# Medical Grounding on TN3K Source: [Medical Grounding on TN3K]() Collection index: [llms.txt]() Data updated: 2026-03-31T11:20:49.811Z Evaluation of Medical Grounding performance on the TN3K dataset, measured using the A@0.5 metric. 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 Grounding | | Dataset (including any stated split) | TN3K | | Metric | A@0.5 | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 16 of 16 | | 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 - [A@0.5](): Higher is better; unit: PERCENT. - [A@0.8](): Higher is better; unit: PERCENT. - [mAP](): Higher is better; unit: PERCENT. ## Recorded results | Method | A@0.5 | Result date | Report paper | | --- | --- | --- | --- | | MedLoc-R1-32Bpiecewise | 68.01 | Not recorded | 69cb55400636e5bf4afca154 | | MedLoc-R1-7Bpiecewise | 67.11 | Not recorded | 69cb55400636e5bf4afca154 | | MedLoc-R1 | 66.18 | Not recorded | 69cb55400636e5bf4afca154 | | MedLoc-R1-3Bpiecewise | 66.18 | Not recorded | 69cb55400636e5bf4afca154 | | MedLoc-R1-3Bcosine | 65.89 | Not recorded | 69cb55400636e5bf4afca154 | | SFT-3B | 62.39 | Not recorded | 69cb55400636e5bf4afca154 | | MedLoc-R1-3Blinear | 62.15 | Not recorded | 69cb55400636e5bf4afca154 | | Raw-IoU-3B | 57.17 | Not recorded | 69cb55400636e5bf4afca154 | | BoxMed-RL | 56.39 | Not recorded | 69cb55400636e5bf4afca154 | | MedGround-R1 | 51.43 | Not recorded | 69cb55400636e5bf4afca154 | | V-Triune-3B | 43.5 | Not recorded | 69cb55400636e5bf4afca154 | | VLM-R1-3B | 43.01 | Not recorded | 69cb55400636e5bf4afca154 | | GroundingDINO-L | 32.7 | Not recorded | 69cb55400636e5bf4afca154 | | Zero Shot-32B | 21.42 | Not recorded | 69cb55400636e5bf4afca154 | | Zero Shot-7B | 19.12 | Not recorded | 69cb55400636e5bf4afca154 | | Zero Shot-3B | 9.74 | Not recorded | 69cb55400636e5bf4afca154 | ### MedLoc-R1-32Bpiecewise A@0.5: 68.01. Result ID: 69cbadf947f34ce924b4dc79. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Scale":"32B","Scheduling Strategy":"piecewise"}. This record does not assert independent reproduction. ### MedLoc-R1-7Bpiecewise A@0.5: 67.11. Result ID: 69cbadf947f34ce924b4dc78. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Scale":"7B","Scheduling Strategy":"piecewise"}. This record does not assert independent reproduction. ### MedLoc-R1 A@0.5: 66.18. Result ID: 69cbadf947f34ce924b4dc57. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### MedLoc-R1-3Bpiecewise A@0.5: 66.18. Result ID: 69cbadf947f34ce924b4dc77. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Scale":"3B","Scheduling Strategy":"piecewise"}. This record does not assert independent reproduction. ### MedLoc-R1-3Bcosine A@0.5: 65.89. Result ID: 69cbadf947f34ce924b4dc7b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Scale":"3B","Scheduling Strategy":"cosine"}. This record does not assert independent reproduction. ### SFT-3B A@0.5: 62.39. Result ID: 69cbadf947f34ce924b4dc73. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Scale":"3B","Training Strategy":"SFT"}. This record does not assert independent reproduction. ### MedLoc-R1-3Blinear A@0.5: 62.15. Result ID: 69cbadf947f34ce924b4dc7a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Scale":"3B","Scheduling Strategy":"linear"}. This record does not assert independent reproduction. ### Raw-IoU-3B A@0.5: 57.17. Result ID: 69cbadf947f34ce924b4dc76. [Reporting paper]() Recorded evaluation configuration: {"Model Scale":"3B","Training Strategy":"RL","Scheduling Strategy":"Continuous IoU"}. This record does not assert independent reproduction. ### BoxMed-RL A@0.5: 56.39. Result ID: 69cbadf947f34ce924b4dc55. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"status":"reproduced","setting":"bbox-only"}. This record does not assert independent reproduction. ### MedGround-R1 A@0.5: 51.43. Result ID: 69cbadf947f34ce924b4dc56. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### V-Triune-3B A@0.5: 43.5. Result ID: 69cbadf947f34ce924b4dc75. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Scale":"3B","Training Strategy":"RL","Scheduling Strategy":"V-Triune-style"}. This record does not assert independent reproduction. ### VLM-R1-3B A@0.5: 43.01. Result ID: 69cbadf947f34ce924b4dc74. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Scale":"3B","Training Strategy":"RL","Scheduling Strategy":"Fixed"}. This record does not assert independent reproduction. ### GroundingDINO-L A@0.5: 32.7. Result ID: 69cbadf947f34ce924b4dc54. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"status":"fine-tuned"}. This record does not assert independent reproduction. ### Zero Shot-32B A@0.5: 21.42. Result ID: 69cbadf947f34ce924b4dc72. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Scale":"32B","Training Strategy":"Zero-shot"}. This record does not assert independent reproduction. ### Zero Shot-7B A@0.5: 19.12. Result ID: 69cbadf947f34ce924b4dc71. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Scale":"7B","Training Strategy":"Zero-shot"}. This record does not assert independent reproduction. ### Zero Shot-3B A@0.5: 9.74. Result ID: 69cbadf947f34ce924b4dc70. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Scale":"3B","Training Strategy":"Zero-shot"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.