Update tiny models for BridgeTowerModel
Browse files- config.json +61 -0
- model.safetensors +3 -0
- tokenizer.json +0 -0
- tokenizer_config.json +7 -0
config.json
ADDED
|
@@ -0,0 +1,61 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"BridgeTowerModel"
|
| 4 |
+
],
|
| 5 |
+
"contrastive_hidden_size": 512,
|
| 6 |
+
"dtype": "float32",
|
| 7 |
+
"hidden_act": "gelu",
|
| 8 |
+
"hidden_size": 64,
|
| 9 |
+
"init_layernorm_from_vision_encoder": false,
|
| 10 |
+
"initializer_factor": 1,
|
| 11 |
+
"intermediate_size": 128,
|
| 12 |
+
"layer_norm_eps": 1e-05,
|
| 13 |
+
"link_tower_type": "add",
|
| 14 |
+
"logit_scale_init_value": 2.6592,
|
| 15 |
+
"model_type": "bridgetower",
|
| 16 |
+
"num_attention_heads": 4,
|
| 17 |
+
"num_hidden_layers": 2,
|
| 18 |
+
"share_cross_modal_transformer_layers": true,
|
| 19 |
+
"share_link_tower_layers": false,
|
| 20 |
+
"text_config": {
|
| 21 |
+
"add_cross_attention": false,
|
| 22 |
+
"attention_probs_dropout_prob": 0.1,
|
| 23 |
+
"bos_token_id": 0,
|
| 24 |
+
"eos_token_id": 2,
|
| 25 |
+
"hidden_act": "gelu",
|
| 26 |
+
"hidden_dropout_prob": 0.1,
|
| 27 |
+
"hidden_size": 64,
|
| 28 |
+
"initializer_factor": 1e-10,
|
| 29 |
+
"intermediate_size": 128,
|
| 30 |
+
"is_decoder": false,
|
| 31 |
+
"layer_norm_eps": 1e-05,
|
| 32 |
+
"max_position_embeddings": 514,
|
| 33 |
+
"model_type": "bridgetower_text_model",
|
| 34 |
+
"num_attention_heads": 4,
|
| 35 |
+
"num_hidden_layers": 2,
|
| 36 |
+
"pad_token_id": 1,
|
| 37 |
+
"tie_word_embeddings": false,
|
| 38 |
+
"type_vocab_size": 1,
|
| 39 |
+
"use_cache": true,
|
| 40 |
+
"vocab_size": 50265
|
| 41 |
+
},
|
| 42 |
+
"tie_word_embeddings": false,
|
| 43 |
+
"transformers_version": "5.16.0.dev0",
|
| 44 |
+
"vision_config": {
|
| 45 |
+
"batch_size": 1,
|
| 46 |
+
"hidden_size": 64,
|
| 47 |
+
"image_size": 64,
|
| 48 |
+
"init_layernorm_from_vision_encoder": false,
|
| 49 |
+
"initializer_factor": 1e-10,
|
| 50 |
+
"is_training": false,
|
| 51 |
+
"layer_norm_eps": 1e-05,
|
| 52 |
+
"model_type": "bridgetower_vision_model",
|
| 53 |
+
"num_channels": 3,
|
| 54 |
+
"num_hidden_layers": 2,
|
| 55 |
+
"num_image_features": 17,
|
| 56 |
+
"patch_size": 16,
|
| 57 |
+
"remove_last_layer": false,
|
| 58 |
+
"share_layernorm": true,
|
| 59 |
+
"stop_gradient": false
|
| 60 |
+
}
|
| 61 |
+
}
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d5492add6fb6634c161976607e9cd422acb7bfee694c845d9c63dde61c3c9bbe
|
| 3 |
+
size 14782104
|
tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"backend": "tokenizers",
|
| 3 |
+
"is_local": true,
|
| 4 |
+
"local_files_only": false,
|
| 5 |
+
"model_max_length": 1000000000000000019884624838656,
|
| 6 |
+
"tokenizer_class": "TokenizersBackend"
|
| 7 |
+
}
|