cllm/deepseekcoder_7b_codesearch_net_python
134
1{2 "_name_or_path": "../../models/deepseek-coder-6.7b-instruct/",3 "architectures": [4 "LlamaForCausalLM"5 ],6 "attention_bias": false,7 "attention_dropout": 0.0,8 "bos_token_id": 32013,9 "eos_token_id": 32021,10 "hidden_act": "silu",11 "hidden_size": 4096,12 "initializer_range": 0.02,13 "intermediate_size": 11008,14 "max_position_embeddings": 16384,15 "model_type": "llama",16 "num_attention_heads": 32,17 "num_hidden_layers": 32,18 "num_key_value_heads": 32,19 "pad_token_id": 32018,20 "pretraining_tp": 1,21 "rms_norm_eps": 1e-06,22 "rope_scaling": {23 "factor": 4.0,24 "type": "linear"25 },26 "rope_theta": 100000,27 "tie_word_embeddings": false,28 "torch_dtype": "bfloat16",29 "transformers_version": "4.36.2",30 "use_cache": true,31 "vocab_size": 3225632}33 