_type": "t5",
|
206 |
+
"n_positions": 512,
|
207 |
+
"no_repeat_ngram_size": 0,
|
208 |
+
"num_beam_groups": 1,
|
209 |
+
"num_beams": 1,
|
210 |
+
"num_decoder_layers": 12,
|
211 |
+
"num_heads": 12,
|
212 |
+
"num_layers": 12,
|
213 |
+
"num_return_sequences": 1,
|
214 |
+
"output_attentions": false,
|
215 |
+
"output_hidden_states": false,
|
216 |
+
"output_past": true,
|
217 |
+
"output_scores": false,
|
218 |
+
"pad_token_id": 0,
|
219 |
+
"prefix": null,
|
220 |
+
"problem_type": null,
|
221 |
+
"pruned_heads": {},
|
222 |
+
"relative_attention_max_distance": 128,
|
223 |
+
"relative_attention_num_buckets": 32,
|
224 |
+
"remove_invalid_values": false,
|
225 |
+
"repetition_penalty": 1.0,
|
226 |
+
"return_dict": true,
|
227 |
+
"return_dict_in_generate": false,
|
228 |
+
"sep_token_id": null,
|
229 |
+
"suppress_tokens": null,
|
230 |
+
"task_specific_params": {
|
231 |
+
"summarization": {
|
232 |
+
"early_stopping": true,
|
233 |
+
"length_penalty": 2.0,
|
234 |
+
"max_length": 200,
|
235 |
+
"min_length": 30,
|
236 |
+
"no_repeat_ngram_size": 3,
|
237 |
+
"num_beams": 4,
|
238 |
+
"prefix": "summarize: "
|
239 |
+
},
|
240 |
+
"translation_en_to_de": {
|
241 |
+
"early_stopping": true,
|
242 |
+
"max_length": 300,
|
243 |
+
"num_beams": 4,
|
244 |
+
"prefix": "translate English to German: "
|
245 |
+
},
|
246 |
+
"translation_en_to_fr": {
|
247 |
+
"early_stopping": true,
|
248 |
+
"max_length": 300,
|
249 |
+
"num_beams": 4,
|
250 |
+
"prefix": "translate English to French: "
|
251 |
+
},
|
252 |
+
"translation_en_to_ro": {
|
253 |
+
"early_stopping": true,
|
254 |
+
"max_length": 300,
|
255 |
+
"num_beams": 4,
|
256 |
+
"prefix": "translate English to Romanian: "
|
257 |
+
}
|
258 |
+
},
|
259 |
+
"temperature": 1.0,
|
260 |
+
"tf_legacy_loss": false,
|
261 |
+
"tie_encoder_decoder": false,
|
262 |
+
"tie_word_embeddings": false,
|
263 |
+
"tokenizer_class": null,
|
264 |
+
"top_k": 50,
|
265 |
+
"top_p": 1.0,
|
266 |
+
"torch_dtype": null,
|
267 |
+
"torchscript": false,
|
268 |
+
"typical_p": 1.0,
|
269 |
+
"use_bfloat16": false,
|
270 |
+
"use_cache": true,
|
271 |
+
"vocab_size": 32128
|
272 |
+
},
|
273 |
+
"torch_dtype": "float32",
|
274 |
+
"transformers_version": "4.41.0.dev0",
|
275 |
+
"vocab_size": 32128
|
276 |
+
}
|
checkpoint-144-epoch-3/generation_config.json
ADDED
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_from_model_config": true,
|
3 |
+
"bos_token_id": 1025,
|
4 |
+
"decoder_start_token_id": 1025,
|
5 |
+
"do_sample": true,
|
6 |
+
"eos_token_id": 1024,
|
7 |
+
"guidance_scale": 1.0,
|
8 |
+
"max_length": 2580,
|
9 |
+
"min_new_tokens": 50,
|
10 |
+
"pad_token_id": 1024,
|
11 |
+
"transformers_version": "4.41.0.dev0"
|
12 |
+
}
|
checkpoint-216-epoch-5/config.json
ADDED
@@ -0,0 +1,276 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_name_or_path": "/fsx/yoach/tmp/artefacts/training-400M-punctuated-v2/",
|
3 |
+
"architectures": [
|
4 |
+
"ParlerTTSForConditionalGeneration"
|
5 |
+
],
|
6 |
+
"audio_encoder": {
|
7 |
+
"_name_or_path": "ylacombe/dac_44khZ_8kbps",
|
8 |
+
"add_cross_attention": false,
|
9 |
+
"architectures": [
|
10 |
+
"DACModel"
|
11 |
+
],
|
12 |
+
"bad_words_ids": null,
|
13 |
+
"begin_suppress_tokens": null,
|
14 |
+
"bos_token_id": null,
|
15 |
+
"chunk_size_feed_forward": 0,
|
16 |
+
"codebook_size": 1024,
|
17 |
+
"cross_attention_hidden_size": null,
|
18 |
+
"decoder_start_token_id": null,
|
19 |
+
"diversity_penalty": 0.0,
|
20 |
+
"do_sample": false,
|
21 |
+
"early_stopping": false,
|
22 |
+
"encoder_no_repeat_ngram_size": 0,
|
23 |
+
"eos_token_id": null,
|
24 |
+
"exponential_decay_length_penalty": null,
|
25 |
+
"finetuning_task": null,
|
26 |
+
"forced_bos_token_id": null,
|
27 |
+
"forced_eos_token_id": null,
|
28 |
+
"frame_rate": 86,
|
29 |
+
"id2label": {
|
30 |
+
"0": "LABEL_0",
|
31 |
+
"1": "LABEL_1"
|
32 |
+
},
|
33 |
+
"is_decoder": false,
|
34 |
+
"is_encoder_decoder": false,
|
35 |
+
"label2id": {
|
36 |
+
"LABEL_0": 0,
|
37 |
+
"LABEL_1": 1
|
38 |
+
},
|
39 |
+
"latent_dim": 1024,
|
40 |
+
"length_penalty": 1.0,
|
41 |
+
"max_length": 20,
|
42 |
+
"min_length": 0,
|
43 |
+
"model_bitrate": 8,
|
44 |
+
"model_type": "dac",
|
45 |
+
"no_repeat_ngram_size": 0,
|
46 |
+
"num_beam_groups": 1,
|
47 |
+
"num_beams": 1,
|
48 |
+
"num_codebooks": 9,
|
49 |
+
"num_return_sequences": 1,
|
50 |
+
"output_attentions": false,
|
51 |
+
"output_hidden_states": false,
|
52 |
+
"output_scores": false,
|
53 |
+
"pad_token_id": null,
|
54 |
+
"prefix": null,
|
55 |
+
"problem_type": null,
|
56 |
+
"pruned_heads": {},
|
57 |
+
"remove_invalid_values": false,
|
58 |
+
"repetition_penalty": 1.0,
|
59 |
+
"return_dict": true,
|
60 |
+
"return_dict_in_generate": false,
|
61 |
+
"sampling_rate": 44100,
|
62 |
+
"sep_token_id": null,
|
63 |
+
"suppress_tokens": null,
|
64 |
+
"task_specific_params": null,
|
65 |
+
"temperature": 1.0,
|
66 |
+
"tf_legacy_loss": false,
|
67 |
+
"tie_encoder_decoder": false,
|
68 |
+
"tie_word_embeddings": true,
|
69 |
+
"tokenizer_class": null,
|
70 |
+
"top_k": 50,
|
71 |
+
"top_p": 1.0,
|
72 |
+
"torch_dtype": "float32",
|
73 |
+
"torchscript": false,
|
74 |
+
"typical_p": 1.0,
|
75 |
+
"use_bfloat16": false
|
76 |
+
},
|
77 |
+
"decoder": {
|
78 |
+
"_name_or_path": "/fsx/yoach/tmp/artefacts/decoder_400M/",
|
79 |
+
"activation_dropout": 0.0,
|
80 |
+
"activation_function": "gelu",
|
81 |
+
"add_cross_attention": true,
|
82 |
+
"architectures": [
|
83 |
+
"ParlerTTSForCausalLM"
