# Medical Question Answering on Medical TF-QA (test) Source: [Medical Question Answering on Medical TF-QA (test)]() Collection index: [llms.txt]() Data updated: 2026-02-26T05:30:59.445Z Evaluation of model performance on the test split of the Medical TF-QA dataset for Medical Question Answering, measured using metrics including Accuracy (Acc), FC, SER, and SG. A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. | Scope | Recorded value | | --- | --- | | Task | Medical Question Answering | | Dataset (including any stated split) | Medical TF-QA (test) | | Metric | Accuracy | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 18 of 18 | | Coverage | All results currently returned for this benchmark | Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol. ## Metrics - [Accuracy](): Higher is better; unit: PERCENT. - [FC Score](): Higher is better; unit: PERCENT. - [SER](): Lower is better; unit: PERCENT. - [SG Score](): Higher is better; unit: PERCENT. ## Recorded results | Method | Accuracy | Result date | Report paper | | --- | --- | --- | --- | | PubMedBERT + MA-X | 85 | Not recorded | 69705d94e25ad2fa9d1ed173 | | Gemma-3-27B + MA-X | 84 | Not recorded | 69705d94e25ad2fa9d1ed173 | | Flan-T5-XL + MA-X | 83 | Not recorded | 69705d94e25ad2fa9d1ed173 | | BioBERT + MA-X | 83 | Not recorded | 69705d94e25ad2fa9d1ed173 | | BioGPT + MA-X | 83 | Not recorded | 69705d94e25ad2fa9d1ed173 | | Llama-3-8B-Instruct + MA-X | 82 | Not recorded | 69705d94e25ad2fa9d1ed173 | | T5-Large + MA-X | 81 | Not recorded | 69705d94e25ad2fa9d1ed173 | | PubMedBERT | 81 | Not recorded | 69705d94e25ad2fa9d1ed173 | | Mistral-7B-Instruct-v0.3 + MA-X | 80 | Not recorded | 69705d94e25ad2fa9d1ed173 | | BioBERT | 80 | Not recorded | 69705d94e25ad2fa9d1ed173 | | Gemma-3-27B | 79 | Not recorded | 69705d94e25ad2fa9d1ed173 | | DeepSeek-7B + MA-X | 78 | Not recorded | 69705d94e25ad2fa9d1ed173 | | Flan-T5-XL | 78 | Not recorded | 69705d94e25ad2fa9d1ed173 | | BioGPT | 78 | Not recorded | 69705d94e25ad2fa9d1ed173 | | Llama-3-8B-Instruct | 77 | Not recorded | 69705d94e25ad2fa9d1ed173 | | T5-Large | 76 | Not recorded | 69705d94e25ad2fa9d1ed173 | | Mistral-7B-Instruct-v0.3 | 75 | Not recorded | 69705d94e25ad2fa9d1ed173 | | DeepSeek-7B | 73 | Not recorded | 69705d94e25ad2fa9d1ed173 | ### PubMedBERT + MA-X Accuracy: 85. Result ID: 69967bb15698665ec4c36a5c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"Specialised Medical","Activation Steering":"MEDASSESS-X"}. This record does not assert independent reproduction. ### Gemma-3-27B + MA-X Accuracy: 84. Result ID: 69967bb15698665ec4c36a4e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"Decoder-Only","Activation Steering":"MEDASSESS-X"}. This record does not assert independent reproduction. ### Flan-T5-XL + MA-X Accuracy: 83. Result ID: 69967bb15698665ec4c36a58. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"Encoder-Decoder","Activation Steering":"MEDASSESS-X"}. This record does not assert independent reproduction. ### BioBERT + MA-X Accuracy: 83. Result ID: 69967bb15698665ec4c36a5a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"Specialised Medical","Activation Steering":"MEDASSESS-X"}. This record does not assert independent reproduction. ### BioGPT + MA-X Accuracy: 83. Result ID: 69967bb15698665ec4c36a5e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"Specialised Medical","Activation Steering":"MEDASSESS-X"}. This record does not assert independent reproduction. ### Llama-3-8B-Instruct + MA-X Accuracy: 82. Result ID: 69967bb15698665ec4c36a50. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"Decoder-Only","Activation Steering":"MEDASSESS-X"}. This record does not assert independent reproduction. ### T5-Large + MA-X Accuracy: 81. Result ID: 69967bb15698665ec4c36a56. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"Encoder-Decoder","Activation Steering":"MEDASSESS-X"}. This record does not assert independent reproduction. ### PubMedBERT Accuracy: 81. Result ID: 69967bb15698665ec4c36a5b. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Architecture":"Specialised Medical","Activation Steering":"None"}. This record does not assert independent reproduction. ### Mistral-7B-Instruct-v0.3 + MA-X Accuracy: 80. Result ID: 69967bb15698665ec4c36a52. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"Decoder-Only","Activation Steering":"MEDASSESS-X"}. This record does not assert independent reproduction. ### BioBERT Accuracy: 80. Result ID: 69967bb15698665ec4c36a59. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"Specialised Medical","Activation Steering":"None"}. This record does not assert independent reproduction. ### Gemma-3-27B Accuracy: 79. Result ID: 69967bb15698665ec4c36a4d. [Reporting paper]() Recorded evaluation configuration: {"Architecture":"Decoder-Only","Activation Steering":"None"}. This record does not assert independent reproduction. ### DeepSeek-7B + MA-X Accuracy: 78. Result ID: 69967bb15698665ec4c36a54. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"Decoder-Only","Activation Steering":"MEDASSESS-X"}. This record does not assert independent reproduction. ### Flan-T5-XL Accuracy: 78. Result ID: 69967bb15698665ec4c36a57. [Reporting paper]() Recorded evaluation configuration: {"Architecture":"Encoder-Decoder","Activation Steering":"None"}. This record does not assert independent reproduction. ### BioGPT Accuracy: 78. Result ID: 69967bb15698665ec4c36a5d. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Architecture":"Specialised Medical","Activation Steering":"None"}. This record does not assert independent reproduction. ### Llama-3-8B-Instruct Accuracy: 77. Result ID: 69967bb15698665ec4c36a4f. [Reporting paper]() Recorded evaluation configuration: {"Architecture":"Decoder-Only","Activation Steering":"None"}. This record does not assert independent reproduction. ### T5-Large Accuracy: 76. Result ID: 69967bb15698665ec4c36a55. [Reporting paper]() Recorded evaluation configuration: {"Architecture":"Encoder-Decoder","Activation Steering":"None"}. This record does not assert independent reproduction. ### Mistral-7B-Instruct-v0.3 Accuracy: 75. Result ID: 69967bb15698665ec4c36a51. [Reporting paper]() Recorded evaluation configuration: {"Architecture":"Decoder-Only","Activation Steering":"None"}. This record does not assert independent reproduction. ### DeepSeek-7B Accuracy: 73. Result ID: 69967bb15698665ec4c36a53. [Reporting paper]() Recorded evaluation configuration: {"Architecture":"Decoder-Only","Activation Steering":"None"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.