reset
Browse files- config.json +11 -10
- pytorch_model.bin +2 -2
- run_speech_recognition_seq2seq.py +0 -1
config.json
CHANGED
@@ -3,7 +3,7 @@
|
|
3 |
"SpeechEncoderDecoderModel"
|
4 |
],
|
5 |
"decoder": {
|
6 |
-
"_name_or_path": "facebook/bart-
|
7 |
"activation_dropout": 0.1,
|
8 |
"activation_function": "gelu",
|
9 |
"add_bias_logits": false,
|
@@ -20,20 +20,20 @@
|
|
20 |
"classif_dropout": 0.1,
|
21 |
"classifier_dropout": 0.0,
|
22 |
"cross_attention_hidden_size": null,
|
23 |
-
"d_model":
|
24 |
-
"decoder_attention_heads":
|
25 |
-
"decoder_ffn_dim":
|
26 |
"decoder_layerdrop": 0.0,
|
27 |
-
"decoder_layers":
|
28 |
"decoder_start_token_id": 2,
|
29 |
"diversity_penalty": 0.0,
|
30 |
"do_sample": false,
|
31 |
"dropout": 0.1,
|
32 |
"early_stopping": true,
|
33 |
-
"encoder_attention_heads":
|
34 |
-
"encoder_ffn_dim":
|
35 |
"encoder_layerdrop": 0.0,
|
36 |
-
"encoder_layers":
|
37 |
"encoder_no_repeat_ngram_size": 0,
|
38 |
"eos_token_id": 2,
|
39 |
"exponential_decay_length_penalty": null,
|
@@ -61,9 +61,10 @@
|
|
61 |
"model_type": "bart",
|
62 |
"no_repeat_ngram_size": 3,
|
63 |
"normalize_before": false,
|
|
|
64 |
"num_beam_groups": 1,
|
65 |
"num_beams": 4,
|
66 |
-
"num_hidden_layers":
|
67 |
"num_return_sequences": 1,
|
68 |
"output_attentions": false,
|
69 |
"output_hidden_states": false,
|
@@ -106,7 +107,7 @@
|
|
106 |
"tokenizer_class": null,
|
107 |
"top_k": 50,
|
108 |
"top_p": 1.0,
|
109 |
-
"torch_dtype":
|
110 |
"torchscript": false,
|
111 |
"typical_p": 1.0,
|
112 |
"use_bfloat16": false,
|
|
|
3 |
"SpeechEncoderDecoderModel"
|
4 |
],
|
5 |
"decoder": {
|
6 |
+
"_name_or_path": "facebook/bart-base",
|
7 |
"activation_dropout": 0.1,
|
8 |
"activation_function": "gelu",
|
9 |
"add_bias_logits": false,
|
|
|
20 |
"classif_dropout": 0.1,
|
21 |
"classifier_dropout": 0.0,
|
22 |
"cross_attention_hidden_size": null,
|
23 |
+
"d_model": 768,
|
24 |
+
"decoder_attention_heads": 12,
|
25 |
+
"decoder_ffn_dim": 3072,
|
26 |
"decoder_layerdrop": 0.0,
|
27 |
+
"decoder_layers": 6,
|
28 |
"decoder_start_token_id": 2,
|
29 |
"diversity_penalty": 0.0,
|
30 |
"do_sample": false,
|
31 |
"dropout": 0.1,
|
32 |
"early_stopping": true,
|
33 |
+
"encoder_attention_heads": 12,
|
34 |
+
"encoder_ffn_dim": 3072,
|
35 |
"encoder_layerdrop": 0.0,
|
36 |
+
"encoder_layers": 6,
|
37 |
"encoder_no_repeat_ngram_size": 0,
|
38 |
"eos_token_id": 2,
|
39 |
"exponential_decay_length_penalty": null,
|
|
|
61 |
"model_type": "bart",
|
62 |
"no_repeat_ngram_size": 3,
|
63 |
"normalize_before": false,
|
64 |
+
"normalize_embedding": true,
|
65 |
"num_beam_groups": 1,
|
66 |
"num_beams": 4,
|
67 |
+
"num_hidden_layers": 6,
|
68 |
"num_return_sequences": 1,
|
69 |
"output_attentions": false,
|
70 |
"output_hidden_states": false,
|
|
|
107 |
"tokenizer_class": null,
|
108 |
"top_k": 50,
|
109 |
"top_p": 1.0,
|
110 |
+
"torch_dtype": "float32",
|
111 |
"torchscript": false,
|
112 |
"typical_p": 1.0,
|
113 |
"use_bfloat16": false,
|
pytorch_model.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:ca8fd76aa7ed61545e51c6e8f239069cefec1e6c185d200b69b3e7c71bdb84df
|
3 |
+
size 1649512941
|
run_speech_recognition_seq2seq.py
DELETED
@@ -1 +0,0 @@
|
|
1 |
-
/mnt/lscratch/users/gmenon/transformers/examples/pytorch/speech-recognition/run_speech_recognition_seq2seq.py
|
|
|
|