Update tiny models for Kosmos2_5Model
Browse files- config.json +52 -0
- model.safetensors +3 -0
- preprocessor_config.json +18 -0
- tokenizer.json +0 -0
- tokenizer_config.json +21 -0
config.json
ADDED
|
@@ -0,0 +1,52 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"Kosmos2_5Model"
|
| 4 |
+
],
|
| 5 |
+
"dtype": "float32",
|
| 6 |
+
"latent_query_num": 3,
|
| 7 |
+
"model_type": "kosmos-2.5",
|
| 8 |
+
"text_config": {
|
| 9 |
+
"activation_dropout": 0.0,
|
| 10 |
+
"activation_function": "gelu",
|
| 11 |
+
"attention_dropout": 0.0,
|
| 12 |
+
"attention_heads": 4,
|
| 13 |
+
"bos_token_id": 0,
|
| 14 |
+
"dropout": 0.0,
|
| 15 |
+
"embed_dim": 32,
|
| 16 |
+
"eos_token_id": 2,
|
| 17 |
+
"ffn_dim": 64,
|
| 18 |
+
"init_std": 0.02,
|
| 19 |
+
"layer_norm_eps": 1e-05,
|
| 20 |
+
"layerdrop": 0.0,
|
| 21 |
+
"layers": 2,
|
| 22 |
+
"max_position_embeddings": 512,
|
| 23 |
+
"model_type": "kosmos_2_5_text_model",
|
| 24 |
+
"pad_token_id": 1,
|
| 25 |
+
"scale_embedding": true,
|
| 26 |
+
"tie_word_embeddings": true,
|
| 27 |
+
"use_cache": true,
|
| 28 |
+
"vocab_size": 108464
|
| 29 |
+
},
|
| 30 |
+
"tie_word_embeddings": true,
|
| 31 |
+
"transformers_version": "5.16.0.dev0",
|
| 32 |
+
"vision_config": {
|
| 33 |
+
"attention_dropout": 0.0,
|
| 34 |
+
"dense_act_fn": "gelu_new",
|
| 35 |
+
"dropout": 0.0,
|
| 36 |
+
"dropout_rate": 0.0,
|
| 37 |
+
"head_dim": 64,
|
| 38 |
+
"hidden_size": 32,
|
| 39 |
+
"image_size": 32,
|
| 40 |
+
"initializer_factor": 1.0,
|
| 41 |
+
"initializer_range": 0.02,
|
| 42 |
+
"intermediate_size": 64,
|
| 43 |
+
"layer_norm_eps": 1e-06,
|
| 44 |
+
"max_num_patches": 4096,
|
| 45 |
+
"model_type": "kosmos_2_5_vision_model",
|
| 46 |
+
"num_attention_heads": 4,
|
| 47 |
+
"num_channels": 3,
|
| 48 |
+
"num_hidden_layers": 2,
|
| 49 |
+
"patch_embed_hidden_size": 48,
|
| 50 |
+
"patch_size": 4
|
| 51 |
+
}
|
| 52 |
+
}
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:305ab440664709a24ff8f80589531027e5116cc2a50232c6096204dd7f453428
|
| 3 |
+
size 15350000
|
preprocessor_config.json
ADDED
|
@@ -0,0 +1,18 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"crop_size": {
|
| 3 |
+
"height": 32,
|
| 4 |
+
"width": 32
|
| 5 |
+
},
|
| 6 |
+
"do_convert_rgb": true,
|
| 7 |
+
"do_normalize": true,
|
| 8 |
+
"image_processor_type": "Kosmos2_5ImageProcessor",
|
| 9 |
+
"max_patches": 4096,
|
| 10 |
+
"patch_size": {
|
| 11 |
+
"height": 16,
|
| 12 |
+
"width": 16
|
| 13 |
+
},
|
| 14 |
+
"size": {
|
| 15 |
+
"height": 32,
|
| 16 |
+
"width": 32
|
| 17 |
+
}
|
| 18 |
+
}
|
tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1,21 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_prefix_space": false,
|
| 3 |
+
"backend": "tokenizers",
|
| 4 |
+
"boi_token": "<image>",
|
| 5 |
+
"bos_token": "<s>",
|
| 6 |
+
"clean_up_tokenization_spaces": false,
|
| 7 |
+
"eoi_token": "</image>",
|
| 8 |
+
"eos_token": "</s>",
|
| 9 |
+
"image_token": "<s>",
|
| 10 |
+
"is_local": true,
|
| 11 |
+
"local_files_only": false,
|
| 12 |
+
"model_max_length": 4096,
|
| 13 |
+
"model_specific_special_tokens": {
|
| 14 |
+
"boi_token": "<image>",
|
| 15 |
+
"eoi_token": "</image>",
|
| 16 |
+
"image_token": "<s>"
|
| 17 |
+
},
|
| 18 |
+
"pad_token": "<pad>",
|
| 19 |
+
"tokenizer_class": "TokenizersBackend",
|
| 20 |
+
"unk_token": "<unk>"
|
| 21 |
+
}
|