| { | |
| "architectures": [ | |
| "SwetaForCausalLM" | |
| ], | |
| "bos_token_id": 2, | |
| "d_ff": 2048, | |
| "d_model": 768, | |
| "dropout": 0.0, | |
| "dtype": "float32", | |
| "eos_token_id": 3, | |
| "max_seq_len": 1024, | |
| "model_type": "sweta", | |
| "n_heads": 12, | |
| "n_layers": 12, | |
| "pad_token_id": 0, | |
| "tie_embeddings": true, | |
| "transformers_version": "5.0.0", | |
| "vocab_size": 64000 | |
| } | |