| { | |
| "architectures": [ | |
| "TRMForConditionalGeneration" | |
| ], | |
| "hidden_size": 512, | |
| "num_hidden_layers": 2, | |
| "num_attention_heads": 8, | |
| "recursive_steps": 16, | |
| "model_type": "trm", | |
| "torch_dtype": "float32", | |
| "transformers_version": "4.40.0" | |
| } |
| { | |
| "architectures": [ | |
| "TRMForConditionalGeneration" | |
| ], | |
| "hidden_size": 512, | |
| "num_hidden_layers": 2, | |
| "num_attention_heads": 8, | |
| "recursive_steps": 16, | |
| "model_type": "trm", | |
| "torch_dtype": "float32", | |
| "transformers_version": "4.40.0" | |
| } |