GuilhermeNunes commited on
Commit
a8eba42
·
verified ·
1 Parent(s): 7436d0e

Upload LlavaNextForConditionalGeneration

Browse files
config.json CHANGED
@@ -66,7 +66,7 @@
66
  "multimodal_projector_bias": true,
67
  "projector_hidden_act": "gelu",
68
  "text_config": {
69
- "_name_or_path": "/mnt/scratch-artemis/gviveiros/TowerVision/llava-next-native/towerp_2b_instruct/",
70
  "add_faster_video": false,
71
  "add_time_instruction": false,
72
  "architectures": [
@@ -76,12 +76,10 @@
76
  "attention_dropout": 0.0,
77
  "attn_logit_softcapping": 50.0,
78
  "cache_implementation": "hybrid",
79
- "eos_token_id": 107,
80
  "faster_token_stride": 10,
81
  "final_logit_softcapping": 30.0,
82
  "force_sample": false,
83
  "head_dim": 256,
84
- "hidden_act": "gelu_pytorch_tanh",
85
  "hidden_activation": "gelu_pytorch_tanh",
86
  "hidden_size": 2304,
87
  "image_aspect_ratio": "anyres",
@@ -147,7 +145,7 @@
147
  "image_split_resolution": null,
148
  "initializer_range": 0.02,
149
  "intermediate_size": 9216,
150
- "max_position_embeddings": 8192,
151
  "mm_hidden_size": 1152,
152
  "mm_newline_position": "grid",
153
  "mm_patch_merge_type": "spatial_unpad",
@@ -174,7 +172,7 @@
174
  "sliding_window": 4096,
175
  "tokenizer_model_max_length": 8192,
176
  "tokenizer_padding_side": "right",
177
- "torch_dtype": "float32",
178
  "use_cache": true,
179
  "use_mm_proj": true,
180
  "use_pos_skipping": false,
@@ -182,7 +180,7 @@
182
  "vocab_size": 256001
183
  },
184
  "tie_word_embeddings": false,
185
- "torch_dtype": "float32",
186
  "transformers_version": "4.52.0.dev0",
187
  "use_image_newline_parameter": true,
188
  "vision_config": {
@@ -197,7 +195,7 @@
197
  "num_channels": 3,
198
  "num_hidden_layers": 26,
199
  "patch_size": 14,
200
- "torch_dtype": "float32",
201
  "vision_use_head": false
202
  },
203
  "vision_feature_layer": -1,
 
66
  "multimodal_projector_bias": true,
67
  "projector_hidden_act": "gelu",
68
  "text_config": {
69
+ "_name_or_path": "/mnt/scratch-artemis/gviveiros/TowerVision/llava-next-native/towerp_2b_base/",
70
  "add_faster_video": false,
71
  "add_time_instruction": false,
72
  "architectures": [
 
76
  "attention_dropout": 0.0,
77
  "attn_logit_softcapping": 50.0,
78
  "cache_implementation": "hybrid",
 
79
  "faster_token_stride": 10,
80
  "final_logit_softcapping": 30.0,
81
  "force_sample": false,
82
  "head_dim": 256,
 
83
  "hidden_activation": "gelu_pytorch_tanh",
84
  "hidden_size": 2304,
85
  "image_aspect_ratio": "anyres",
 
145
  "image_split_resolution": null,
146
  "initializer_range": 0.02,
147
  "intermediate_size": 9216,
148
+ "max_position_embeddings": 4096,
149
  "mm_hidden_size": 1152,
150
  "mm_newline_position": "grid",
151
  "mm_patch_merge_type": "spatial_unpad",
 
172
  "sliding_window": 4096,
173
  "tokenizer_model_max_length": 8192,
174
  "tokenizer_padding_side": "right",
175
+ "torch_dtype": "float16",
176
  "use_cache": true,
177
  "use_mm_proj": true,
178
  "use_pos_skipping": false,
 
180
  "vocab_size": 256001
181
  },
182
  "tie_word_embeddings": false,
183
+ "torch_dtype": "float16",
184
  "transformers_version": "4.52.0.dev0",
185
  "use_image_newline_parameter": true,
186
  "vision_config": {
 
195
  "num_channels": 3,
196
  "num_hidden_layers": 26,
197
  "patch_size": 14,
198
+ "torch_dtype": "float16",
199
  "vision_use_head": false
200
  },
201
  "vision_feature_layer": -1,
generation_config.json CHANGED
@@ -2,7 +2,7 @@
2
  "_from_model_config": true,
3
  "bos_token_id": 2,
4
  "cache_implementation": "hybrid",
5
- "eos_token_id": 107,
6
  "pad_token_id": 0,
7
  "transformers_version": "4.52.0.dev0"
8
  }
 
2
  "_from_model_config": true,
3
  "bos_token_id": 2,
4
  "cache_implementation": "hybrid",
5
+ "eos_token_id": 1,
6
  "pad_token_id": 0,
7
  "transformers_version": "4.52.0.dev0"
8
  }
model-00001-of-00002.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a337ac8a8db3caab6b1b452762056b731a5a057f3121ab0928244e1d08722246
3
+ size 4978392440
model-00002-of-00002.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:aadce153b8abd872f2e01d65324a98621db80b32209c58df8a7c5ed8cb66de98
3
+ size 1061826432
model.safetensors.index.json CHANGED
The diff for this file is too large to render. See raw diff