LoRA adapter fine-tuned from GanjinZero/biobart-large on the MTS-Dialog dataset using the original project prompt and target construction.
Configuration: LoRA rank 8, alpha 64, dropout 0.01, targeting BART query/value projections. Use with GanjinZero/biobart-large through PEFT.
This is an educational research artifact. Generated notes require clinician review and are not clinically validated.
LoRA adapter fine-tuned from GanjinZero/biobart-large on the MTS-Dialog dataset using the original project prompt and target construction.
Configuration: LoRA rank 8, alpha 64, dropout 0.01, targeting BART query/value projections. Use with GanjinZero/biobart-large through PEFT.
This is an educational research artifact. Generated notes require clinician review and are not clinically validated.