|
84 |
+
],
|
85 |
+
"attention_dropout": 0.0,
|
86 |
+
"bad_words_ids": null,
|
87 |
+
"begin_suppress_tokens": null,
|
88 |
+
"bos_token_id": 1025,
|
89 |
+
"chunk_size_feed_forward": 0,
|
90 |
+
"cross_attention_hidden_size": null,
|
91 |
+
"decoder_start_token_id": null,
|
92 |
+
"diversity_penalty": 0.0,
|
93 |
+
"do_sample": false,
|
94 |
+
"dropout": 0.1,
|
95 |
+
"early_stopping": false,
|
96 |
+
"encoder_no_repeat_ngram_size": 0,
|
97 |
+
"eos_token_id": 1024,
|
98 |
+
"exponential_decay_length_penalty": null,
|
99 |
+
"ffn_dim": 4096,
|
100 |
+
"finetuning_task": null,
|
101 |
+
"forced_bos_token_id": null,
|
102 |
+
"forced_eos_token_id": null,
|
103 |
+
"hidden_size": 1024,
|
104 |
+
"id2label": {
|
105 |
+
"0": "LABEL_0",
|
106 |
+
"1": "LABEL_1"
|
107 |
+
},
|
108 |
+
"initializer_factor": 0.02,
|
109 |
+
"is_decoder": true,
|
110 |
+
"is_encoder_decoder": false,
|
111 |
+
"label2id": {
|
112 |
+
"LABEL_0": 0,
|
113 |
+
"LABEL_1": 1
|
114 |
+
},
|
115 |
+
"layerdrop": 0.0,
|
116 |
+
"length_penalty": 1.0,
|
117 |
+
"max_length": 20,
|
118 |
+
"max_position_embeddings": 4096,
|
119 |
+
"min_length": 0,
|
120 |
+
"model_type": "parler_tts_decoder",
|
121 |
+
"no_repeat_ngram_size": 0,
|
122 |
+
"num_attention_heads": 16,
|
123 |
+
"num_beam_groups": 1,
|
124 |
+
"num_beams": 1,
|
125 |
+
"num_codebooks": 9,
|
126 |
+
"num_hidden_layers": 24,
|
127 |
+
"num_return_sequences": 1,
|
128 |
+
"output_attentions": false,
|
129 |
+
"output_hidden_states": false,
|
130 |
+
"output_scores": false,
|
131 |
+
"pad_token_id": 1024,
|
132 |
+
"prefix": null,
|
133 |
+
"problem_type": null,
|
134 |
+
"pruned_heads": {},
|
135 |
+
"remove_invalid_values": false,
|
136 |
+
"repetition_penalty": 1.0,
|
137 |
+
"return_dict": true,
|
138 |
+
"return_dict_in_generate": false,
|
139 |
+
"scale_embedding": false,
|
140 |
+
"sep_token_id": null,
|
141 |
+
"suppress_tokens": null,
|
142 |
+
"task_specific_params": null,
|
143 |
+
"temperature": 1.0,
|
144 |
+
"tf_legacy_loss": false,
|
145 |
+
"tie_encoder_decoder": false,
|
146 |
+
"tie_word_embeddings": false,
|
147 |
+
"tokenizer_class": null,
|
148 |
+
"top_k": 50,
|
149 |
+
"top_p": 1.0,
|
150 |
+
"torch_dtype": "float32",
|
151 |
+
"torchscript": false,
|
152 |
+
"typical_p": 1.0,
|
153 |
+
"use_bfloat16": false,
|
154 |
+
"use_cache": true,
|
155 |
+
"vocab_size": 1088
|
156 |
+
},
|
157 |
+
"decoder_start_token_id": 1025,
|
158 |
+
"is_encoder_decoder": true,
|
159 |
+
"model_type": "parler_tts",
|
160 |
+
"pad_token_id": 1024,
|
161 |
+
"text_encoder": {
|
162 |
+
"_name_or_path": "google/flan-t5-base",
|
163 |
+
"add_cross_attention": false,
|
164 |
+
"architectures": [
|
165 |
+
"T5ForConditionalGeneration"
|
166 |
+
],
|
167 |
+
"bad_words_ids": null,
|
168 |
+
"begin_suppress_tokens": null,
|
169 |
+
"bos_token_id": null,
|
170 |
+
"chunk_size_feed_forward": 0,
|
171 |
+
"classifier_dropout": 0.0,
|
172 |
+
"cross_attention_hidden_size": null,
|
173 |
+
"d_ff": 2048,
|
174 |
+
"d_kv": 64,
|
175 |
+
"d_model": 768,
|
176 |
+
"decoder_start_token_id": 0,
|
177 |
+
"dense_act_fn": "gelu_new",
|
178 |
+
"diversity_penalty": 0.0,
|
179 |
+
"do_sample": false,
|
180 |
+
"dropout_rate": 0.1,
|
181 |
+
"early_stopping": false,
|
182 |
+
"encoder_no_repeat_ngram_size": 0,
|
183 |
+
"eos_token_id": 1,
|
184 |
+
"exponential_decay_length_penalty": null,
|
185 |
+
"feed_forward_proj": "gated-gelu",
|
186 |
+
"finetuning_task": null,
|
187 |
+
"forced_bos_token_id": null,
|
188 |
+
"forced_eos_token_id": null,
|
189 |
+
"id2label": {
|
190 |
+
"0": "LABEL_0",
|
191 |
+
"1": "LABEL_1"
|
192 |
+
},
|
193 |
+
"initializer_factor": 1.0,
|
194 |
+
"is_decoder": false,
|
195 |
+
"is_encoder_decoder": true,
|
196 |
+
"is_gated_act": true,
|
197 |
+
"label2id": {
|
198 |
+
"LABEL_0": 0,
|
199 |
+
"LABEL_1": 1
|
200 |
+
},
|
201 |
+
"layer_norm_epsilon": 1e-06,
|
202 |
+
"length_penalty": 1.0,
|
203 |
+
"max_length": 20,
|
204 |
+
"min_length": 0,
|
205 |
+
"model_type": "t5",
|
206 |
+
"n_positions": 512,
|
207 |
+
"no_repeat_ngram_size": 0,
|
208 |
+
"num_beam_groups": 1,
|
209 |
+
"num_beams": 1,
|
210 |
+
"num_decoder_layers": 12,
|
211 |
+
"num_heads": 12,
|
212 |
+
"num_layers": 12,
|
213 |
+
"num_return_sequences": 1,
|
214 |
+
"output_attentions": false,
|
215 |
+
"output_hidden_states": false,
|
216 |
+
"output_past": true,
|
217 |
+
"output_scores": false,
|
218 |
+
"pad_token_id": 0,
|
219 |
+
"prefix": null,
|
220 |
+
"problem_type": null,
|
221 |
+
"pruned_heads": {},
|
222 |
+
"relative_attention_max_distance": 128,
|
223 |
+
"relative_attention_num_buckets": 32,
|
224 |
+
"remove_invalid_values": false,
|
225 |
+
"repetition_penalty": 1.0,
|
226 |
+
"return_dict": true,
|
227 |
+
"return_dict_in_generate": false,
|
228 |
+
"sep_token_id": null,
|
229 |
+
"suppress_tokens": null,
|
230 |
+
"task_specific_params": {
|
231 |
+
"summarization": {
|
232 |
+
"early_stopping": true,
|
233 |
+
"length_penalty": 2.0,
|
234 |
+
"max_length": 200,
|
235 |
+
"min_length": 30,
|
236 |
+
"no_repeat_ngram_size": 3,
|
237 |
+
"num_beams": 4,
|
238 |
+
"prefix": "summarize: "
|
239 |
+
},
|
240 |
+
"translation_en_to_de": {
|
241 |
+
"early_stopping": true,
|
242 |
+
"max_length": 300,
|
243 |
+
"num_beams": 4,
|
244 |
+
"prefix": "translate English to German: "
|
245 |
+
},
|
246 |
+
"translation_en_to_fr": {
|
247 |
+
"early_stopping": true,
|
248 |
+
"max_length": 300,
|
249 |
+
"num_beams": 4,
|
250 |
+
"prefix": "translate English to French: "
