{ "apply_patch_attention_mask": false, "architectures": [ "Yasa2ForConditionalGeneration" ], "auto_map": { "AutoConfig": "configuration_yasa2.Yasa2Config", "AutoModel": "modeling_yasa2.Yasa2Model", "AutoModelForImageTextToText": "modeling_yasa2.Yasa2ForConditionalGeneration" }, "dtype": "bfloat16", "image_token_id": 100278, "label_ignore_index": 100257, "model_type": "yasa2", "num_query_tokens": 64, "text_config": { "add_position_embedding": false, "architectures": [ "YasaCausalLM" ], "attention_bias": false, "attention_dropout": 0.0, "dtype": "bfloat16", "head_dim": 128, "hidden_act": "silu", "hidden_size": 4096, "initializer_range": 0.02, "intermediate_size": 10880, "layer_norm_eps": 1e-05, "linear_no_bias": true, "max_position_embeddings": 16384, "mlp_bias": false, "model_architecture": "decoder_only", "model_type": "yasa_model", "naming_convention": "language_model", "normalization": "RMSNorm", "num_attention_heads": 32, "num_hidden_layers": 32, "num_key_value_heads": 8, "num_query_groups": 8, "pretraining_tp": 1, "rms_norm_eps": 1e-05, "rope_scaling": { "rope_type": "yarn", "factor": 2.0, "original_max_position_embeddings": 8192 }, "rope_theta": 10000, "transformer_impl": "transformer_engine", "use_cache": true, "use_parallel_residual": false, "vocab_size": 100352, "partial_rotary_factor": 1.0 }, "transformers_version": "4.57.3", "use_vision_pos_embed": true, "vision_config": { "depths": [ 3, 3, 27, 3 ], "drop_path_rate": 0.0, "hidden_act": "gelu", "hidden_size": 2816, "hidden_sizes": [ 352, 704, 1408, 2816 ], "image_size": 512, "initializer_range": 0.02, "layer_norm_eps": 1e-12, "layer_scale_init_value": 0, "model_type": "convnext", "num_channels": 3, "patch_size": 4, "use_grn": true }, "vision_pooling": "adaptive_avg" }