Cambrian-S-3B / config.json
ShushengYang's picture
Upload folder using huggingface_hub
5006d23 verified
{
"_name_or_path": "Qwen/Qwen2.5-3B-Instruct",
"anyres_max_subimages": 9,
"architectures": [
"CambrianQwenForCausalLM"
],
"attention_dropout": 0.0,
"bos_token_id": 151643,
"connector_only": true,
"do_sample": true,
"eos_token_id": 151645,
"freeze_mm_mlp_adapter": false,
"hidden_act": "silu",
"hidden_size": 2048,
"image_aspect_ratio": "anyres",
"image_position": 14,
"initializer_range": 0.02,
"intermediate_size": 11008,
"max_images_per_sample": 128,
"max_position_embeddings": 32768,
"max_window_layers": 70,
"miv_token_len": 64,
"mm_hidden_size": 1152,
"mm_projector_lr": null,
"mm_projector_type": "mlp2x_gelu",
"mm_use_im_newline_token": true,
"mm_use_im_patch_token": false,
"mm_use_im_start_end": false,
"mm_vision_sampler_lr": null,
"mm_vision_select_feature": "patch",
"mm_vision_select_layer": -2,
"mm_vision_tower_aux_list": [
"google/siglip2-so400m-patch14-384"
],
"mm_vision_tower_aux_token_len_list": [
729
],
"mm_vision_tower_lr": null,
"model_type": "cambrian_qwen",
"num_attention_heads": 16,
"num_hidden_layers": 36,
"num_key_value_heads": 2,
"rms_norm_eps": 1e-06,
"rope_scaling": null,
"rope_theta": 1000000.0,
"si_token_len": 729,
"sliding_window": 32768,
"tie_word_embeddings": true,
"tokenizer_model_max_length": 11264,
"tokenizer_padding_side": "right",
"torch_dtype": "float32",
"transformers_version": "4.39.2",
"tune_mm_mlp_adapter": false,
"unfreeze_mm_vision_tower": false,
"use_cache": true,
"use_mm_proj": true,
"use_sliding_window": false,
"video_max_frames": 128,
"vision_hidden_size": 1152,
"vision_tower_aux_token_len_list": [
729
],
"vocab_size": 151936
}