|
251 |
+
},
|
252 |
+
"translation_en_to_ro": {
|
253 |
+
"early_stopping": true,
|
254 |
+
"max_length": 300,
|
255 |
+
"num_beams": 4,
|
256 |
+
"prefix": "translate English to Romanian: "
|
257 |
+
}
|
258 |
+
},
|
259 |
+
"temperature": 1.0,
|
260 |
+
"tf_legacy_loss": false,
|
261 |
+
"tie_encoder_decoder": false,
|
262 |
+
"tie_word_embeddings": false,
|
263 |
+
"tokenizer_class": null,
|
264 |
+
"top_k": 50,
|
265 |
+
"top_p": 1.0,
|
266 |
+
"torch_dtype": null,
|
267 |
+
"torchscript": false,
|
268 |
+
"typical_p": 1.0,
|
269 |
+
"use_bfloat16": false,
|
270 |
+
"use_cache": true,
|
271 |
+
"vocab_size": 32128
|
272 |
+
},
|
273 |
+
"torch_dtype": "float32",
|
274 |
+
"transformers_version": "4.41.0.dev0",
|
275 |
+
"vocab_size": 32128
|
276 |
+
}
|
checkpoint-216-epoch-5/generation_config.json
ADDED
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_from_model_config": true,
|
3 |
+
"bos_token_id": 1025,
|
4 |
+
"decoder_start_token_id": 1025,
|
5 |
+
"do_sample": true,
|
6 |
+
"eos_token_id": 1024,
|
7 |
+
"guidance_scale": 1.0,
|
8 |
+
"max_length": 2580,
|
9 |
+
"min_new_tokens": 50,
|
10 |
+
"pad_token_id": 1024,
|
11 |
+
"transformers_version": "4.41.0.dev0"
|
12 |
+
}
|
checkpoint-288-epoch-7/config.json
ADDED
@@ -0,0 +1,276 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_name_or_path": "/fsx/yoach/tmp/artefacts/training-400M-punctuated-v2/",
|
3 |
+
"architectures": [
|
4 |
+
"ParlerTTSForConditionalGeneration"
|
5 |
+
],
|
6 |
+
"audio_encoder": {
|
7 |
+
"_name_or_path": "ylacombe/dac_44khZ_8kbps",
|
8 |
+
"add_cross_attention": false,
|
9 |
+
"architectures": [
|
10 |
+
"DACModel"
|
11 |
+
],
|
12 |
+
"bad_words_ids": null,
|
13 |
+
"begin_suppress_tokens": null,
|
14 |
+
"bos_token_id": null,
|
15 |
+
"chunk_size_feed_forward": 0,
|
16 |
+
"codebook_size": 1024,
|
17 |
+
"cross_attention_hidden_size": null,
|
18 |
+
"decoder_start_token_id": null,
|
19 |
+
"diversity_penalty": 0.0,
|
20 |
+
"do_sample": false,
|
21 |
+
"early_stopping": false,
|
22 |
+
"encoder_no_repeat_ngram_size": 0,
|
23 |
+
"eos_token_id": null,
|
24 |
+
"exponential_decay_length_penalty": null,
|
25 |
+
"finetuning_task": null,
|
26 |
+
"forced_bos_token_id": null,
|
27 |
+
"forced_eos_token_id": null,
|
28 |
+
"frame_rate": 86,
|
29 |
+
"id2label": {
|
30 |
+
"0": "LABEL_0",
|
31 |
+
"1": "LABEL_1"
|
32 |
+
},
|
33 |
+
"is_decoder": false,
|
34 |
+
"is_encoder_decoder": false,
|
35 |
+
"label2id": {
|
36 |
+
"LABEL_0": 0,
|
37 |
+
"LABEL_1": 1
|
38 |
+
},
|
39 |
+
"latent_dim": 1024,
|
40 |
+
"length_penalty": 1.0,
|
41 |
+
"max_length": 20,
|
42 |
+
"min_length": 0,
|
43 |
+
"model_bitrate": 8,
|
44 |
+
"model_type": "dac",
|
45 |
+
"no_repeat_ngram_size": 0,
|
46 |
+
"num_beam_groups": 1,
|
47 |
+
"num_beams": 1,
|
48 |
+
"num_codebooks": 9,
|
49 |
+
"num_return_sequences": 1,
|
50 |
+
"output_attentions": false,
|
51 |
+
"output_hidden_states": false,
|
52 |
+
"output_scores": false,
|
53 |
+
"pad_token_id": null,
|
54 |
+
"prefix": null,
|
55 |
+
"problem_type": null,
|
56 |
+
"pruned_heads": {},
|
57 |
+
"remove_invalid_values": false,
|
58 |
+
"repetition_penalty": 1.0,
|
59 |
+
"return_dict": true,
|
60 |
+
"return_dict_in_generate": false,
|
61 |
+
"sampling_rate": 44100,
|
62 |
+
"sep_token_id": null,
|
63 |
+
"suppress_tokens": null,
|
64 |
+
"task_specific_params": null,
|
65 |
+
"temperature": 1.0,
|
66 |
+
"tf_legacy_loss": false,
|
67 |
+
"tie_encoder_decoder": false,
|
68 |
+
"tie_word_embeddings": true,
|
69 |
+
"tokenizer_class": null,
|
70 |
+
"top_k": 50,
|
71 |
+
"top_p": 1.0,
|
72 |
+
"torch_dtype": "float32",
|
73 |
+
"torchscript": false,
|
74 |
+
"typical_p": 1.0,
|
75 |
+
"use_bfloat16": false
|
76 |
+
},
|
77 |
+
"decoder": {
|
78 |
+
"_name_or_path": "/fsx/yoach/tmp/artefacts/decoder_400M/",
|
79 |
+
"activation_dropout": 0.0,
|
80 |
+
"activation_function": "gelu",
|
81 |
+
"add_cross_attention": true,
|
82 |
+
"architectures": [
|
83 |
+
"ParlerTTSForCausalLM"
|
84 |
+
],
|
85 |
+
"attention_dropout": 0.0,
|
86 |
+
"bad_words_ids": null,
|
87 |
+
"begin_suppress_tokens": null,
|
88 |
+
"bos_token_id": 1025,
|
89 |
+
"chunk_size_feed_forward": 0,
|
90 |
+
"cross_attention_hidden_size": null,
|
91 |
+
"decoder_start_token_id": null,
|
92 |
+
"diversity_penalty": 0.0,
|
93 |
+
"do_sample": false,
|
94 |
+
"dropout": 0.1,
|
95 |
+
"early_stopping": false,
|
96 |
+
"encoder_no_repeat_ngram_size": 0,
|
97 |
+
"eos_token_id": 1024,
|
98 |
+
"exponential_decay_length_penalty": null,
|
99 |
+
"ffn_dim": 4096,
|
100 |
+
"finetuning_task": null,
|
101 |
+
"forced_bos_token_id": null,
|
102 |
+
"forced_eos_token_id": null,
|
103 |
+
"hidden_size": 1024,
|
104 |
+
"id2label": {
|
105 |
+
"0": "LABEL_0",
|
106 |
+
"1": "LABEL_1"
|
107 |
+
},
|
108 |
+
"initializer_factor": 0.02,
|
109 |
+
"is_decoder": true,
|
110 |
+
"is_encoder_decoder": false,
|
111 |
+
"label2id": {
|
112 |
+
"LABEL_0": 0,
|
113 |
+
"LABEL_1": 1
|
114 |
+
},
|
115 |
+
"layerdrop": 0.0,
|
116 |
+
"length_penalty": 1.0,
|
117 |
+
"max_length": 20,
|
118 |
+
"max_position_embeddings": 4096,
|
119 |
+
"min_length": 0,
|
120 |
+
"model_type": "parler_tts_decoder",
|
121 |
+
"no_repeat_ngram_size": 0,
|
122 |
+
"num_attention_heads": 16,
|
123 |
+
"num_beam_groups": 1,
|
124 |
+
"num_beams": 1,
|
125 |
+
"num_codebooks": 9,
|
126 |
+
"num_hidden_layers": 24,
|
127 |
+
"num_return_sequences": 1,
|
128 |
+
"output_attentions": false,
|
129 |
+
"output_hidden_states": false,
|
130 |
+
