Safetensors
English
qwen2
video-SALMONN-o1 / config.json
BrianatCambridge's picture
Upload folder using huggingface_hub
e157e18 verified
raw
history blame
8.17 kB
{
"_name_or_path": "/mnt/bn/tiktok-mm-4/aiic/public/model/OV-Qwen2-7B-AM9",
"add_time_token": true,
"architectures": [
"LlavaAVQwenForCausalLM"
],
"attention_dropout": 0.0,
"bos_token_id": 151643,
"eos_token_id": 151645,
"hidden_act": "silu",
"hidden_size": 3584,
"image_aspect_ratio": "anyres_max_9",
"image_crop_resolution": null,
"image_grid_pinpoints": [
[
384,
384
],
[
384,
768
],
[
384,
1152
],
[
384,
1536
],
[
768,
768
],
[
384,
1920
],
[
768,
768
],
[
384,
2304
],
[
768,
1152
],
[
384,
2688
],
[
768,
1152
],
[
384,
3072
],
[
768,
1536
],
[
384,
3456
],
[
768,
1536
],
[
1152,
1152
],
[
768,
1920
],
[
1152,
1152
],
[
768,
1920
],
[
1152,
1152
],
[
768,
2304
],
[
1152,
1536
],
[
768,
2304
],
[
1152,
1536
],
[
768,
2688
],
[
1152,
1536
],
[
768,
2688
],
[
1152,
1920
],
[
768,
3072
],
[
1152,
1920
],
[
1536,
1536
],
[
768,
3072
],
[
1152,
1920
],
[
1536,
1536
],
[
768,
3456
],
[
1152,
2304
],
[
1536,
1536
],
[
768,
3456
],
[
1152,
2304
],
[
1536,
1536
],
[
768,
3840
],
[
1152,
2304
],
[
1536,
1920
],
[
768,
3840
],
[
1152,
2688
],
[
1536,
1920
],
[
768,
4224
],
[
1152,
2688
],
[
1536,
1920
],
[
768,
4224
],
[
1152,
2688
],
[
1536,
1920
],
[
768,
4608
],
[
1152,
3072
],
[
1536,
2304
],
[
768,
4608
],
[
1152,
3072
],
[
1536,
2304
],
[
1920,
1920
],
[
768,
4992
],
[
1152,
3072
],
[
1536,
2304
],
[
1920,
1920
],
[
768,
4992
],
[
1152,
3456
],
[
1536,
2304
],
[
1920,
1920
],
[
768,
5376
],
[
1152,
3456
],
[
1536,
2688
],
[
1920,
1920
],
[
768,
5376
],
[
1152,
3456
],
[
1536,
2688
],
[
1920,
1920
],
[
768,
5760
],
[
1152,
3840
],
[
1536,
2688
],
[
1920,
2304
],
[
768,
5760
],
[
1152,
3840
],
[
1536,
2688
],
[
1920,
2304
],
[
768,
6144
],
[
1152,
3840
],
[
1536,
3072
],
[
1920,
2304
],
[
768,
6144
],
[
1152,
4224
],
[
1536,
3072
],
[
1920,
2304
],
[
768,
6528
],
[
1152,
4224
],
[
1536,
3072
],
[
1920,
2304
],
[
768,
6528
],
[
1152,
4224
],
[
1536,
3072
],
[
1920,
2688
],
[
768,
6912
],
[
1152,
4608
],
[
1536,
3456
],
[
1920,
2688
],
[
2304,
2304
]
],
"image_split_resolution": null,
"initializer_range": 0.02,
"intermediate_size": 18944,
"max_position_embeddings": 32768,
"max_window_layers": 28,
"mf_split_init": false,
"mm_hidden_size": 1152,
"mm_patch_merge_type": "spatial_unpad",
"mm_projector_lr": null,
"mm_projector_type": "mlp2x_gelu",
"mm_resampler_type": null,
"mm_tunable_parts": "mm_vision_tower,mm_mlp_adapter,mm_language_model",
"mm_use_im_patch_token": false,
"mm_use_im_start_end": false,
"mm_vision_select_feature": "patch",
"mm_vision_select_layer": -2,
"mm_vision_tower": "google/siglip-so400m-patch14-384",
"mm_vision_tower_lr": 2e-06,
"model_args": {
"add_time_token": true,
"audio_visual": true,
"beats_path": null,
"do_rag": false,
"flash_attn": false,
"fps": 2,
"freeze_backbone": true,
