Summarization
Transformers
PyTorch
English
pegasus_x
text2text-generation
Inference Endpoints
Joemgu commited on
Commit
687ffc1
1 Parent(s): f50c38c

Update config.json

Browse files
Files changed (1) hide show
  1. config.json +17 -5
config.json CHANGED
@@ -1,13 +1,13 @@
1
  {
2
  "_name_or_path": "/app/models/pegasus_run_en/checkpoint-400",
3
- "activation_dropout": 0.1,
4
  "activation_function": "relu",
5
  "add_bias_logits": false,
6
  "add_final_layer_norm": true,
7
  "architectures": [
8
  "PegasusXForConditionalGeneration"
9
  ],
10
- "attention_dropout": 0.1,
11
  "block_size": 512,
12
  "bos_token_id": 0,
13
  "classif_dropout": 0.0,
@@ -18,8 +18,8 @@
18
  "decoder_layerdrop": 0.0,
19
  "decoder_layers": 16,
20
  "decoder_start_token_id": 0,
21
- "dropout": 0.1,
22
- "dropout_rate": 0.1,
23
  "encoder_attention_heads": 16,
24
  "encoder_ffn_dim": 4096,
25
  "encoder_layerdrop": 0.0,
@@ -56,6 +56,18 @@
56
  "static_position_embeddings": true,
57
  "torch_dtype": "float32",
58
  "transformers_version": "4.28.1",
59
- "use_cache": false,
 
 
 
 
 
 
 
 
 
 
 
 
60
  "vocab_size": 96103
61
  }
 
1
  {
2
  "_name_or_path": "/app/models/pegasus_run_en/checkpoint-400",
3
+ "activation_dropout": 0.0,
4
  "activation_function": "relu",
5
  "add_bias_logits": false,
6
  "add_final_layer_norm": true,
7
  "architectures": [
8
  "PegasusXForConditionalGeneration"
9
  ],
10
+ "attention_dropout": 0.0,
11
  "block_size": 512,
12
  "bos_token_id": 0,
13
  "classif_dropout": 0.0,
 
18
  "decoder_layerdrop": 0.0,
19
  "decoder_layers": 16,
20
  "decoder_start_token_id": 0,
21
+ "dropout": 0.0,
22
+ "dropout_rate": 0.0,
23
  "encoder_attention_heads": 16,
24
  "encoder_ffn_dim": 4096,
25
  "encoder_layerdrop": 0.0,
 
56
  "static_position_embeddings": true,
57
  "torch_dtype": "float32",
58
  "transformers_version": "4.28.1",
59
+ "task_specific_params": {
60
+ "summarization": {
61
+ "early_stopping": true,
62
+ "length_penalty": 2.0,
63
+ "repetition_penalty": 3.5,
64
+ "max_length": 1024,
65
+ "min_length": 8,
66
+ "encoder_no_repeat_ngram_size": 4,
67
+ "no_repeat_ngram_size": 3,
68
+ "num_beams": 8
69
+ }
70
+ },
71
+ "use_cache": true,
72
  "vocab_size": 96103
73
  }