MSALab/PerceptionDLM

1{2  "architectures": [3    "PDMLLM"4  ],5  "auto_map": {6    "AutoConfig": "configuration_pdmllm.PDMLLMConfig",7    "AutoModel": "modeling_pdmllm.PDMLLM",8    "AutoModelForCausalLM": "modeling_pdmllm.PDMLLM"9  },10  "downsample_ratio": 0.5,11  "image_size": 512,12  "image_token_id": 126349,13  "kernel_size": [14    16,15    1616  ],17  "language_model_config": {18    "_attn_implementation_autoset": true,19    "_name_or_path": "bitersun/LLaDA-8B-Instruct-HF",20    "add_cross_attention": false,21    "architectures": [22      "LLaDAModelLM"23    ],24    "attention_bias": false,25    "attention_dropout": 0.0,26    "auto_map": {27      "AutoConfig": "configuration_llada.LLaDAConfig",28      "AutoModel": "modeling_llada.LLaDAModelLM",29      "AutoModelForCausalLM": "modeling_llada.LLaDAModelLM"30    },31    "bad_words_ids": null,32    "begin_suppress_tokens": null,33    "bos_token_id": 128000,34    "chunk_size_feed_forward": 0,35    "cross_attention_hidden_size": null,36    "decoder_start_token_id": null,37    "diversity_penalty": 0.0,38    "do_sample": false,39    "early_stopping": false,40    "encoder_no_repeat_ngram_size": 0,41    "eos_token_id": 126081,42    "exponential_decay_length_penalty": null,43    "finetuning_task": null,44    "forced_bos_token_id": null,45    "forced_eos_token_id": null,46    "hidden_act": "silu",47    "hidden_size": 4096,48    "id2label": {49      "0": "LABEL_0",50      "1": "LABEL_1"51    },52    "initializer_range": 0.02,53    "intermediate_size": 12288,54    "is_decoder": false,55    "is_encoder_decoder": false,56    "label2id": {57      "LABEL_0": 0,58      "LABEL_1": 159    },60    "length_penalty": 1.0,61    "max_length": 20,62    "max_position_embeddings": 16384,63    "min_length": 0,64    "model_type": "llada",65    "no_repeat_ngram_size": 0,66    "num_attention_heads": 32,67    "num_beam_groups": 1,68    "num_beams": 1,69    "num_hidden_layers": 32,70    "num_key_value_heads": 32,71    "num_return_sequences": 1,72    "output_attentions": false,73    "output_hidden_states": false,74    "output_scores": false,75    "pad_token_id": null,76    "prefix": null,77    "pretraining_tp": 1,78    "problem_type": null,79    "pruned_heads": {},80    "remove_invalid_values": false,81    "repetition_penalty": 1.0,82    "return_dict": true,83    "return_dict_in_generate": false,84    "rms_norm_eps": 1e-05,85    "rope_scaling": null,86    "rope_theta": 500000.0,87    "sep_token_id": null,88    "suppress_tokens": null,89    "task_specific_params": null,90    "temperature": 1.0,91    "tf_legacy_loss": false,92    "tie_encoder_decoder": false,93    "tie_word_embeddings": false,94    "tokenizer_class": null,95    "top_k": 50,96    "top_p": 1.0,97    "torch_dtype": "bfloat16",98    "torchscript": false,99    "typical_p": 1.0,100    "use_bfloat16": false,101    "use_cache": false,102    "vocab_size": 126464103  },104  "mask_patch_embedding_in_channels": 3,105  "mask_patch_embedding_out_channels": 1152,106  "model_type": "pdmllm",107  "num_image_token": 256,108  "patch_size": 16,109  "prompt_numbers": 6,110  "replacement_noise_mode": false,111  "roi_output_size": 4,112  "torch_dtype": "bfloat16",113  "transformers_version": "4.51.3",114  "vision_abstractor_config": {115    "projection_type": "mlp2x_gelu"116  },117  "vision_model_config": {118    "_attn_implementation_autoset": true,119    "_name_or_path": "google/siglip2-so400m-patch16-512",120    "add_cross_attention": false,121    "architectures": null,122    "bad_words_ids": null,123    "begin_suppress_tokens": null,124    "bos_token_id": null,125    "chunk_size_feed_forward": 0,126    "cross_attention_hidden_size": null,127    "decoder_start_token_id": null,128    "diversity_penalty": 0.0,129    "do_sample": false,130    "early_stopping": false,131    "encoder_no_repeat_ngram_size": 0,132    "eos_token_id": null,133    "exponential_decay_length_penalty": null,134    "finetuning_task": null,135    "forced_bos_token_id": null,136    "forced_eos_token_id": null,137    "hidden_size": 1152,138    "id2label": {139      "0": "LABEL_0",140      "1": "LABEL_1"141    },142    "initializer_factor": 1.0,143    "is_decoder": false,144    "is_encoder_decoder": false,145    "label2id": {146      "LABEL_0": 0,147      "LABEL_1": 1148    },149    "length_penalty": 1.0,150    "max_length": 20,151    "min_length": 0,152    "model_type": "siglip",153    "no_repeat_ngram_size": 0,154    "num_beam_groups": 1,155    "num_beams": 1,156    "num_return_sequences": 1,157    "output_attentions": false,158    "output_hidden_states": false,159    "output_scores": false,160    "pad_token_id": null,161    "prefix": null,162    "problem_type": null,163    "pruned_heads": {},164    "remove_invalid_values": false,165    "repetition_penalty": 1.0,166    "return_dict": true,167    "return_dict_in_generate": false,168    "sep_token_id": null,169    "suppress_tokens": null,170    "task_specific_params": null,171    "temperature": 1.0,172    "text_config": {173      "_attn_implementation_autoset": false,174      "_name_or_path": "",175      "add_cross_attention": false,176      "architectures": null,177      "attention_dropout": 0.0,178      "bad_words_ids": null,179      "begin_suppress_tokens": null,180      "bos_token_id": 49406,181      "chunk_size_feed_forward": 0,182      "cross_attention_hidden_size": null,183      "decoder_start_token_id": null,184      "diversity_penalty": 0.0,185      "do_sample": false,186      "early_stopping": false,187      "encoder_no_repeat_ngram_size": 0,188      "eos_token_id": 49407,189      "exponential_decay_length_penalty": null,190      "finetuning_task": null,191      "forced_bos_token_id": null,192      "forced_eos_token_id": null,193      "hidden_act": "gelu_pytorch_tanh",194      "hidden_size": 1152,195      "id2label": {196        "0": "LABEL_0",197        "1": "LABEL_1"198      },199      "intermediate_size": 4304,200      "is_decoder": false,201      "is_encoder_decoder": false,202      "label2id": {203        "LABEL_0": 0,204        "LABEL_1": 1205      },206      "layer_norm_eps": 1e-06,207      "length_penalty": 1.0,208      "max_length": 20,209      "max_position_embeddings": 64,210      "min_length": 0,211      "model_type": "siglip_text_model",212      "no_repeat_ngram_size": 0,213      "num_attention_heads": 16,214      "num_beam_groups": 1,215      "num_beams": 1,216      "num_hidden_layers": 27,217      "num_return_sequences": 1,218      "output_attentions": false,219      "output_hidden_states": false,220      "output_scores": false,221      "pad_token_id": 1,222      "prefix": null,223      "problem_type": null,224      "projection_size": 1152,225      "pruned_heads": {},226      "remove_invalid_values": false,227      "repetition_penalty": 1.0,228      "return_dict": true,229      "return_dict_in_generate": false,230      "sep_token_id": null,231      "suppress_tokens": null,232      "task_specific_params": null,233      "temperature": 1.0,234      "tf_legacy_loss": false,235      "tie_encoder_decoder": false,236      "tie_word_embeddings": true,237      "tokenizer_class": null,238      "top_k": 50,239      "top_p": 1.0,240      "torch_dtype": "bfloat16",241      "torchscript": false,242      "typical_p": 1.0,243      "use_bfloat16": false,244      "vocab_size": 256000245    },246    "tf_legacy_loss": false,247    "tie_encoder_decoder": false,248    "tie_word_embeddings": true,249    "tokenizer_class": null,250    "top_k": 50,251    "top_p": 1.0,252    "torch_dtype": "bfloat16",253    "torchscript": false,254    "typical_p": 1.0,255    "use_bfloat16": false,256    "vision_config": {257      "_attn_implementation_autoset": false,258      "_name_or_path": "",259      "add_cross_attention": false,260      "architectures": null,261      "attention_dropout": 0.0,262      "bad_words_ids": null,263      "begin_suppress_tokens": null,264      "bos_token_id": null,265      "chunk_size_feed_forward": 0,266      "cross_attention_hidden_size": null,267      "decoder_start_token_id": null,268      "diversity_penalty": 0.0,269      "do_sample": false,270      "early_stopping": false,271      "encoder_no_repeat_ngram_size": 0,272      "eos_token_id": null,273      "exponential_decay_length_penalty": null,274      "finetuning_task": null,275      "forced_bos_token_id": null,276      "forced_eos_token_id": null,277      "hidden_act": "gelu_pytorch_tanh",278      "hidden_size": 1152,279      "id2label": {280        "0": "LABEL_0",281        "1": "LABEL_1"282      },283      "image_size": 512,284      "intermediate_size": 4304,285      "is_decoder": false,286      "is_encoder_decoder": false,287      "label2id": {288        "LABEL_0": 0,289        "LABEL_1": 1290      },291      "layer_norm_eps": 1e-06,292      "length_penalty": 1.0,293      "max_length": 20,294      "min_length": 0,295      "model_type": "siglip_vision_model",296      "no_repeat_ngram_size": 0,297      "num_attention_heads": 16,298      "num_beam_groups": 1,299      "num_beams": 1,300      "num_channels": 3,301      "num_hidden_layers": 27,302      "num_return_sequences": 1,303      "output_attentions": false,304      "output_hidden_states": false,305      "output_scores": false,306      "pad_token_id": null,307      "patch_size": 16,308      "prefix": null,309      "problem_type": null,310      "pruned_heads": {},311      "remove_invalid_values": false,312      "repetition_penalty": 1.0,313      "return_dict": true,314      "return_dict_in_generate": false,315      "sep_token_id": null,316      "suppress_tokens": null,317      "task_specific_params": null,318      "temperature": 1.0,319      "tf_legacy_loss": false,320      "tie_encoder_decoder": false,321      "tie_word_embeddings": true,322      "tokenizer_class": null,323      "top_k": 50,324      "top_p": 1.0,325      "torch_dtype": "bfloat16",326      "torchscript": false,327      "typical_p": 1.0,328      "use_bfloat16": false329    }330  },331  "vision_output_key": null,332  "vision_select_layer": -2333}334