| { | |
| "architectures": [ | |
| "CustomTransformerModel" | |
| ], | |
| "bos_token_id": 2, | |
| "d_ff": 2048, | |
| "d_model": 512, | |
| "dropout": 0.05, | |
| "eos_token_id": 3, | |
| "max_position_embeddings": 300, | |
| "model_type": "miscovery", | |
| "num_decoder_layers": 8, | |
| "num_encoder_layers": 8, | |
| "num_heads": 8, | |
| "pad_token_id": 0, | |
| "torch_dtype": "float32", | |
| "transformers_version": "4.35.2", | |
| "use_flash_attn": true, | |
| "vocab_size": 50000 | |
| } | |