| { |
| "adapter_config": { |
| "attention_dropout": 0.0, |
| "attn_implementation": "sdpa", |
| "float32_attention": true, |
| "head_dim": 72, |
| "hidden_act": "silu", |
| "hidden_size": 1152, |
| "image_feature_dropout": 0.0, |
| "initializer_range": 0.02, |
| "intermediate_size": 12288, |
| "model_type": "molmo2", |
| "num_attention_heads": 16, |
| "num_key_value_heads": 16, |
| "pooling_attention_mask": true, |
| "residual_dropout": 0.0, |
| "text_hidden_size": 4096, |
| "vit_layers": [ |
| -3, |
| -9 |
| ] |
| }, |
| "architectures": [ |
| "Molmo2ForConditionalGeneration" |
| ], |
| "auto_map": { |
| "AutoConfig": "configuration_molmo2.Molmo2Config", |
| "AutoModelForImageTextToText": "modeling_molmo2.Molmo2ForConditionalGeneration" |
| }, |
| "dtype": "float32", |
| "frame_end_token_id": 151944, |
| "frame_start_token_id": 151943, |
| "image_col_id": 151939, |
| "image_end_token_id": 151937, |
| "image_high_res_id": 151938, |
| "image_low_res_id": 151942, |
| "image_patch_id": 151938, |
| "image_start_token_id": 151936, |
| "initializer_range": 0.02, |
| "low_res_image_start_token_id": 151940, |
| "model_type": "molmo2", |
| "quantization_config": { |
| "config_groups": { |
| "group_0": { |
| "format": "float-quantized", |
| "input_activations": { |
| "actorder": null, |
| "block_structure": null, |
| "dynamic": false, |
| "group_size": null, |
| "num_bits": 8, |
| "observer": "minmax", |
| "observer_kwargs": {}, |
| "scale_dtype": null, |
| "strategy": "tensor", |
| "symmetric": true, |
| "type": "float", |
| "zp_dtype": null |
| }, |
| "output_activations": null, |
| "targets": [ |
| "Linear" |
| ], |
| "weights": { |
| "actorder": null, |
| "block_structure": null, |
| "dynamic": false, |
| "group_size": null, |
| "num_bits": 8, |
| "observer": "minmax", |
| "observer_kwargs": {}, |
| "scale_dtype": null, |
| "strategy": "tensor", |
| "symmetric": true, |
| "type": "float", |
| "zp_dtype": null |
| } |
| } |
| }, |
| "format": "float-quantized", |
| "global_compression_ratio": null, |
| "ignore": [ |
| "model.vision_backbone.image_vit.patch_embedding", |
| "model.vision_backbone.image_vit.transformer.resblocks.0.attention.wq", |
| "model.vision_backbone.image_vit.transformer.resblocks.0.attention.wk", |
| "model.vision_backbone.image_vit.transformer.resblocks.0.attention.wv", |
| "model.vision_backbone.image_vit.transformer.resblocks.0.attention.wo", |
| "model.vision_backbone.image_vit.transformer.resblocks.0.feed_forward.w1", |
| "model.vision_backbone.image_vit.transformer.resblocks.0.feed_forward.w2", |
| "model.vision_backbone.image_vit.transformer.resblocks.1.attention.wq", |
| "model.vision_backbone.image_vit.transformer.resblocks.1.attention.wk", |
| "model.vision_backbone.image_vit.transformer.resblocks.1.attention.wv", |
| "model.vision_backbone.image_vit.transformer.resblocks.1.attention.wo", |
| "model.vision_backbone.image_vit.transformer.resblocks.1.feed_forward.w1", |
| "model.vision_backbone.image_vit.transformer.resblocks.1.feed_forward.w2", |
| "model.vision_backbone.image_vit.transformer.resblocks.2.attention.wq", |
| "model.vision_backbone.image_vit.transformer.resblocks.2.attention.wk", |
| "model.vision_backbone.image_vit.transformer.resblocks.2.attention.wv", |
| "model.vision_backbone.image_vit.transformer.resblocks.2.attention.wo", |
| "model.vision_backbone.image_vit.transformer.resblocks.2.feed_forward.w1", |
| "model.vision_backbone.image_vit.transformer.resblocks.2.feed_forward.w2", |
| "model.vision_backbone.image_vit.transformer.resblocks.3.attention.wq", |
| "model.vision_backbone.image_vit.transformer.resblocks.3.attention.wk", |
| "model.vision_backbone.image_vit.transformer.resblocks.3.attention.wv", |
| "model.vision_backbone.image_vit.transformer.resblocks.3.attention.wo", |
| "model.vision_backbone.image_vit.transformer.resblocks.3.feed_forward.w1", |
| "model.vision_backbone.image_vit.transformer.resblocks.3.feed_forward.w2", |
