| { |
| "allow_neg_eigval": false, |
| "architectures": [ |
| "PanoVLMForDecision" |
| ], |
| "attn_mode": "chunk", |
| "auto_map": { |
| "AutoConfig": "configuration_panovlm_decision.PanoVLMDecisionConfig", |
| "AutoImageProcessor": "image_processing_panovlm.PanoVLMImageProcessor", |
| "AutoModel": "modeling_panovlm_decision.PanoVLMForDecision", |
| "AutoProcessor": "processing_panovlm_decision.PanoVLMDecisionProcessor" |
| }, |
| "bos_token_id": 1, |
| "calibration_temperatures": { |
| "choice": 2.2702925205230713, |
| "noul": 1.8340929746627808, |
| "score": 1.017610788345337 |
| }, |
| "conv_bias": false, |
| "conv_size": 4, |
| "dtype": "float32", |
| "eos_token_id": 2, |
| "expand_v": 1.0, |
| "fuse_bitlinear": true, |
| "fuse_norm": true, |
| "head_dim": 128, |
| "head_dropout": 0.1, |
| "head_num_layers": 0, |
| "hidden_size": 1024, |
| "image_seq_length": 144, |
| "image_size": 768, |
| "image_token_id": 32001, |
| "kv_lora_rank": 512, |
| "layer_types": [ |
| "linear_attention", |
| "linear_attention", |
| "linear_attention", |
| "linear_attention", |
| "linear_attention", |
| "full_attention", |
| "linear_attention", |
| "linear_attention", |
| "linear_attention", |
| "linear_attention", |
| "linear_attention", |
| "full_attention", |
| "linear_attention", |
| "linear_attention", |
| "linear_attention", |
| "linear_attention", |
| "linear_attention", |
| "full_attention", |
| "linear_attention", |
| "linear_attention", |
| "linear_attention", |
| "linear_attention", |
| "linear_attention", |
| "full_attention" |
| ], |
| "lower_bound": -1.0, |
| "max_position_embeddings": 131072, |
| "mla_norm_eps": 1e-05, |
| "mla_num_heads": 8, |
| "mlp_hidden_dim": 2816, |
| "model_type": "panovlm_decision", |
| "num_heads": 8, |
| "num_hidden_layers": 24, |
| "num_reserved_token_slots": 128, |
| "num_v_heads": null, |
| "pad_token_id": null, |
| "projector_type": "linear", |
| "q_lora_rank": 512, |
| "qk_nope_head_dim": 128, |
| "qk_rope_head_dim": 64, |
| "rms_norm_eps": 1e-05, |
| "safe_gate": true, |
| "tie_word_embeddings": false, |
| "transformers_version": "5.8.1", |
| "use_bitlinear": false, |
| "use_lower_bound": false, |
| "use_short_conv": true, |
| "v_head_dim": 128, |
| "vision_config": { |
| "cls_ratio": 2.0, |
| "downsamples": [ |
| false, |
| true, |
| true, |
| true, |
| true |
| ], |
| "embed_dims": [ |
| 96, |
| 192, |
| 384, |
| 768, |
| 1536 |
| ], |
| "inference_mode": true, |
| "layers": [ |
| 2, |
| 12, |
| 24, |
| 4, |
| 2 |
| ], |
| "lkc_use_act": true, |
| "mlp_ratios": [ |
| 4, |
| 4, |
| 4, |
| 4, |
| 4 |
| ], |
| "norm_layer": { |
| "func_args": [], |
| "func_kwargs": { |
| "eps": 1e-05 |
| }, |
| "func_name": "LayerNorm2d" |
| }, |
| "num_classes": 0, |
| "pos_embs": [ |
| null, |
| null, |
| null, |
| { |
| "func_args": [], |
| "func_kwargs": { |
| "spatial_shape": [ |
| 7, |
| 7 |
| ] |
| }, |
| "func_name": "RepConditionalPosEnc" |
| }, |
| { |
| "func_args": [], |
| "func_kwargs": { |
| "spatial_shape": [ |
| 7, |
| 7 |
| ] |
| }, |
| "func_name": "RepConditionalPosEnc" |
| } |
| ], |
| "se_downsamples": [ |
| false, |
| false, |
| false, |
| false, |
| false |
| ], |
| "stem_use_scale_branch": false, |
| "token_mixers": [ |
| "repmixer", |
| "repmixer", |
| "repmixer", |
| "attention", |
| "attention" |
| ] |
| }, |
| "vision_feature_dim": 3072, |
| "vocab_size": 32000 |
| } |
|
|
