tarekziade HF Staff
Consolidate MIC usage, slim weights and reviewed framework expectations
c35a1e6 verified | { | |
| "architectures": [ | |
| "GotOcr2ForConditionalGeneration" | |
| ], | |
| "dtype": "bfloat16", | |
| "ignore_index": -100, | |
| "image_seq_length": 576, | |
| "image_token_index": 151859, | |
| "model_type": "got_ocr2", | |
| "text_config": { | |
| "attention_dropout": 0.0, | |
| "bos_token_id": null, | |
| "dtype": "float32", | |
| "eos_token_id": null, | |
| "hidden_act": "silu", | |
| "hidden_size": 64, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 128, | |
| "layer_types": [ | |
| "full_attention", | |
| "full_attention" | |
| ], | |
| "max_position_embeddings": 32768, | |
| "max_window_layers": 21, | |
| "model_type": "qwen2", | |
| "num_attention_heads": 4, | |
| "num_hidden_layers": 2, | |
| "num_key_value_heads": 4, | |
| "pad_token_id": null, | |
| "rms_norm_eps": 1e-06, | |
| "rope_parameters": { | |
| "rope_theta": 1000000.0, | |
| "rope_type": "default" | |
| }, | |
| "sliding_window": null, | |
| "tie_word_embeddings": true, | |
| "use_cache": true, | |
| "use_sliding_window": false, | |
| "vocab_size": 151860 | |
| }, | |
| "tie_word_embeddings": true, | |
| "transformers_version": "5.19.0.dev0", | |
| "vision_config": { | |
| "attention_dropout": 0.0, | |
| "dtype": "float32", | |
| "global_attn_indexes": [ | |
| 2, | |
| 5, | |
| 8, | |
| 11 | |
| ], | |
| "hidden_act": "gelu", | |
| "hidden_size": 64, | |
| "image_size": 1024, | |
| "initializer_range": 1e-10, | |
| "layer_norm_eps": 1e-06, | |
| "mlp_dim": 3072, | |
| "model_type": "", | |
| "num_attention_heads": 4, | |
| "num_channels": 3, | |
| "num_hidden_layers": 2, | |
| "output_channels": 256, | |
| "patch_size": 16, | |
| "qkv_bias": true, | |
| "use_abs_pos": true, | |
| "use_rel_pos": true, | |
| "window_size": 14 | |
| } | |
| } | |