| "model.vision_backbone.image_vit.transformer.resblocks.4.attention.wq", |
| "model.vision_backbone.image_vit.transformer.resblocks.4.attention.wk", |
| "model.vision_backbone.image_vit.transformer.resblocks.4.attention.wv", |
| "model.vision_backbone.image_vit.transformer.resblocks.4.attention.wo", |
| "model.vision_backbone.image_vit.transformer.resblocks.4.feed_forward.w1", |
| "model.vision_backbone.image_vit.transformer.resblocks.4.feed_forward.w2", |
| "model.vision_backbone.image_vit.transformer.resblocks.5.attention.wq", |
| "model.vision_backbone.image_vit.transformer.resblocks.5.attention.wk", |
| "model.vision_backbone.image_vit.transformer.resblocks.5.attention.wv", |
| "model.vision_backbone.image_vit.transformer.resblocks.5.attention.wo", |
| "model.vision_backbone.image_vit.transformer.resblocks.5.feed_forward.w1", |
| "model.vision_backbone.image_vit.transformer.resblocks.5.feed_forward.w2", |
| "model.vision_backbone.image_vit.transformer.resblocks.6.attention.wq", |
| "model.vision_backbone.image_vit.transformer.resblocks.6.attention.wk", |
| "model.vision_backbone.image_vit.transformer.resblocks.6.attention.wv", |
| "model.vision_backbone.image_vit.transformer.resblocks.6.attention.wo", |
| "model.vision_backbone.image_vit.transformer.resblocks.6.feed_forward.w1", |
| "model.vision_backbone.image_vit.transformer.resblocks.6.feed_forward.w2", |
| "model.vision_backbone.image_vit.transformer.resblocks.7.attention.wq", |
| "model.vision_backbone.image_vit.transformer.resblocks.7.attention.wk", |
| "model.vision_backbone.image_vit.transformer.resblocks.7.attention.wv", |
| "model.vision_backbone.image_vit.transformer.resblocks.7.attention.wo", |
| "model.vision_backbone.image_vit.transformer.resblocks.7.feed_forward.w1", |
| "model.vision_backbone.image_vit.transformer.resblocks.7.feed_forward.w2", |
| "model.vision_backbone.image_vit.transformer.resblocks.8.attention.wq", |
| "model.vision_backbone.image_vit.transformer.resblocks.8.attention.wk", |
| "model.vision_backbone.image_vit.transformer.resblocks.8.attention.wv", |
| "model.vision_backbone.image_vit.transformer.resblocks.8.attention.wo", |
| "model.vision_backbone.image_vit.transformer.resblocks.8.feed_forward.w1", |
| "model.vision_backbone.image_vit.transformer.resblocks.8.feed_forward.w2", |
| "model.vision_backbone.image_vit.transformer.resblocks.9.attention.wq", |
| "model.vision_backbone.image_vit.transformer.resblocks.9.attention.wk", |
| "model.vision_backbone.image_vit.transformer.resblocks.9.attention.wv", |
| "model.vision_backbone.image_vit.transformer.resblocks.9.attention.wo", |
| "model.vision_backbone.image_vit.transformer.resblocks.9.feed_forward.w1", |
| "model.vision_backbone.image_vit.transformer.resblocks.9.feed_forward.w2", |
| "model.vision_backbone.image_vit.transformer.resblocks.10.attention.wq", |
| "model.vision_backbone.image_vit.transformer.resblocks.10.attention.wk", |
| "model.vision_backbone.image_vit.transformer.resblocks.10.attention.wv", |
| "model.vision_backbone.image_vit.transformer.resblocks.10.attention.wo", |
| "model.vision_backbone.image_vit.transformer.resblocks.10.feed_forward.w1", |
| "model.vision_backbone.image_vit.transformer.resblocks.10.feed_forward.w2", |
| "model.vision_backbone.image_vit.transformer.resblocks.11.attention.wq", |
| "model.vision_backbone.image_vit.transformer.resblocks.11.attention.wk", |
| "model.vision_backbone.image_vit.transformer.resblocks.11.attention.wv", |
| "model.vision_backbone.image_vit.transformer.resblocks.11.attention.wo", |
| "model.vision_backbone.image_vit.transformer.resblocks.11.feed_forward.w1", |
| "model.vision_backbone.image_vit.transformer.resblocks.11.feed_forward.w2", |
| "model.vision_backbone.image_vit.transformer.resblocks.12.attention.wq", |
| "model.vision_backbone.image_vit.transformer.resblocks.12.attention.wk", |
| "model.vision_backbone.image_vit.transformer.resblocks.12.attention.wv", |
