Text Generation
Transformers
English
encoder_decoder
code
natural language understanding
machine learning
research
introspection
self-reflection
conversational
Inference Endpoints
Or4cl3-1 commited on
Commit
cd6f29f
1 Parent(s): 7c814dc

Update config.json

Browse files
Files changed (1) hide show
  1. config.json +29 -37
config.json CHANGED
@@ -1,60 +1,52 @@
1
  {
2
- "data_dir": "/path/to/data",
3
  "text_encoder": {
4
- "model_path": "/path/to/text_encoder.h5",
5
- "max_length": 512
6
  },
7
  "image_encoder": {
8
- "model_path": "/path/to/image_encoder.h5",
9
- "input_size": [224, 224, 3]
10
  },
11
  "audio_encoder": {
12
- "model_path": "/path/to/audio_encoder.h5",
13
- "sample_rate": 16000,
14
- "duration": 5.0
15
  },
16
- "hybrid_model": {
17
- "dense_units": [512, 256],
18
- "dropout_rate": 0.2,
19
- "output_units": 1,
20
- "output_activation": "sigmoid"
21
  },
22
- "attention": {
 
 
 
 
 
23
  "traditional_attention": {
24
- "attention_type": "additive"
25
  },
26
  "self_attention": {
27
- "num_heads": 8,
28
- "key_dim": 64
29
  },
30
  "linear_attention": {
31
- "num_heads": 4,
32
- "key_dim": 32
33
  }
34
  },
35
  "belief_desire_intent_tree": {
36
- "max_depth": 5,
37
- "node_units": 128
38
  },
39
- "training": {
40
- "epochs": 10,
41
- "batch_size": 32,
42
- "learning_rate": 0.001
 
 
43
  },
44
  "realtime_learning": {
45
- "update_interval": 100,
46
- "update_batch_size": 16
47
  },
48
  "knowledge_base": {
49
- "embedding_dim": 256,
50
- "max_entities": 100000
51
- },
52
- "explainer": {
53
- "explanation_type": "attention"
54
- },
55
- "iragel": {
56
- "generation_length": 100,
57
- "top_k": 5,
58
- "top_p": 0.9
59
  }
60
  }
 
1
  {
2
+ "model_type": "csumlm",
3
  "text_encoder": {
4
+ "model_name": "bert-base-uncased",
5
+ "pretrained_weights": "path/to/bert-base-uncased"
6
  },
7
  "image_encoder": {
8
+ "model_name": "resnet50",
9
+ "pretrained_weights": "path/to/resnet50"
10
  },
11
  "audio_encoder": {
12
+ "model_name": "wav2vec2-base",
13
+ "pretrained_weights": "path/to/wav2vec2-base"
 
14
  },
15
+ "fusion_layer": {
16
+ "type": "concatenate",
17
+ "output_dim": 1024
 
 
18
  },
19
+ "output_layer": {
20
+ "type": "dense",
21
+ "units": 1,
22
+ "activation": "sigmoid"
23
+ },
24
+ "attention_mechanism": {
25
  "traditional_attention": {
26
+ "type": "bahdanau"
27
  },
28
  "self_attention": {
29
+ "type": "scaled_dot_product"
 
30
  },
31
  "linear_attention": {
32
+ "type": "linear"
 
33
  }
34
  },
35
  "belief_desire_intent_tree": {
36
+ "type": "recursive_neural_network"
 
37
  },
38
+ "iragel": {
39
+ "retrieval_model": "path/to/retrieval_model",
40
+ "generation_model": "path/to/generation_model"
41
+ },
42
+ "explainer": {
43
+ "type": "attention_visualization"
44
  },
45
  "realtime_learning": {
46
+ "type": "online_learning",
47
+ "learning_rate": 0.001
48
  },
49
  "knowledge_base": {
50
+ "type": "knowledge_graph_embeddings"
 
 
 
 
 
 
 
 
 
51
  }
52
  }