"freeze_beats": true,
"freeze_final_linear": true,
"freeze_speech_QFormer": true,
"freeze_whisper": true,
"from_stage_1_5": false,
"image_processor": "google/siglip-so400m-patch14-384",
"lora_alpha": 256,
"lora_dropout": 0.05,
"lora_enable": true,
"lora_r": 64,
"mf_split_init": false,
"mm_mask_drop_mode": "fixed",
"mm_mask_drop_ratio": 0.25,
"mm_mask_drop_ratio_lower": null,
"mm_mask_drop_ratio_upper": null,
"mm_mask_drop_skip_percentage": 0.0,
"mm_newline_position": "grid",
"mm_patch_merge_type": "spatial_unpad",
"mm_perceiver_depth": 3,
"mm_perceiver_ff_mult": 4,
"mm_perceiver_latents": 32,
"mm_perceiver_pretrained": null,
"mm_pooling_position": "before",
"mm_projector_type": "mlp2x_gelu",
"mm_qformer_depth": 3,
"mm_qformer_latents": 32,
"mm_qformer_pretrained": null,
"mm_resampler_type": "spatial_pool",
"mm_spatial_pool_mode": "max",
"mm_spatial_pool_out_channels": 1152,
"mm_spatial_pool_stride": 2,
"mm_use_im_patch_token": false,
"mm_use_im_start_end": false,
"mm_vision_select_feature": "patch",
"mm_vision_select_layer": -2,
"mm_vlmattention_bert_type": "qformer_pretrain",
"mm_vlmattention_compress_type": null,
"mm_vlmattention_num_query": 32,
"mm_vlmattention_pretrained": null,
"modality_max_length": "None",
"model_name_or_path": "/mnt/bn/tiktok-mm-4/aiic/users/guangzhisun/llava-video/output/models/sft_finevideo_with_caption_qa_reasoning/checkpoint_93000_qwen",
"multi_frame_num": 30,
"multi_frame_projector": false,
"niv_cnn_params": "[(10, 5), (3, 2), (3, 2), (3, 2), (3, 2), (2, 2), (2, 2)]",
"niv_in_channels": 4,
"niv_out_channels": 401,
"num_speech_query_token": 1,
"patchify_video_feature": false,
"pretrain_mm_mlp_adapter": null,
"rag_input_frames": 1,
"rag_topk": 5,
"rag_type": "direct",
"salmonn_path": null,
"second_per_window": 0.2,
"second_stride": 0.2,
"segmentation": -1,
"spt_projector": false,
"tune_mm_mlp_adapter": false,
"tune_mm_vision_resampler": false,
"unfreeze_mm_vision_tower": false,
"use_final_linear": true,
"use_flash_tower": true,
"use_mfcnn": false,
"use_mftrans": false,
"use_niv": false,
"use_speech_Qformer": true,
"version": "qwen_1_5",
"vision_tower": "google/siglip-so400m-patch14-384",
"whisper_lora": false,
"whisper_path": "openai/whisper-large-v3",
"window_level_Qformer": true
},
"model_type": "qwen2",
"multi_frame_num": 30,
"multi_frame_projector": false,
"num_attention_heads": 28,
"num_hidden_layers": 28,
"num_key_value_heads": 4,
"pos_skipping_range": 4096,
"rms_norm_eps": 1e-06,
"rope_scaling": null,
"rope_theta": 1000000.0,
"sliding_window": 131072,
"tie_word_embeddings": false,
"tokenizer_model_max_length": 32768,
"tokenizer_padding_side": "right",
"torch_dtype": "bfloat16",
"transformers_version": "4.39.2",
"use_cache": true,
"use_flash_tower": true,
"use_mfcnn": false,
"use_mftrans": false,
"use_mm_proj": true,
"use_pos_skipping": false,
"use_sliding_window": false,
"vision_tower_pretrained": null,
"vocab_size": 152064
}