{ "_name_or_path": "PixelRefer-Lite-7B", "architectures": [ "PixelreferliteMetaForCausalLM" ], "attention_dropout": 0.0, "bos_token_id": 151643, "cross_attn_every_n_layers": 8, "cross_attn_gating_type": "whole-dynamic-tanh-warmup", "cross_attn_implementation": "text-only-vanilla", "eos_token_id": 151645, "fast_token_spatial_stride": 1, "fast_token_temporal_sampling_stride": 1, "fast_token_temporal_stride": 1, "hidden_act": "silu", "hidden_size": 3584, "image_aspect_ratio": "qwen2vl", "image_size": 28, "image_token_index": 151665, "image_token_length": 1, "initializer_range": 0.02, "intermediate_size": 18944, "is_alignment": false, "llm_lr": 1e-05, "mask_num": 32, "max_frames": 180, "max_position_embeddings": 32768, "max_window_layers": 28, "min_fast_frames": null, "mm_hidden_size": 1152, "mm_max_length": 8192, "mm_projector_lr": 1e-05, "mm_projector_type": "mlp2x_gelu", "mm_vision_encoder": "DAMO-NLP-SG/VL3-SigLIP-NaViT", "mm_vision_select_feature": "patch", "mm_vision_select_layer": -2, "model_type": "pixelreferlite_qwen2", "num_attention_heads": 28, "num_hidden_layers": 28, "num_key_value_heads": 4, "region_encoder_lr": 0.001, "region_token_index": 151668, "rms_norm_eps": 1e-06, "rope_scaling": null, "rope_theta": 1000000.0, "sliding_window": null, "spatial_merge_size": 2, "tie_word_embeddings": false, "tokenizer_model_max_length": 16384, "tokenizer_padding_side": "right", "torch_dtype": "bfloat16", "transformers_version": "4.46.3", "use_cache": false, "use_mm_proj": true, "use_sliding_window": false, "use_token_compression": true, "vision_encoder_lr": 0.0, "vocab_size": 151669 }