Team Ai
Modelpublic

Devops-hestabit/deepseek-code

sourceHugging Faceotherupdated 2y agoView on Hugging Face
0likes24downloads
config.json29 linesDownload Raw Back to root
1{2  "architectures": [3    "LlamaForCausalLM"4  ],5  "bos_token_id": 32013,6  "eos_token_id": 32021,7  "hidden_act": "silu",8  "hidden_size": 7168,9  "initializer_range": 0.02,10  "intermediate_size": 19200,11  "max_position_embeddings": 16384,12  "model_type": "llama",13  "num_attention_heads": 56,14  "num_hidden_layers": 62,15  "num_key_value_heads": 8,16  "pretraining_tp": 1,17  "rms_norm_eps": 1e-06,18  "rope_scaling": {19    "factor": 4.0,20    "type": "linear"21  },22  "rope_theta": 100000,23  "tie_word_embeddings": false,24  "torch_dtype": "bfloat16",25  "transformers_version": "4.33.1",26  "use_cache": true,27  "vocab_size": 3225628}29