KwangHwi/quantization
07
1{2 "architectures": [3 "LlamaBidirectionalModel"4 ],5 "attention_bias": false,6 "attention_dropout": 0.0,7 "auto_map": {8 "AutoConfig": "llama_bidirectional_model.LlamaBidirectionalConfig",9 "AutoModel": "llama_bidirectional_model.LlamaBidirectionalModel"10 },11 "bos_token_id": 128000,12 "dtype": "bfloat16",13 "eos_token_id": 128001,14 "head_dim": 64,15 "hidden_act": "silu",16 "hidden_size": 2048,17 "initializer_range": 0.02,18 "intermediate_size": 8192,19 "max_position_embeddings": 131072,20 "mlp_bias": false,21 "model_type": "llama_bidirec",22 "num_attention_heads": 32,23 "num_hidden_layers": 16,24 "num_key_value_heads": 8,25 "pad_token_id": null,26 "pooling": "avg",27 "pretraining_tp": 1,28 "rms_norm_eps": 1e-05,29 "rope_parameters": {30 "factor": 32.0,31 "high_freq_factor": 4.0,32 "low_freq_factor": 1.0,33 "original_max_position_embeddings": 8192,34 "rope_theta": 500000.0,35 "rope_type": "llama3"36 },37 "temperature": 1.0,38 "tie_word_embeddings": true,39 "transformers_version": "5.14.1",40 "use_bidirectional_attention": true,41 "use_cache": true,42 "vocab_size": 12825643}44 