| "model.vision_backbone.image_vit.transformer.resblocks.12.attention.wo", |
| "model.vision_backbone.image_vit.transformer.resblocks.12.feed_forward.w1", |
| "model.vision_backbone.image_vit.transformer.resblocks.12.feed_forward.w2", |
| "model.vision_backbone.image_vit.transformer.resblocks.13.attention.wq", |
| "model.vision_backbone.image_vit.transformer.resblocks.13.attention.wk", |
| "model.vision_backbone.image_vit.transformer.resblocks.13.attention.wv", |
| "model.vision_backbone.image_vit.transformer.resblocks.13.attention.wo", |
| "model.vision_backbone.image_vit.transformer.resblocks.13.feed_forward.w1", |
| "model.vision_backbone.image_vit.transformer.resblocks.13.feed_forward.w2", |
| "model.vision_backbone.image_vit.transformer.resblocks.14.attention.wq", |
| "model.vision_backbone.image_vit.transformer.resblocks.14.attention.wk", |
| "model.vision_backbone.image_vit.transformer.resblocks.14.attention.wv", |
| "model.vision_backbone.image_vit.transformer.resblocks.14.attention.wo", |
| "model.vision_backbone.image_vit.transformer.resblocks.14.feed_forward.w1", |
| "model.vision_backbone.image_vit.transformer.resblocks.14.feed_forward.w2", |
| "model.vision_backbone.image_vit.transformer.resblocks.15.attention.wq", |
| "model.vision_backbone.image_vit.transformer.resblocks.15.attention.wk", |
| "model.vision_backbone.image_vit.transformer.resblocks.15.attention.wv", |
| "model.vision_backbone.image_vit.transformer.resblocks.15.attention.wo", |
| "model.vision_backbone.image_vit.transformer.resblocks.15.feed_forward.w1", |
| "model.vision_backbone.image_vit.transformer.resblocks.15.feed_forward.w2", |
| "model.vision_backbone.image_vit.transformer.resblocks.16.attention.wq", |
| "model.vision_backbone.image_vit.transformer.resblocks.16.attention.wk", |
| "model.vision_backbone.image_vit.transformer.resblocks.16.attention.wv", |
| "model.vision_backbone.image_vit.transformer.resblocks.16.attention.wo", |
| "model.vision_backbone.image_vit.transformer.resblocks.16.feed_forward.w1", |
| "model.vision_backbone.image_vit.transformer.resblocks.16.feed_forward.w2", |
| "model.vision_backbone.image_vit.transformer.resblocks.17.attention.wq", |
| "model.vision_backbone.image_vit.transformer.resblocks.17.attention.wk", |
| "model.vision_backbone.image_vit.transformer.resblocks.17.attention.wv", |
| "model.vision_backbone.image_vit.transformer.resblocks.17.attention.wo", |
| "model.vision_backbone.image_vit.transformer.resblocks.17.feed_forward.w1", |
| "model.vision_backbone.image_vit.transformer.resblocks.17.feed_forward.w2", |
| "model.vision_backbone.image_vit.transformer.resblocks.18.attention.wq", |
| "model.vision_backbone.image_vit.transformer.resblocks.18.attention.wk", |
| "model.vision_backbone.image_vit.transformer.resblocks.18.attention.wv", |
| "model.vision_backbone.image_vit.transformer.resblocks.18.attention.wo", |
| "model.vision_backbone.image_vit.transformer.resblocks.18.feed_forward.w1", |
| "model.vision_backbone.image_vit.transformer.resblocks.18.feed_forward.w2", |
| "model.vision_backbone.image_vit.transformer.resblocks.19.attention.wq", |
| "model.vision_backbone.image_vit.transformer.resblocks.19.attention.wk", |
| "model.vision_backbone.image_vit.transformer.resblocks.19.attention.wv", |
| "model.vision_backbone.image_vit.transformer.resblocks.19.attention.wo", |
| "model.vision_backbone.image_vit.transformer.resblocks.19.feed_forward.w1", |
| "model.vision_backbone.image_vit.transformer.resblocks.19.feed_forward.w2", |
| "model.vision_backbone.image_vit.transformer.resblocks.20.attention.wq", |
| "model.vision_backbone.image_vit.transformer.resblocks.20.attention.wk", |
| "model.vision_backbone.image_vit.transformer.resblocks.20.attention.wv", |
| "model.vision_backbone.image_vit.transformer.resblocks.20.attention.wo", |
| "model.vision_backbone.image_vit.transformer.resblocks.20.feed_forward.w1", |
