| { | |
| "_name_or_path": "/data/sparrow/model/pretrain/", | |
| "architectures": [ | |
| "SparrowModelForCausalLM" | |
| ], | |
| "attention_bias": false, | |
| "dropout": 0.0, | |
| "flash_attn": true, | |
| "hidden_dim": 512, | |
| "hidden_size": 512, | |
| "intermediate_dim": 2048, | |
| "max_seq_len": 512, | |
| "mlp_bias": false, | |
| "model_type": "sparrow", | |
| "norm_eps": "1e-5", | |
| "num_attention_heads": 16, | |
| "num_hidden_layers": 8, | |
| "num_key_value_heads": 16, | |
| "torch_dtype": "float32", | |
| "transformers_version": "4.48.2", | |
| "vocab_size": 32000 | |
| } | |