"output_scores": false,
|
131 |
+
"pad_token_id": 1024,
|
132 |
+
"prefix": null,
|
133 |
+
"problem_type": null,
|
134 |
+
"pruned_heads": {},
|
135 |
+
"remove_invalid_values": false,
|
136 |
+
"repetition_penalty": 1.0,
|
137 |
+
"return_dict": true,
|
138 |
+
"return_dict_in_generate": false,
|
139 |
+
"scale_embedding": false,
|
140 |
+
"sep_token_id": null,
|
141 |
+
"suppress_tokens": null,
|
142 |
+
"task_specific_params": null,
|
143 |
+
"temperature": 1.0,
|
144 |
+
"tf_legacy_loss": false,
|
145 |
+
"tie_encoder_decoder": false,
|
146 |
+
"tie_word_embeddings": false,
|
147 |
+
"tokenizer_class": null,
|
148 |
+
"top_k": 50,
|
149 |
+
"top_p": 1.0,
|
150 |
+
"torch_dtype": "float32",
|
151 |
+
"torchscript": false,
|
152 |
+
"typical_p": 1.0,
|
153 |
+
"use_bfloat16": false,
|
154 |
+
"use_cache": true,
|
155 |
+
"vocab_size": 1088
|
156 |
+
},
|
157 |
+
"decoder_start_token_id": 1025,
|
158 |
+
"is_encoder_decoder": true,
|
159 |
+
"model_type": "parler_tts",
|
160 |
+
"pad_token_id": 1024,
|
161 |
+
"text_encoder": {
|
162 |
+
"_name_or_path": "google/flan-t5-base",
|
163 |
+
"add_cross_attention": false,
|
164 |
+
"architectures": [
|
165 |
+
"T5ForConditionalGeneration"
|
166 |
+
],
|
167 |
+
"bad_words_ids": null,
|
168 |
+
"begin_suppress_tokens": null,
|
169 |
+
"bos_token_id": null,
|
170 |
+
"chunk_size_feed_forward": 0,
|
171 |
+
"classifier_dropout": 0.0,
|
172 |
+
"cross_attention_hidden_size": null,
|
173 |
+
"d_ff": 2048,
|
174 |
+
"d_kv": 64,
|
175 |
+
"d_model": 768,
|
176 |
+
"decoder_start_token_id": 0,
|
177 |
+
"dense_act_fn": "gelu_new",
|
178 |
+
"diversity_penalty": 0.0,
|
179 |
+
"do_sample": false,
|
180 |
+
"dropout_rate": 0.1,
|
181 |
+
"early_stopping": false,
|
182 |
+
"encoder_no_repeat_ngram_size": 0,
|
183 |
+
"eos_token_id": 1,
|
184 |
+
"exponential_decay_length_penalty": null,
|
185 |
+
"feed_forward_proj": "gated-gelu",
|
186 |
+
"finetuning_task": null,
|
187 |
+
"forced_bos_token_id": null,
|
188 |
+
"forced_eos_token_id": null,
|
189 |
+
"id2label": {
|
190 |
+
"0": "LABEL_0",
|
191 |
+
"1": "LABEL_1"
|
192 |
+
},
|
193 |
+
"initializer_factor": 1.0,
|
194 |
+
"is_decoder": false,
|
195 |
+
"is_encoder_decoder": true,
|
196 |
+
"is_gated_act": true,
|
197 |
+
"label2id": {
|
198 |
+
"LABEL_0": 0,
|
199 |
+
"LABEL_1": 1
|
200 |
+
},
|
201 |
+
"layer_norm_epsilon": 1e-06,
|
202 |
+
"length_penalty": 1.0,
|
203 |
+
"max_length": 20,
|
204 |
+
"min_length": 0,
|
205 |
+
"model_type": "t5",
|
206 |
+
"n_positions": 512,
|
207 |
+
"no_repeat_ngram_size": 0,
|
208 |
+
"num_beam_groups": 1,
|
209 |
+
"num_beams": 1,
|
210 |
+
"num_decoder_layers": 12,
|
211 |
+
"num_heads": 12,
|
212 |
+
"num_layers": 12,
|
213 |
+
"num_return_sequences": 1,
|
214 |
+
"output_attentions": false,
|
215 |
+
"output_hidden_states": false,
|
216 |
+
"output_past": true,
|
217 |
+
"output_scores": false,
|
218 |
+
"pad_token_id": 0,
|
219 |
+
"prefix": null,
|
220 |
+
"problem_type": null,
|
221 |
+
"pruned_heads": {},
|
222 |
+
"relative_attention_max_distance": 128,
|
223 |
+
"relative_attention_num_buckets": 32,
|
224 |
+
"remove_invalid_values": false,
|
225 |
+
"repetition_penalty": 1.0,
|
226 |
+
"return_dict": true,
|
227 |
+
"return_dict_in_generate": false,
|
228 |
+
"sep_token_id": null,
|
229 |
+
"suppress_tokens": null,
|
230 |
+
"task_specific_params": {
|
231 |
+
"summarization": {
|
232 |
+
"early_stopping": true,
|
233 |
+
"length_penalty": 2.0,
|
234 |
+
"max_length": 200,
|
235 |
+
"min_length": 30,
|
236 |
+
"no_repeat_ngram_size": 3,
|
237 |
+
"num_beams": 4,
|
238 |
+
"prefix": "summarize: "
|
239 |
+
},
|
240 |
+
"translation_en_to_de": {
|
241 |
+
"early_stopping": true,
|
242 |
+
"max_length": 300,
|
243 |
+
"num_beams": 4,
|
244 |
+
"prefix": "translate English to German: "
|
245 |
+
},
|
246 |
+
"translation_en_to_fr": {
|
247 |
+
"early_stopping": true,
|
248 |
+
"max_length": 300,
|
249 |
+
"num_beams": 4,
|
250 |
+
"prefix": "translate English to French: "
|
251 |
+
},
|
252 |
+
"translation_en_to_ro": {
|
253 |
+
"early_stopping": true,
|
254 |
+
"max_length": 300,
|
255 |
+
"num_beams": 4,
|
256 |
+
"prefix": "translate English to Romanian: "
|
257 |
+
}
|
258 |
+
},
|
259 |
+
"temperature": 1.0,
|
260 |
+
"tf_legacy_loss": false,
|
261 |
+
"tie_encoder_decoder": false,
|
262 |
+
"tie_word_embeddings": false,
|
263 |
+
"tokenizer_class": null,
|
264 |
+
"top_k": 50,
|
265 |
+
"top_p": 1.0,
|
266 |
+
"torch_dtype": null,
|
267 |
+
"torchscript": false,
|
268 |
+
"typical_p": 1.0,
|
269 |
+
"use_bfloat16": false,
|
270 |
+
"use_cache": true,
|
271 |
+
"vocab_size": 32128
|
272 |
+
},
|
273 |
+
"torch_dtype": "float32",
|
274 |
+
"transformers_version": "4.41.0.dev0",
|
275 |
+
"vocab_size": 32128
|
276 |
+
}
|
checkpoint-288-epoch-7/generation_config.json
ADDED
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_from_model_config": true,
|
3 |
+
"bos_token_id": 1025,
|
4 |
+
"decoder_start_token_id": 1025,
|
5 |
+
"do_sample": true,
|
6 |
+
"eos_token_id": 1024,
|
7 |
+
"guidance_scale": 1.0,
|
8 |
+
"max_length": 2580,
|
9 |
+
"min_new_tokens": 50,
|
10 |
+
"pad_token_id": 1024,
|
11 |
+
"transformers_version": "4.41.0.dev0"
|
12 |
+
}
|
checkpoint-312-epoch-7/config.json
ADDED
@@ -0,0 +1,276 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_name_or_path": "/fsx/yoach/tmp/artefacts/training-400M-punctuated-v2/",
|
3 |
+
"architectures": [
|
4 |
+
"ParlerTTSForConditionalGeneration"
|
5 |
+
],
|
6 |
+
"audio_encoder": {