| "model.vision_backbone.image_vit.transformer.resblocks.20.feed_forward.w2", |
| "model.vision_backbone.image_vit.transformer.resblocks.21.attention.wq", |
| "model.vision_backbone.image_vit.transformer.resblocks.21.attention.wk", |
| "model.vision_backbone.image_vit.transformer.resblocks.21.attention.wv", |
| "model.vision_backbone.image_vit.transformer.resblocks.21.attention.wo", |
| "model.vision_backbone.image_vit.transformer.resblocks.21.feed_forward.w1", |
| "model.vision_backbone.image_vit.transformer.resblocks.21.feed_forward.w2", |
| "model.vision_backbone.image_vit.transformer.resblocks.22.attention.wq", |
| "model.vision_backbone.image_vit.transformer.resblocks.22.attention.wk", |
| "model.vision_backbone.image_vit.transformer.resblocks.22.attention.wv", |
| "model.vision_backbone.image_vit.transformer.resblocks.22.attention.wo", |
| "model.vision_backbone.image_vit.transformer.resblocks.22.feed_forward.w1", |
| "model.vision_backbone.image_vit.transformer.resblocks.22.feed_forward.w2", |
| "model.vision_backbone.image_vit.transformer.resblocks.23.attention.wq", |
| "model.vision_backbone.image_vit.transformer.resblocks.23.attention.wk", |
| "model.vision_backbone.image_vit.transformer.resblocks.23.attention.wv", |
| "model.vision_backbone.image_vit.transformer.resblocks.23.attention.wo", |
| "model.vision_backbone.image_vit.transformer.resblocks.23.feed_forward.w1", |
| "model.vision_backbone.image_vit.transformer.resblocks.23.feed_forward.w2", |
| "model.vision_backbone.image_vit.transformer.resblocks.24.attention.wq", |
| "model.vision_backbone.image_vit.transformer.resblocks.24.attention.wk", |
| "model.vision_backbone.image_vit.transformer.resblocks.24.attention.wv", |
| "model.vision_backbone.image_vit.transformer.resblocks.24.attention.wo", |
| "model.vision_backbone.image_vit.transformer.resblocks.24.feed_forward.w1", |
| "model.vision_backbone.image_vit.transformer.resblocks.24.feed_forward.w2", |
| "model.vision_backbone.image_pooling_2d.wq", |
| "model.vision_backbone.image_pooling_2d.wk", |
| "model.vision_backbone.image_pooling_2d.wv", |
| "model.vision_backbone.image_pooling_2d.wo", |
| "model.vision_backbone.image_projector.w1", |
| "model.vision_backbone.image_projector.w2", |
| "model.vision_backbone.image_projector.w3", |
| "lm_head" |
| ], |
| "kv_cache_scheme": null, |
| "quant_method": "compressed-tensors", |
| "quantization_status": "compressed", |
| "sparsity_config": {}, |
| "transform_config": {}, |
| "version": "0.13.0" |
| }, |
| "text_config": { |
| "additional_vocab_size": 128, |
| "attention_dropout": 0.0, |
| "attn_implementation": "sdpa", |
| "embedding_dropout": 0.0, |
| "head_dim": 128, |
| "hidden_act": "silu", |
| "hidden_size": 4096, |
| "initializer_range": 0.02, |
| "intermediate_size": 12288, |
| "layer_norm_eps": 1e-06, |
| "max_position_embeddings": 36864, |
| "model_type": "molmo2_text", |
| "norm_after": false, |
| "num_attention_heads": 32, |
| "num_hidden_layers": 36, |
| "num_key_value_heads": 8, |
| "qk_norm_type": "qwen3", |
| "qkv_bias": false, |
| "residual_dropout": 0.0, |
| "rope_scaling": null, |
| "rope_scaling_layers": null, |
| "rope_theta": 1000000.0, |
| "use_cache": true, |
| "use_qk_norm": true, |
| "vocab_size": 151936 |
| }, |
| "tie_word_embeddings": false, |
| "transformers_version": "4.57.3", |
| "use_cache": true, |
| "use_frame_special_tokens": false, |
| "vit_config": { |
| "attention_dropout": 0.0, |
| "attn_implementation": "sdpa", |
| "float32_attention": true, |
| "head_dim": 72, |
| "hidden_act": "gelu_pytorch_tanh", |
| "hidden_size": 1152, |
| "image_default_input_size": [ |
| 378, |
| 378 |
| ], |
| "image_num_pos": 729, |
| "image_patch_size": 14, |
| "initializer_range": 0.02, |
| "intermediate_size": 4304, |
| "layer_norm_eps": 1e-06, |
| "model_type": "molmo2", |
| "num_attention_heads": 16, |
| "num_hidden_layers": 27, |
| "num_key_value_heads": 16, |
| "residual_dropout": 0.0 |
| } |
| } |
