{ "architectures": ["SmartCoderMoEForCausalLM"], "model_type": "smartcoder_moe", "hidden_size": 2048, "num_hidden_layers": 40, "num_attention_heads": 16, "num_key_value_heads": 4, "intermediate_size": 8192, "vocab_size": 65536, "max_position_embeddings": 16384, "rope_theta": 999999.0, "sliding_window": null, "use_cache": true, "tie_word_embeddings": false, "norm_epsilon": 1e-5, "bos_token_id": 0, "eos_token_id": 0, "pad_token_id": 0, "torch_dtype": "bfloat16", "transformers_version": "5.9.0", "n_experts": 32, "expert_dim": 512, "top_k": 2, "total_experts": 1280, "dense_intermediate": 8192, "auto_map": { "AutoConfig": "modeling_smartcoder_moe.SmartCoderMoEConfig", "AutoModelForCausalLM": "modeling_smartcoder_moe.SmartCoderMoEForCausalLM" } }