| { |
| "architectures": [ |
| "Starcoder2ForCausalLM" |
| ], |
| "attention_dropout": 0.0, |
| "bos_token_id": 50256, |
| "embedding_dropout": 0.0, |
| "eos_token_id": 50256, |
| "hidden_act": "gelu_pytorch_tanh", |
| "hidden_size": 256, |
| "initializer_range": 0.018042, |
| "intermediate_size": 1024, |
| "max_position_embeddings": 4096, |
| "model_type": "starcoder2", |
| "norm_epsilon": 1e-05, |
| "num_attention_heads": 8, |
| "num_hidden_layers": 2, |
| "num_key_value_heads": 4, |
| "residual_dropout": 0.0, |
| "rope_theta": 10000.0, |
| "sliding_window": null, |
| "torch_dtype": "float32", |
| "transformers_version": "4.42.4", |
| "use_bias": true, |
| "use_cache": true, |
| "vocab_size": 2000 |
| } |
|
|
