Fix config files
#3
by
Matthijs
- opened
- added_tokens.json +4 -0
- config.json +5 -2
- generation_config.json +1 -1
- preprocessor_config.json +0 -5
- special_tokens_map.json +7 -0
added_tokens.json
ADDED
@@ -0,0 +1,4 @@
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"<ctc_blank>": 80,
|
3 |
+
"<mask>": 79
|
4 |
+
}
|
config.json
CHANGED
@@ -38,7 +38,6 @@
|
|
38 |
"decoder_ffn_dim": 3072,
|
39 |
"decoder_layerdrop": 0.1,
|
40 |
"decoder_layers": 6,
|
41 |
-
"decoder_max_relative_position": 160,
|
42 |
"decoder_start_token_id": 2,
|
43 |
"encoder_attention_heads": 12,
|
44 |
"encoder_ffn_dim": 3072,
|
@@ -49,6 +48,9 @@
|
|
49 |
"feat_extract_activation": "gelu",
|
50 |
"feat_extract_norm": "group",
|
51 |
"feat_proj_dropout": 0.0,
|
|
|
|
|
|
|
52 |
"hidden_act": "gelu",
|
53 |
"hidden_dropout": 0.1,
|
54 |
"hidden_size": 768,
|
@@ -82,7 +84,8 @@
|
|
82 |
"speech_decoder_prenet_layers": 2,
|
83 |
"speech_decoder_prenet_units": 256,
|
84 |
"torch_dtype": "float32",
|
85 |
-
"transformers_version": "4.
|
86 |
"use_cache": true,
|
|
|
87 |
"vocab_size": 81
|
88 |
}
|
|
|
38 |
"decoder_ffn_dim": 3072,
|
39 |
"decoder_layerdrop": 0.1,
|
40 |
"decoder_layers": 6,
|
|
|
41 |
"decoder_start_token_id": 2,
|
42 |
"encoder_attention_heads": 12,
|
43 |
"encoder_ffn_dim": 3072,
|
|
|
48 |
"feat_extract_activation": "gelu",
|
49 |
"feat_extract_norm": "group",
|
50 |
"feat_proj_dropout": 0.0,
|
51 |
+
"guided_attention_loss_num_heads": 2,
|
52 |
+
"guided_attention_loss_scale": 10.0,
|
53 |
+
"guided_attention_loss_sigma": 0.4,
|
54 |
"hidden_act": "gelu",
|
55 |
"hidden_dropout": 0.1,
|
56 |
"hidden_size": 768,
|
|
|
84 |
"speech_decoder_prenet_layers": 2,
|
85 |
"speech_decoder_prenet_units": 256,
|
86 |
"torch_dtype": "float32",
|
87 |
+
"transformers_version": "4.28.0.dev0",
|
88 |
"use_cache": true,
|
89 |
+
"use_guided_attention_loss": true,
|
90 |
"vocab_size": 81
|
91 |
}
|
generation_config.json
CHANGED
@@ -5,5 +5,5 @@
|
|
5 |
"eos_token_id": 2,
|
6 |
"max_length": 450,
|
7 |
"pad_token_id": 1,
|
8 |
-
"transformers_version": "4.
|
9 |
}
|
|
|
5 |
"eos_token_id": 2,
|
6 |
"max_length": 450,
|
7 |
"pad_token_id": 1,
|
8 |
+
"transformers_version": "4.28.0.dev0"
|
9 |
}
|
preprocessor_config.json
CHANGED
@@ -7,16 +7,11 @@
|
|
7 |
"frame_signal_scale": 1.0,
|
8 |
"hop_length": 16,
|
9 |
"mel_floor": 1e-10,
|
10 |
-
"n_fft": 1024,
|
11 |
-
"n_freqs": 513,
|
12 |
"num_mel_bins": 80,
|
13 |
"padding_side": "right",
|
14 |
"padding_value": 0.0,
|
15 |
"processor_class": "SpeechT5Processor",
|
16 |
-
"reduction_factor": 2,
|
17 |
"return_attention_mask": true,
|
18 |
-
"sample_size": 1024,
|
19 |
-
"sample_stride": 256,
|
20 |
"sampling_rate": 16000,
|
21 |
"win_function": "hann_window",
|
22 |
"win_length": 64
|
|
|
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 |
"return_attention_mask": true,
|
|
|
|
|
15 |
"sampling_rate": 16000,
|
16 |
"win_function": "hann_window",
|
17 |
"win_length": 64
|
special_tokens_map.json
CHANGED
@@ -1,6 +1,13 @@
|
|
1 |
{
|
2 |
"bos_token": "<s>",
|
3 |
"eos_token": "</s>",
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
4 |
"pad_token": "<pad>",
|
5 |
"unk_token": "<unk>"
|
6 |
}
|
|
|
1 |
{
|
2 |
"bos_token": "<s>",
|
3 |
"eos_token": "</s>",
|
4 |
+
"mask_token": {
|
5 |
+
"content": "<mask>",
|
6 |
+
"lstrip": true,
|
7 |
+
"normalized": true,
|
8 |
+
"rstrip": false,
|
9 |
+
"single_word": false
|
10 |
+
},
|
11 |
"pad_token": "<pad>",
|
12 |
"unk_token": "<unk>"
|
13 |
}
|