Hugging Face's logo

trl-internal-testing
/
tiny-Idefics2ForConditionalGeneration

Image-to-Text
Transformers
Safetensors
idefics2
trl
text-generation-inference
Model card Files
xet
Community
qgallouedec HF Staff commited on
Commit
ed05be4
·
verified ·
1 Parent(s): ef1ec3a

Upload Idefics2ForConditionalGeneration

Browse files
Files changed (3) hide show
  1. config.json +27 -6
  2. generation_config.json +1 -1
  3. model.safetensors +2 -2
config.json CHANGED
@@ -5,31 +5,52 @@
5
  "image_token_id": 32001,
6
  "model_type": "idefics2",
7
  "perceiver_config": {
 
 
8
  "hidden_size": 8,
9
- "model_type": "idefics2_perceiver"
 
 
 
 
 
 
 
10
  },
11
  "text_config": {
12
- "head_dim": 2,
 
 
13
  "hidden_size": 8,
 
14
  "intermediate_size": 32,
15
  "max_position_embeddings": 131072,
16
  "model_type": "mistral",
17
  "num_attention_heads": 4,
18
  "num_hidden_layers": 2,
19
  "num_key_value_heads": 2,
20
- "pad_token_id": null,
21
  "rms_norm_eps": 1e-06,
 
 
 
22
  "vocab_size": 32006
23
  },
24
  "tie_word_embeddings": false,
25
  "torch_dtype": "float32",
26
- "transformers_version": "4.47.0.dev0",
27
  "use_cache": true,
28
  "vision_config": {
29
- "hidden_size": 8,
 
 
 
 
30
  "intermediate_size": 32,
 
31
  "model_type": "idefics2_vision",
32
  "num_attention_heads": 4,
33
- "num_hidden_layers": 2
 
 
34
  }
35
  }
 
5
  "image_token_id": 32001,
6
  "model_type": "idefics2",
7
  "perceiver_config": {
8
+ "attention_dropout": 0.0,
9
+ "hidden_act": "silu",
10
  "hidden_size": 8,
11
+ "initializer_range": 0.02,
12
+ "model_type": "idefics2_perceiver",
13
+ "num_key_value_heads": 4,
14
+ "resampler_depth": 3,
15
+ "resampler_head_dim": 96,
16
+ "resampler_n_heads": 16,
17
+ "resampler_n_latents": 64,
18
+ "rms_norm_eps": 1e-06
19
  },
20
  "text_config": {
21
+ "attention_dropout": 0.0,
22
+ "head_dim": null,
23
+ "hidden_act": "silu",
24
  "hidden_size": 8,
25
+ "initializer_range": 0.02,
26
  "intermediate_size": 32,
27
  "max_position_embeddings": 131072,
28
  "model_type": "mistral",
29
  "num_attention_heads": 4,
30
  "num_hidden_layers": 2,
31
  "num_key_value_heads": 2,
 
32
  "rms_norm_eps": 1e-06,
33
+ "rope_theta": 10000.0,
34
+ "sliding_window": 4096,
35
+ "use_cache": true,
36
  "vocab_size": 32006
37
  },
38
  "tie_word_embeddings": false,
39
  "torch_dtype": "float32",
40
+ "transformers_version": "4.54.0.dev0",
41
  "use_cache": true,
42
  "vision_config": {
43
+ "attention_dropout": 0.0,
44
+ "hidden_act": "gelu_pytorch_tanh",
45
+ "hidden_size": 16,
46
+ "image_size": 224,
47
+ "initializer_range": 0.02,
48
  "intermediate_size": 32,
49
+ "layer_norm_eps": 1e-06,
50
  "model_type": "idefics2_vision",
51
  "num_attention_heads": 4,
52
+ "num_channels": 3,
53
+ "num_hidden_layers": 2,
54
+ "patch_size": 32
55
  }
56
  }
generation_config.json CHANGED
@@ -2,5 +2,5 @@
2
  "_from_model_config": true,
3
  "bos_token_id": 1,
4
  "eos_token_id": 2,
5
- "transformers_version": "4.47.0.dev0"
6
  }
 
2
  "_from_model_config": true,
3
  "bos_token_id": 1,
4
  "eos_token_id": 2,
5
+ "transformers_version": "4.54.0.dev0"
6
  }
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:82e251f3847db0322d90733e1f743038b935e0081dbea66bc683f32219e83c36
3
- size 2557856
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:334e2c579065024ec44dcfff381929abff4adc37b1bfe053482d9c3b10366cd3
3
+ size 2670736