We introduce CyberSolve, the flan-t5-large model fintuned on all 2M records of the DeepMind LinAlg 1D dataset. This is the first model checkpoint, scoring 86.56 on the eval dataset
71a7616
verified
| { | |
| "_from_model_config": true, | |
| "decoder_start_token_id": 0, | |
| "eos_token_id": 1, | |
| "pad_token_id": 0, | |
| "transformers_version": "4.47.1" | |
| } | |