Upload folder using huggingface_hub

config.json

@@ -5,7 +5,7 @@
     "Video",
     "Image"
   ],
-  "_name_or_path": "./checkpoints/OmniFusion-8B",
+  "_name_or_path": "/mnt/bn/tns-algo-video-public-my2/wangpeng.an/model/Meta-Llama-3.1-8B-Instruct",
   "architectures": [
     "LlavaLlamaForCausalLM"
   ],
@@ -24,16 +24,17 @@
   "image_grid_pinpoints": null,
   "initializer_range": 0.02,
   "intermediate_size": 14336,
+  "is_fusion": false,
   "max_position_embeddings": 131072,
   "mlp_bias": false,
-  "mm_audio_caption_tower": "LanguageBind/LanguageBind_Audio",
-  "mm_audio_tower": "/mnt/bn/algo-masp-nas-2/multimodal/multimodal/LanguageBind_Audio_Asr",
+  "mm_audio_caption_tower": "/mnt/bn/tns-algo-video-public-my2/wangpeng.an/model/pretrained_model/LanguageBind_Audio",
+  "mm_audio_tower": "/mnt/bn/tns-algo-video-public-my2/wangpeng.an/model/LanguageBind_Audio_Asr",
   "mm_hidden_size": 1024,
-  "mm_image_tower": "LanguageBind/LanguageBind_Image",
+  "mm_image_tower": "/mnt/bn/tns-algo-video-public-my2/wangpeng.an/model/pretrained_model/LanguageBind_Image",
   "mm_projector_type": "mlp2x_gelu",
   "mm_use_x_patch_token": false,
   "mm_use_x_start_end": false,
-  "mm_video_tower": "LanguageBind/LanguageBind_Video_merge",
+  "mm_video_tower": "/mnt/bn/tns-algo-video-public-my2/wangpeng.an/model/pretrained_model/LanguageBind_Video_merge",
   "mm_vision_select_feature": "patch",
   "mm_vision_select_layer": -2,
   "model_type": "llava_llama",
@@ -51,9 +52,9 @@
   },
   "rope_theta": 500000.0,
   "tie_word_embeddings": false,
-  "tokenizer_model_max_length": 3072,
+  "tokenizer_model_max_length": 32768,
   "torch_dtype": "bfloat16",
-  "transformers_version": "4.44.2",
+  "transformers_version": "4.43.1",
   "tune_mm_mlp_adapter": false,
   "use_cache": true,
   "use_mm_proj": true,

generation_config.json

@@ -8,5 +8,5 @@
   ],
   "temperature": 0.6,
   "top_p": 0.9,
-  "transformers_version": "4.44.2"
+  "transformers_version": "4.43.1"
 }

model-00001-of-00004.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c5cee007928cb055adf18279f82822ce847ea3b40c9bccc7248cc5ffc639a294
+oid sha256:5205f56e1ec50d58b2f223f31bc6af8f02c7b087f93eb7244797a181208a3a5e
 size 4976698672

model-00002-of-00004.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8f2f432671240849b970a7d52936cc2d6ab4bc7a4d39cda583a2c363b7c6b818
+oid sha256:67ac5e3e6f723059947dcd16f659f3f7a22cb0b0c82786cf6a21c5e5b5c1958a
 size 4999802720

model-00003-of-00004.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:413b47cf86c1a751cd2498ddd89dd0c43e94f7be545abf4b5e50216e803769ab
+oid sha256:4c836a764885acdec63a321cefd5af936311e905af53c1bf37e09f3f6cae6334
 size 4915916176

model-00004-of-00004.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ae90cda28ab614512056e2ae2860ba0471a9ab8f613b571b713cbab37e8e1f4a
+oid sha256:5721641119b930b11809b98d3fd13579453ca5da051842931a0dad04623c147f
 size 3851682320

tokenizer_config.json

@@ -2057,7 +2057,7 @@
     "input_ids",
     "attention_mask"
   ],
-  "model_max_length": 2048,
+  "model_max_length": 32768,
   "pad_token": "<|finetune_right_pad_id|>",
   "padding_side": "right",
   "tokenizer_class": "PreTrainedTokenizerFast"

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6a30b64ecb2fb872dad2460933c188c222dbafd5b7b87c581e5390dd01e6c4f4
-size 6840
+oid sha256:b0de789be30021d2f653e953a3dd52c0ab7a9bbaf31abd714d16a3eed37f4882
+size 6776