|
7 |
+
"_name_or_path": "ylacombe/dac_44khZ_8kbps",
|
8 |
+
"add_cross_attention": false,
|
9 |
+
"architectures": [
|
10 |
+
"DACModel"
|
11 |
+
],
|
12 |
+
"bad_words_ids": null,
|
13 |
+
"begin_suppress_tokens": null,
|
14 |
+
"bos_token_id": null,
|
15 |
+
"chunk_size_feed_forward": 0,
|
16 |
+
"codebook_size": 1024,
|
17 |
+
"cross_attention_hidden_size": null,
|
18 |
+
"decoder_start_token_id": null,
|
19 |
+
"diversity_penalty": 0.0,
|
20 |
+
"do_sample": false,
|
21 |
+
"early_stopping": false,
|
22 |
+
"encoder_no_repeat_ngram_size": 0,
|
23 |
+
"eos_token_id": null,
|
24 |
+
"exponential_decay_length_penalty": null,
|
25 |
+
"finetuning_task": null,
|
26 |
+
"forced_bos_token_id": null,
|
27 |
+
"forced_eos_token_id": null,
|
28 |
+
"frame_rate": 86,
|
29 |
+
"id2label": {
|
30 |
+
"0": "LABEL_0",
|
31 |
+
"1": "LABEL_1"
|
32 |
+
},
|
33 |
+
"is_decoder": false,
|
34 |
+
"is_encoder_decoder": false,
|
35 |
+
"label2id": {
|
36 |
+
"LABEL_0": 0,
|
37 |
+
"LABEL_1": 1
|
38 |
+
},
|
39 |
+
"latent_dim": 1024,
|
40 |
+
"length_penalty": 1.0,
|
41 |
+
"max_length": 20,
|
42 |
+
"min_length": 0,
|
43 |
+
"model_bitrate": 8,
|
44 |
+
"model_type": "dac",
|
45 |
+
"no_repeat_ngram_size": 0,
|
46 |
+
"num_beam_groups": 1,
|
47 |
+
"num_beams": 1,
|
48 |
+
"num_codebooks": 9,
|
49 |
+
"num_return_sequences": 1,
|
50 |
+
"output_attentions": false,
|
51 |
+
"output_hidden_states": false,
|
52 |
+
"output_scores": false,
|
53 |
+
"pad_token_id": null,
|
54 |
+
"prefix": null,
|
55 |
+
"problem_type": null,
|
56 |
+
"pruned_heads": {},
|
57 |
+
"remove_invalid_values": false,
|
58 |
+
"repetition_penalty": 1.0,
|
59 |
+
"return_dict": true,
|
60 |
+
"return_dict_in_generate": false,
|
61 |
+
"sampling_rate": 44100,
|
62 |
+
"sep_token_id": null,
|
63 |
+
"suppress_tokens": null,
|
64 |
+
"task_specific_params": null,
|
65 |
+
"temperature": 1.0,
|
66 |
+
"tf_legacy_loss": false,
|
67 |
+
"tie_encoder_decoder": false,
|
68 |
+
"tie_word_embeddings": true,
|
69 |
+
"tokenizer_class": null,
|
70 |
+
"top_k": 50,
|
71 |
+
"top_p": 1.0,
|
72 |
+
"torch_dtype": "float32",
|
73 |
+
"torchscript": false,
|
74 |
+
"typical_p": 1.0,
|
75 |
+
"use_bfloat16": false
|
76 |
+
},
|
77 |
+
"decoder": {
|
78 |
+
"_name_or_path": "/fsx/yoach/tmp/artefacts/decoder_400M/",
|
79 |
+
"activation_dropout": 0.0,
|
80 |
+
"activation_function": "gelu",
|
81 |
+
"add_cross_attention": true,
|
82 |
+
"architectures": [
|
83 |
+
"ParlerTTSForCausalLM"
|
84 |
+
],
|
85 |
+
"attention_dropout": 0.0,
|
86 |
+
"bad_words_ids": null,
|
87 |
+
"begin_suppress_tokens": null,
|
88 |
+
"bos_token_id": 1025,
|
89 |
+
"chunk_size_feed_forward": 0,
|
90 |
+
"cross_attention_hidden_size": null,
|
91 |
+
"decoder_start_token_id": null,
|
92 |
+
"diversity_penalty": 0.0,
|
93 |
+
"do_sample": false,
|
94 |
+
"dropout": 0.1,
|
95 |
+
"early_stopping": false,
|
96 |
+
"encoder_no_repeat_ngram_size": 0,
|
97 |
+
"eos_token_id": 1024,
|
98 |
+
"exponential_decay_length_penalty": null,
|
99 |
+
"ffn_dim": 4096,
|
100 |
+
"finetuning_task": null,
|
101 |
+
"forced_bos_token_id": null,
|
102 |
+
"forced_eos_token_id": null,
|
103 |
+
"hidden_size": 1024,
|
104 |
+
"id2label": {
|
105 |
+
"0": "LABEL_0",
|
106 |
+
"1": "LABEL_1"
|
107 |
+
},
|
108 |
+
"initializer_factor": 0.02,
|
109 |
+
"is_decoder": true,
|
110 |
+
"is_encoder_decoder": false,
|
111 |
+
"label2id": {
|
112 |
+
"LABEL_0": 0,
|
113 |
+
"LABEL_1": 1
|
114 |
+
},
|
115 |
+
"layerdrop": 0.0,
|
116 |
+
"length_penalty": 1.0,
|
117 |
+
"max_length": 20,
|
118 |
+
"max_position_embeddings": 4096,
|
119 |
+
"min_length": 0,
|
120 |
+
"model_type": "parler_tts_decoder",
|
121 |
+
"no_repeat_ngram_size": 0,
|
122 |
+
"num_attention_heads": 16,
|
123 |
+
"num_beam_groups": 1,
|
124 |
+
"num_beams": 1,
|
125 |
+
"num_codebooks": 9,
|
126 |
+
"num_hidden_layers": 24,
|
127 |
+
"num_return_sequences": 1,
|
128 |
+
"output_attentions": false,
|
129 |
+
"output_hidden_states": false,
|
130 |
+
"output_scores": false,
|
131 |
+
"pad_token_id": 1024,
|
132 |
+
"prefix": null,
|
133 |
+
"problem_type": null,
|
134 |
+
"pruned_heads": {},
|
135 |
+
"remove_invalid_values": false,
|
136 |
+
"repetition_penalty": 1.0,
|
137 |
+
"return_dict": true,
|
138 |
+
"return_dict_in_generate": false,
|
139 |
+
"scale_embedding": false,
|
140 |
+
"sep_token_id": null,
|
141 |
+
"suppress_tokens": null,
|
142 |
+
"task_specific_params": null,
|
143 |
+
"temperature": 1.0,
|
144 |
+
"tf_legacy_loss": false,
|
145 |
+
"tie_encoder_decoder": false,
|
146 |
+
"tie_word_embeddings": false,
|
147 |
+
"tokenizer_class": null,
|
148 |
+
"top_k": 50,
|
149 |
+
"top_p": 1.0,
|
150 |
+
"torch_dtype": "float32",
|
151 |
+
"torchscript": false,
|
152 |
+
"typical_p": 1.0,
|
153 |
+
"use_bfloat16": false,
|
154 |
+
"use_cache": true,
|
155 |
+
"vocab_size": 1088
|
156 |
+
},
|
157 |
+
"decoder_start_token_id": 1025,
|
158 |
+
"is_encoder_decoder": true,
|
159 |
+
"model_type": "parler_tts",
|
160 |
+
"pad_token_id": 1024,
|
161 |
+
"text_encoder": {
|
162 |
+
"_name_or_path": "google/flan-t5-base",
|
163 |
+
"add_cross_attention": false,
|
164 |
+
"architectures": [
|
165 |
+
"T5ForConditionalGeneration"
|
166 |
+
],
|
167 |
+
"bad_words_ids": null,
|
168 |
+
"begin_suppress_tokens": null,
|
169 |
+
"bos_token_id": null,
|
170 |
+
"chunk_size_feed_forward": 0,
|
171 |
+
"classifier_dropout": 0.0,
|
172 |
+
"cross_attention_hidden_size": null,
|
173 |
+
"d_ff": 2048,
|
174 |
+
"d_kv": 64,
|
175 |
+
