felixoder/bug_fixer_model

1{2  "_name_or_path": "deepseek-ai/deepseek-coder-1.3b-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  "head_dim": 128,11  "hidden_act": "silu",12  "hidden_size": 2048,13  "initializer_range": 0.02,14  "intermediate_size": 5504,15  "max_position_embeddings": 16384,16  "mlp_bias": false,17  "model_type": "llama",18  "num_attention_heads": 16,19  "num_hidden_layers": 24,20  "num_key_value_heads": 16,21  "pretraining_tp": 1,22  "rms_norm_eps": 1e-06,23  "rope_scaling": {24    "factor": 4.0,25    "rope_type": "linear",26    "type": "linear"27  },28  "rope_theta": 100000,29  "tie_word_embeddings": false,30  "torch_dtype": "float16",31  "transformers_version": "4.48.3",32  "use_cache": true,33  "vocab_size": 3225634}35