{ "architectures": [ "Starcoder2ForCausalLM" ], "attention_dropout": 0.1, "bos_token_id": 0, "dtype": "float32", "embedding_dropout": 0.1, "eos_token_id": 0, "hidden_act": "gelu_pytorch_tanh", "hidden_size": 64, "initializer_range": 0.018042, "intermediate_size": 128, "max_position_embeddings": 16384, "mlp_type": "default", "model_type": "starcoder2", "norm_epsilon": 1e-05, "norm_type": "layer_norm", "num_attention_heads": 4, "num_hidden_layers": 2, "num_key_value_heads": 2, "pad_token_id": null, "residual_dropout": 0.1, "rope_parameters": { "rope_theta": 999999.4420358813, "rope_type": "default" }, "sliding_window": 4096, "tie_word_embeddings": true, "transformers_version": "5.22.0.dev0", "use_bias": true, "use_cache": true, "vocab_size": 49152 }