"d_model": 768,
|
176 |
+
"decoder_start_token_id": 0,
|
177 |
+
"dense_act_fn": "gelu_new",
|
178 |
+
"diversity_penalty": 0.0,
|
179 |
+
"do_sample": false,
|
180 |
+
"dropout_rate": 0.1,
|
181 |
+
"early_stopping": false,
|
182 |
+
"encoder_no_repeat_ngram_size": 0,
|
183 |
+
"eos_token_id": 1,
|
184 |
+
"exponential_decay_length_penalty": null,
|
185 |
+
"feed_forward_proj": "gated-gelu",
|
186 |
+
"finetuning_task": null,
|
187 |
+
"forced_bos_token_id": null,
|
188 |
+
"forced_eos_token_id": null,
|
189 |
+
"id2label": {
|
190 |
+
"0": "LABEL_0",
|
191 |
+
"1": "LABEL_1"
|
192 |
+
},
|
193 |
+
"initializer_factor": 1.0,
|
194 |
+
"is_decoder": false,
|
195 |
+
"is_encoder_decoder": true,
|
196 |
+
"is_gated_act": true,
|
197 |
+
"label2id": {
|
198 |
+
"LABEL_0": 0,
|
199 |
+
"LABEL_1": 1
|
200 |
+
},
|
201 |
+
"layer_norm_epsilon": 1e-06,
|
202 |
+
"length_penalty": 1.0,
|
203 |
+
"max_length": 20,
|
204 |
+
"min_length": 0,
|
205 |
+
"model_type": "t5",
|
206 |
+
"n_positions": 512,
|
207 |
+
"no_repeat_ngram_size": 0,
|
208 |
+
"num_beam_groups": 1,
|
209 |
+
"num_beams": 1,
|
210 |
+
"num_decoder_layers": 12,
|
211 |
+
"num_heads": 12,
|
212 |
+
"num_layers": 12,
|
213 |
+
"num_return_sequences": 1,
|
214 |
+
"output_attentions": false,
|
215 |
+
"output_hidden_states": false,
|
216 |
+
"output_past": true,
|
217 |
+
"output_scores": false,
|
218 |
+
"pad_token_id": 0,
|
219 |
+
"prefix": null,
|
220 |
+
"problem_type": null,
|
221 |
+
"pruned_heads": {},
|
222 |
+
"relative_attention_max_distance": 128,
|
223 |
+
"relative_attention_num_buckets": 32,
|
224 |
+
"remove_invalid_values": false,
|
225 |
+
"repetition_penalty": 1.0,
|
226 |
+
"return_dict": true,
|
227 |
+
"return_dict_in_generate": false,
|
228 |
+
"sep_token_id": null,
|
229 |
+
"suppress_tokens": null,
|
230 |
+
"task_specific_params": {
|
231 |
+
"summarization": {
|
232 |
+
"early_stopping": true,
|
233 |
+
"length_penalty": 2.0,
|
234 |
+
"max_length": 200,
|
235 |
+
"min_length": 30,
|
236 |
+
"no_repeat_ngram_size": 3,
|
237 |
+
"num_beams": 4,
|
238 |
+
"prefix": "summarize: "
|
239 |
+
},
|
240 |
+
"translation_en_to_de": {
|
241 |
+
"early_stopping": true,
|
242 |
+
"max_length": 300,
|
243 |
+
"num_beams": 4,
|
244 |
+
"prefix": "translate English to German: "
|
245 |
+
},
|
246 |
+
"translation_en_to_fr": {
|
247 |
+
"early_stopping": true,
|
248 |
+
"max_length": 300,
|
249 |
+
"num_beams": 4,
|
250 |
+
"prefix": "translate English to French: "
|
251 |
+
},
|
252 |
+
"translation_en_to_ro": {
|
253 |
+
"early_stopping": true,
|
254 |
+
"max_length": 300,
|
255 |
+
"num_beams": 4,
|
256 |
+
"prefix": "translate English to Romanian: "
|
257 |
+
}
|
258 |
+
},
|
259 |
+
"temperature": 1.0,
|
260 |
+
"tf_legacy_loss": false,
|
261 |
+
"tie_encoder_decoder": false,
|
262 |
+
"tie_word_embeddings": false,
|
263 |
+
"tokenizer_class": null,
|
264 |
+
"top_k": 50,
|
265 |
+
"top_p": 1.0,
|
266 |
+
"torch_dtype": null,
|
267 |
+
"torchscript": false,
|
268 |
+
"typical_p": 1.0,
|
269 |
+
"use_bfloat16": false,
|
270 |
+
"use_cache": true,
|
271 |
+
"vocab_size": 32128
|
272 |
+
},
|
273 |
+
"torch_dtype": "float32",
|
274 |
+
"transformers_version": "4.41.0.dev0",
|
275 |
+
"vocab_size": 32128
|
276 |
+
}
|
checkpoint-312-epoch-7/generation_config.json
ADDED
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_from_model_config": true,
|
3 |
+
"bos_token_id": 1025,
|
4 |
+
"decoder_start_token_id": 1025,
|
5 |
+
"do_sample": true,
|
6 |
+
"eos_token_id": 1024,
|
7 |
+
"guidance_scale": 1.0,
|
8 |
+
"max_length": 2580,
|
9 |
+
"min_new_tokens": 50,
|
10 |
+
"pad_token_id": 1024,
|
11 |
+
"transformers_version": "4.41.0.dev0"
|
12 |
+
}
|
checkpoint-72-epoch-1/config.json
ADDED
@@ -0,0 +1,276 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_name_or_path": "/fsx/yoach/tmp/artefacts/training-400M-punctuated-v2/",
|
3 |
+
"architectures": [
|
4 |
+
"ParlerTTSForConditionalGeneration"
|
5 |
+
],
|
6 |
+
"audio_encoder": {
|
7 |
+
"_name_or_path": "ylacombe/dac_44khZ_8kbps",
|
8 |
+
"add_cross_attention": false,
|
9 |
+
"architectures": [
|
10 |
+
"DACModel"
|
11 |
+
],
|
12 |
+
"bad_words_ids": null,
|
13 |
+
"begin_suppress_tokens": null,
|
14 |
+
"bos_token_id": null,
|
15 |
+
"chunk_size_feed_forward": 0,
|
16 |
+
"codebook_size": 1024,
|
17 |
+
"cross_attention_hidden_size": null,
|
18 |
+
"decoder_start_token_id": null,
|
19 |
+
"diversity_penalty": 0.0,
|
20 |
+
"do_sample": false,
|
21 |
+
"early_stopping": false,
|
22 |
+
"encoder_no_repeat_ngram_size": 0,
|
23 |
+
"eos_token_id": null,
|
24 |
+
"exponential_decay_length_penalty": null,
|
25 |
+
"finetuning_task": null,
|
26 |
+
"forced_bos_token_id": null,
|
27 |
+
"forced_eos_token_id": null,
|
28 |
+
"frame_rate": 86,
|
29 |
+
"id2label": {
|
30 |
+
"0": "LABEL_0",
|
31 |
+
"1": "LABEL_1"
|
32 |
+
},
|
33 |
+
"is_decoder": false,
|
34 |
+
"is_encoder_decoder": false,
|
35 |
+
"label2id": {
|
36 |
+
"LABEL_0": 0,
|
37 |
+
"LABEL_1": 1
|
38 |
+
},
|
39 |
+
"latent_dim": 1024,
|
40 |
+
"length_penalty": 1.0,
|
41 |
+
"max_length": 20,
|
42 |
+
"min_length": 0,
|
43 |
+
"model_bitrate": 8,
|
44 |
+
"model_type": "dac",
|
45 |
+
"no_repeat_ngram_size": 0,
|
46 |
+
"num_beam_groups": 1,
|
47 |
+
"num_beams": 1,
|
48 |
+
"num_codebooks": 9,
|
49 |
+
"num_return_sequences": 1,
|
50 |
+
"output_attentions": false,
|
51 |
+
"output_hidden_states": false,
|
52 |
+
"output_scores": false,
|
53 |
+
"pad_token_id": null,
|
54 |
+
"prefix": null,
|
55 |
