{ "params": { "decoder_param": "decoder.ncnn.param", "decoder_bin": "decoder.ncnn.bin", "embed_token_param": "embed.ncnn.param", "embed_token_bin": "embed.ncnn.bin", "proj_out_param": "lm_head.ncnn.param", "proj_out_bin": "lm_head.ncnn.bin" }, "tokenizer": { "type": "bbpe", "vocab_file": "vocab.txt", "merges_file": "merges.txt", "eos": "<|im_end|>", "bos": "<|im_start|>", "additional_special_tokens": [ "<|im_start|>", "<|im_end|>", "<|image_pad|>", "<|vision_start|>", "<|vision_end|>", "<|vision_pad|>", "<|video_pad|>", "", "", "", "", "", "", "<|object_ref_start|>", "<|object_ref_end|>", "<|box_start|>", "<|box_end|>", "<|quad_start|>", "<|quad_end|>", "<|audio|>", "<|audio_start|>", "<|audio_end|>", "<|endoftext|>", "<|file_sep|>", "<|repo_name|>", "<|stream_start|>", "<|stream_end|>", "<|fim_prefix|>", "<|fim_middle|>", "<|fim_suffix|>", "<|fim_pad|>", "<|image|>" ] }, "setting": { "attn_cnt": 28, "sconv_cnt": 0, "gdr_cnt": 0, "rope": { "type": "RoPE", "rope_head_dim": 128, "rope_theta": 1000000.0 }, "vision": { "type": "vit", "vision_embed_patch_param": "vision_embed_patch.ncnn.param", "vision_embed_patch_bin": "vision_embed_patch.ncnn.bin", "vision_encoder_param": "vision_encoder.ncnn.param", "vision_encoder_bin": "vision_encoder.ncnn.bin", "vision_projector_param": "vision_projector.ncnn.param", "vision_projector_bin": "vision_projector.ncnn.bin", "patch_size": 14, "patch_dim": 1024, "max_num_patches": 16384, "spatial_merge_size": 1, "rope": { "type": "mRoPE", "mrope_section": [0, 32, 32] } } } }