CoolFace
Modelpublic

neuralabs/deepseek_ocr_de

sourceHugging Faceapache-2.0updated 10mo agoView on Hugging Face
5likes29downloads
config.json134 linesDownload Raw Back to root
1{2    "architectures": [3        "DeepseekOCRForCausalLM"4    ],5    "attention_bias": false,6    "attention_dropout": 0.0,7    "auto_map": {8        "AutoConfig": "modeling_deepseekocr.DeepseekOCRConfig",9        "AutoModel": "modeling_deepseekocr.DeepseekOCRForCausalLM"10    },11    "aux_loss_alpha": 0.001,12    "bos_token_id": 0,13    "candidate_resolutions": [14        [15            1024,16            102417        ]18    ],19    "torch_dtype": "float16",20    "eos_token_id": 1,21    "first_k_dense_replace": 1,22    "global_view_pos": "head",23    "head_dim": 128,24    "hidden_act": "silu",25    "hidden_size": 1280,26    "initializer_range": 0.02,27    "intermediate_size": 6848,28    "kv_lora_rank": null,29    "language_config": {30        "architectures": [31            "DeepseekV2ForCausalLM"32        ],33        "auto_map": {34            "AutoConfig": "configuration_deepseekv2.DeepseekV2Config",35            "AutoModel": "modeling_deepseek.DeepseekV2Model",36            "AutoModelForCausalLM": "modeling_deepseek.DeepseekV2ForCausalLM"37        },38        "bos_token_id": 0,39        "eos_token_id": 1,40        "first_k_dense_replace": 1,41        "hidden_size": 1280,42        "intermediate_size": 6848,43        "kv_lora_rank": null,44        "lm_head": true,45        "max_position_embeddings": 8192,46        "moe_intermediate_size": 896,47        "n_group": 1,48        "n_routed_experts": 64,49        "n_shared_experts": 2,50        "num_attention_heads": 10,51        "num_experts_per_tok": 6,52        "num_hidden_layers": 12,53        "num_key_value_heads": 10,54        "q_lora_rank": null,55        "qk_nope_head_dim": 0,56        "qk_rope_head_dim": 0,57        "rm_head": false,58        "topk_group": 1,59        "topk_method": "greedy",60        "torch_dtype": "bfloat16",61        "use_mla": false,62        "v_head_dim": 0,63        "vocab_size": 12928064    },65    "lm_head": true,66    "max_position_embeddings": 8192,67    "mlp_bias": false,68    "model_type": "DeepseekOCR",69    "moe_intermediate_size": 896,70    "n_group": 1,71    "n_routed_experts": 64,72    "n_shared_experts": 2,73    "norm_topk_prob": false,74    "num_attention_heads": 10,75    "num_experts_per_tok": 6,76    "num_hidden_layers": 12,77    "num_key_value_heads": 10,78    "pad_token_id": 2,79    "projector_config": {80        "input_dim": 2048,81        "model_type": "mlp_projector",82        "n_embed": 1280,83        "projector_type": "linear"84    },85    "q_lora_rank": null,86    "qk_nope_head_dim": 0,87    "qk_rope_head_dim": 0,88    "rm_head": false,89    "rms_norm_eps": 1e-06,90    "rope_scaling": null,91    "rope_theta": 10000.0,92    "routed_scaling_factor": 1.0,93    "seq_aux": true,94    "tie_word_embeddings": false,95    "tile_tag": "2D",96    "topk_group": 1,97    "topk_method": "greedy",98    "transformers_version": "4.56.2",99    "unsloth_version": "2025.11.3",100    "use_cache": true,101    "use_mla": false,102    "v_head_dim": 0,103    "vision_config": {104        "image_size": 1024,105        "mlp_ratio": 3.7362,106        "model_name": "deeplip_b_l",107        "model_type": "vision",108        "width": {109            "clip-l-14-224": {110                "heads": 16,111                "image_size": 224,112                "layers": 24,113                "patch_size": 14,114                "width": 1024115            },116            "sam_vit_b": {117                "downsample_channels": [118                    512,119                    1024120                ],121                "global_attn_indexes": [122                    2,123                    5,124                    8,125                    11126                ],127                "heads": 12,128                "layers": 12,129                "width": 768130            }131        }132    },133    "vocab_size": 129280134}