+
"problem_type": null,
|
56 |
+
"pruned_heads": {},
|
57 |
+
"remove_invalid_values": false,
|
58 |
+
"repetition_penalty": 1.0,
|
59 |
+
"return_dict": true,
|
60 |
+
"return_dict_in_generate": false,
|
61 |
+
"sampling_rate": 44100,
|
62 |
+
"sep_token_id": null,
|
63 |
+
"suppress_tokens": null,
|
64 |
+
"task_specific_params": null,
|
65 |
+
"temperature": 1.0,
|
66 |
+
"tf_legacy_loss": false,
|
67 |
+
"tie_encoder_decoder": false,
|
68 |
+
"tie_word_embeddings": true,
|
69 |
+
"tokenizer_class": null,
|
70 |
+
"top_k": 50,
|
71 |
+
"top_p": 1.0,
|
72 |
+
"torch_dtype": "float32",
|
73 |
+
"torchscript": false,
|
74 |
+
"typical_p": 1.0,
|
75 |
+
"use_bfloat16": false
|
76 |
+
},
|
77 |
+
"decoder": {
|
78 |
+
"_name_or_path": "/fsx/yoach/tmp/artefacts/decoder_400M/",
|
79 |
+
"activation_dropout": 0.0,
|
80 |
+
"activation_function": "gelu",
|
81 |
+
"add_cross_attention": true,
|
82 |
+
"architectures": [
|
83 |
+
"ParlerTTSForCausalLM"
|
84 |
+
],
|
85 |
+
"attention_dropout": 0.0,
|
86 |
+
"bad_words_ids": null,
|
87 |
+
"begin_suppress_tokens": null,
|
88 |
+
"bos_token_id": 1025,
|
89 |
+
"chunk_size_feed_forward": 0,
|
90 |
+
"cross_attention_hidden_size": null,
|
91 |
+
"decoder_start_token_id": null,
|
92 |
+
"diversity_penalty": 0.0,
|
93 |
+
"do_sample": false,
|
94 |
+
"dropout": 0.1,
|
95 |
+
"early_stopping": false,
|
96 |
+
"encoder_no_repeat_ngram_size": 0,
|
97 |
+
"eos_token_id": 1024,
|
98 |
+
"exponential_decay_length_penalty": null,
|
99 |
+
"ffn_dim": 4096,
|
100 |
+
"finetuning_task": null,
|
101 |
+
"forced_bos_token_id": null,
|
102 |
+
"forced_eos_token_id": null,
|
103 |
+
"hidden_size": 1024,
|
104 |
+
"id2label": {
|
105 |
+
"0": "LABEL_0",
|
106 |
+
"1": "LABEL_1"
|
107 |
+
},
|
108 |
+
"initializer_factor": 0.02,
|
109 |
+
"is_decoder": true,
|
110 |
+
"is_encoder_decoder": false,
|
111 |
+
"label2id": {
|
112 |
+
"LABEL_0": 0,
|
113 |
+
"LABEL_1": 1
|
114 |
+
},
|
115 |
+
"layerdrop": 0.0,
|
116 |
+
"length_penalty": 1.0,
|
117 |
+
"max_length": 20,
|
118 |
+
"max_position_embeddings": 4096,
|
119 |
+
"min_length": 0,
|
120 |
+
"model_type": "parler_tts_decoder",
|
121 |
+
"no_repeat_ngram_size": 0,
|
122 |
+
"num_attention_heads": 16,
|
123 |
+
"num_beam_groups": 1,
|
124 |
+
"num_beams": 1,
|
125 |
+
"num_codebooks": 9,
|
126 |
+
"num_hidden_layers": 24,
|
127 |
+
"num_return_sequences": 1,
|
128 |
+
"output_attentions": false,
|
129 |
+
"output_hidden_states": false,
|
130 |
+
"output_scores": false,
|
131 |
+
"pad_token_id": 1024,
|
132 |
+
"prefix": null,
|
133 |
+
"problem_type": null,
|
134 |
+
"pruned_heads": {},
|
135 |
+
"remove_invalid_values": false,
|
136 |
+
"repetition_penalty": 1.0,
|
137 |
+
"return_dict": true,
|
138 |
+
"return_dict_in_generate": false,
|
139 |
+
"scale_embedding": false,
|
140 |
+
"sep_token_id": null,
|
141 |
+
"suppress_tokens": null,
|
142 |
+
"task_specific_params": null,
|
143 |
+
"temperature": 1.0,
|
144 |
+
"tf_legacy_loss": false,
|
145 |
+
"tie_encoder_decoder": false,
|
146 |
+
"tie_word_embeddings": false,
|
147 |
+
"tokenizer_class": null,
|
148 |
+
"top_k": 50,
|
149 |
+
"top_p": 1.0,
|
150 |
+
"torch_dtype": "float32",
|
151 |
+
"torchscript": false,
|
152 |
+
"typical_p": 1.0,
|
153 |
+
"use_bfloat16": false,
|
154 |
+
"use_cache": true,
|
155 |
+
"vocab_size": 1088
|
156 |
+
},
|
157 |
+
"decoder_start_token_id": 1025,
|
158 |
+
"is_encoder_decoder": true,
|
159 |
+
"model_type": "parler_tts",
|
160 |
+
"pad_token_id": 1024,
|
161 |
+
"text_encoder": {
|
162 |
+
"_name_or_path": "google/flan-t5-base",
|
163 |
+
"add_cross_attention": false,
|
164 |
+
"architectures": [
|
165 |
+
"T5ForConditionalGeneration"
|
166 |
+
],
|
167 |
+
"bad_words_ids": null,
|
168 |
+
"begin_suppress_tokens": null,
|
169 |
+
"bos_token_id": null,
|
170 |
+
"chunk_size_feed_forward": 0,
|
171 |
+
"classifier_dropout": 0.0,
|
172 |
+
"cross_attention_hidden_size": null,
|
173 |
+
"d_ff": 2048,
|
174 |
+
"d_kv": 64,
|
175 |
+
"d_model": 768,
|
176 |
+
"decoder_start_token_id": 0,
|
177 |
+
"dense_act_fn": "gelu_new",
|
178 |
+
"diversity_penalty": 0.0,
|
179 |
+
"do_sample": false,
|
180 |
+
"dropout_rate": 0.1,
|
181 |
+
"early_stopping": false,
|
182 |
+
"encoder_no_repeat_ngram_size": 0,
|
183 |
+
"eos_token_id": 1,
|
184 |
+
"exponential_decay_length_penalty": null,
|
185 |
+
"feed_forward_proj": "gated-gelu",
|
186 |
+
"finetuning_task": null,
|
187 |
+
"forced_bos_token_id": null,
|
188 |
+
"forced_eos_token_id": null,
|
189 |
+
"id2label": {
|
190 |
+
"0": "LABEL_0",
|
191 |
+
"1": "LABEL_1"
|
192 |
+
},
|
193 |
+
"initializer_factor": 1.0,
|
194 |
+
"is_decoder": false,
|
195 |
+
"is_encoder_decoder": true,
|
196 |
+
"is_gated_act": true,
|
197 |
+
"label2id": {
|
198 |
+
"LABEL_0": 0,
|
199 |
+
"LABEL_1": 1
|
200 |
+
},
|
201 |
+
"layer_norm_epsilon": 1e-06,
|
202 |
+
"length_penalty": 1.0,
|
203 |
+
"max_length": 20,
|
204 |
+
"min_length": 0,
|
205 |
+
"model_type": "t5",
|
206 |
+
"n_positions": 512,
|
207 |
+
"no_repeat_ngram_size": 0,
|
208 |
+
"num_beam_groups": 1,
|
209 |
+
"num_beams": 1,
|
210 |
+
"num_decoder_layers": 12,
|
211 |
+
"num_heads": 12,
|
212 |
+
"num_layers": 12,
|
213 |
+
"num_return_sequences": 1,
|
214 |
+
"output_attentions": false,
|
215 |
+
"output_hidden_states": false,
|
216 |
+
"output_past": true,
|
217 |
+
"output_scores": false,
|
218 |
+
"pad_token_id": 0,
|
219 |
+
"prefix": null,
|
220 |
+
"problem_type": null,
|
221 |
+
"pruned_heads": {},
|
222 |
+
