{ "architectures": [ "FalconPerceptionForSegmentation" ], "auto_map": { "AutoConfig": "configuration_falcon_perception.FalconPerceptionConfig", "AutoModelForCausalLM": "modeling_falcon_perception.FalconPerceptionForSegmentation" }, "model_type": "falcon_perception", "torch_dtype": "float32", "dim": 1024, "n_layers": 28, "n_heads": 16, "head_dim": 128, "n_kv_heads": 8, "vocab_size": 65536, "ffn_dim": 3072, "norm_eps": 1e-05, "max_seq_len": 8192, "rope_theta": 10000, "channel_size": 3, "spatial_patch_size": 16, "temporal_patch_size": 1, "do_segmentation": true, "segm_out_dim": 256, "num_segm_layers": 3, "coord_enc_dim": 512, "coord_dec_dim": 8192, "coord_out_dim": 2048, "coord_token_id": 240, "size_enc_dim": 512, "size_dec_dim": 8192, "size_out_dim": 2048, "size_token_id": 241, "seg_token_id": 262, "eos_id": 11, "img_id": 227, "image_cls_token_id": 244, "image_reg_1_token_id": 245, "image_reg_2_token_id": 246, "image_reg_3_token_id": 247, "image_reg_4_token_id": 248, "img_end_id": 230 }