| { | |
| "architectures": [ | |
| "UnifiedParallelMLPForCausalLM" | |
| ], | |
| "base_model_name_or_path": "Qwen/Qwen2.5-1.5B-Instruct", | |
| "dtype": "bfloat16", | |
| "eos_token_id": 151645, | |
| "hidden_size": 1536, | |
| "mlp_intermediate_size": 12288, | |
| "mlp_positions": [ | |
| 24, | |
| 25, | |
| 26, | |
| 27, | |
| 28 | |
| ], | |
| "model_type": "parallel_mlp_causal_lm", | |
| "num_hidden_layers": 28, | |
| "pad_token_id": 151654, | |
| "rms_norm_eps": 1e-06, | |
| "transformers_version": "4.57.6" | |
| } | |