"relative_attention_max_distance": 128,
|
223 |
+
"relative_attention_num_buckets": 32,
|
224 |
+
"remove_invalid_values": false,
|
225 |
+
"repetition_penalty": 1.0,
|
226 |
+
"return_dict": true,
|
227 |
+
"return_dict_in_generate": false,
|
228 |
+
"sep_token_id": null,
|
229 |
+
"suppress_tokens": null,
|
230 |
+
"task_specific_params": {
|
231 |
+
"summarization": {
|
232 |
+
"early_stopping": true,
|
233 |
+
"length_penalty": 2.0,
|
234 |
+
"max_length": 200,
|
235 |
+
"min_length": 30,
|
236 |
+
"no_repeat_ngram_size": 3,
|
237 |
+
"num_beams": 4,
|
238 |
+
"prefix": "summarize: "
|
239 |
+
},
|
240 |
+
"translation_en_to_de": {
|
241 |
+
"early_stopping": true,
|
242 |
+
"max_length": 300,
|
243 |
+
"num_beams": 4,
|
244 |
+
"prefix": "translate English to German: "
|
245 |
+
},
|
246 |
+
"translation_en_to_fr": {
|
247 |
+
"early_stopping": true,
|
248 |
+
"max_length": 300,
|
249 |
+
"num_beams": 4,
|
250 |
+
"prefix": "translate English to French: "
|
251 |
+
},
|
252 |
+
"translation_en_to_ro": {
|
253 |
+
"early_stopping": true,
|
254 |
+
"max_length": 300,
|
255 |
+
"num_beams": 4,
|
256 |
+
"prefix": "translate English to Romanian: "
|
257 |
+
}
|
258 |
+
},
|
259 |
+
"temperature": 1.0,
|
260 |
+
"tf_legacy_loss": false,
|
261 |
+
"tie_encoder_decoder": false,
|
262 |
+
"tie_word_embeddings": false,
|
263 |
+
"tokenizer_class": null,
|
264 |
+
"top_k": 50,
|
265 |
+
"top_p": 1.0,
|
266 |
+
"torch_dtype": null,
|
267 |
+
"torchscript": false,
|
268 |
+
"typical_p": 1.0,
|
269 |
+
"use_bfloat16": false,
|
270 |
+
"use_cache": true,
|
271 |
+
"vocab_size": 32128
|
272 |
+
},
|
273 |
+
"torch_dtype": "float32",
|
274 |
+
"transformers_version": "4.41.0.dev0",
|
275 |
+
"vocab_size": 32128
|
276 |
+
}
|
checkpoint-72-epoch-1/generation_config.json
ADDED
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_from_model_config": true,
|
3 |
+
"bos_token_id": 1025,
|
4 |
+
"decoder_start_token_id": 1025,
|
5 |
+
"do_sample": true,
|
6 |
+
"eos_token_id": 1024,
|
7 |
+
"guidance_scale": 1.0,
|
8 |
+
"max_length": 2580,
|
9 |
+
"min_new_tokens": 50,
|
10 |
+
"pad_token_id": 1024,
|
11 |
+
"transformers_version": "4.41.0.dev0"
|
12 |
+
}
|
run.sh
CHANGED
@@ -49,4 +49,5 @@ accelerate launch ./run_parler_tts_training.py \
|
|
49 |
--save_steps 72 \
|
50 |
--evaluation_strategy "epoch" \
|
51 |
--save_total_limit 5 \
|
52 |
-
--group_by_length true
|
|
|
|
49 |
--save_steps 72 \
|
50 |
--evaluation_strategy "epoch" \
|
51 |
--save_total_limit 5 \
|
52 |
+
--group_by_length true \
|
53 |
+
--push_to_hub
|
run_parler_tts_training.py
CHANGED
@@ -942,6 +942,7 @@ def main():
|
|
942 |
columns_to_keep = {
|
943 |
"target_audio_column_name": data_args.target_audio_column_name,
|
944 |
"prompt_column_name": data_args.prompt_column_name,
|
|
|
945 |
}
|
946 |
if data_args.description_column_name is not None:
|
947 |
columns_to_keep["description_column_name"] = data_args.description_column_name
|
@@ -1019,6 +1020,12 @@ def main():
|
|
1019 |
),
|
1020 |
}
|
1021 |
)
|
|
|
|
|
|
|
|
|
|
|
|
|
1022 |
|
1023 |
# create model
|
1024 |
model = ParlerTTSForConditionalGeneration.from_pretrained(
|
@@ -1028,6 +1035,7 @@ def main():
|
|
1028 |
token=data_args.token,
|
1029 |
trust_remote_code=data_args.trust_remote_code,
|
1030 |
)
|
|
|
1031 |
|
1032 |
# enable gradient checkpointing if necessary
|
1033 |
if training_args.gradient_checkpointing:
|
@@ -1619,6 +1627,8 @@ def main():
|
|
1619 |
# https://github.com/huggingface/transformers/issues/27293#issuecomment-1872560074
|
1620 |
accelerator.save_state(output_dir=intermediate_dir, safe_serialization=False)
|
1621 |
accelerator.wait_for_everyone()
|
|
|
|
|
1622 |
if accelerator.is_main_process:
|
1623 |
rotate_checkpoints(training_args.save_total_limit, output_dir=training_args.output_dir)
|
1624 |
|
@@ -1761,3 +1771,4 @@ def main():
|
|
1761 |
if __name__ == "__main__":
|
1762 |
set_start_method("spawn")
|
1763 |
main()
|
|
|
|
942 |
columns_to_keep = {
|
943 |
"target_audio_column_name": data_args.target_audio_column_name,
|
944 |
"prompt_column_name": data_args.prompt_column_name,
|
945 |
+
"style": "style",
|
946 |
}
|
947 |
if data_args.description_column_name is not None:
|
948 |
columns_to_keep["description_column_name"] = data_args.description_column_name
|
|
|
1020 |
),
|
1021 |
}
|
1022 |
)
|
1023 |
+
|
1024 |
+
if not dataset_was_precomputed:
|
1025 |
+
def remove_whispering(style):
|
1026 |
+
return style != "whisper"
|
1027 |
+
|
1028 |
+
raw_datasets = raw_datasets.filter(remove_whispering)
|
1029 |
|
1030 |
# create model
|
1031 |
model = ParlerTTSForConditionalGeneration.from_pretrained(
|
|
|
1035 |
token=data_args.token,
|
1036 |
trust_remote_code=data_args.trust_remote_code,
|
1037 |
)
|
1038 |
+
generation_config = model.generation_config
|
1039 |
|
1040 |
# enable gradient checkpointing if necessary
|
1041 |
if training_args.gradient_checkpointing:
|
|
|
1627 |
# https://github.com/huggingface/transformers/issues/27293#issuecomment-1872560074
|
1628 |
accelerator.save_state(output_dir=intermediate_dir, safe_serialization=False)
|
1629 |
accelerator.wait_for_everyone()
|
1630 |
+
config.save_pretrained(intermediate_dir)
|
1631 |
+
generation_config.save_pretrained(intermediate_dir)
|
1632 |
if accelerator.is_main_process:
|
1633 |
rotate_checkpoints(training_args.save_total_limit, output_dir=training_args.output_dir)
|
1634 |
|
|
|
1771 |
if __name__ == "__main__":
|
1772 |
set_start_method("spawn")
|
1773 |
main()
|
1774 |
+
|