diffusion-reasoning/gdsd_countdown_dream
026
1{2 "architectures": [3 "DreamModel"4 ],5 "attention_dropout": 0.0,6 "auto_map": {7 "AutoConfig": "configuration_dream.DreamConfig",8 "AutoModel": "modeling_dream.DreamModel"9 },10 "bos_token_id": 151643,11 "dtype": "bfloat16",12 "eos_token_id": 151643,13 "hidden_act": "silu",14 "hidden_size": 3584,15 "initializer_range": 0.02,16 "intermediate_size": 18944,17 "mask_token_id": 151666,18 "max_position_embeddings": 131072,19 "max_window_layers": 28,20 "model_type": "Dream",21 "num_attention_heads": 28,22 "num_hidden_layers": 28,23 "num_key_value_heads": 4,24 "pad_token_id": 151643,25 "rms_norm_eps": 1e-06,26 "rope_scaling": null,27 "rope_theta": 1000000.0,28 "sliding_window": null,29 "tie_word_embeddings": false,30 "transformers_version": "4.57.3",31 "use_cache": true,32 "use_mrope": false,33 "use_sliding_window": false,34 "vocab_size": 15206435}36 