Mechanistically guided fine-tuning, for PSF - MedGemma-4B (targeted + full) and LLaVA-RAD at n=12K multi-task
AI & ML interests
None defined yet.
Recent Activity
models 18
saillab/llava-rad-targeted-lora-mimic-mt-12k
Image-Text-to-Text • Updated
saillab/medgemma-4b-full-lora-mimic-mt-12k
Image-Text-to-Text • Updated
saillab/medgemma-4b-targeted-lora-mimic-mt-12k
Image-Text-to-Text • Updated
saillab/x-guard
3B • Updated • 202 • 3
saillab/mbart-x-guard
0.6B • Updated • 9
saillab/nepali-llama3-8b-taco
Updated
saillab/taco-sanskrit-33b
Updated
saillab/taco-persian-33b
Updated • 3
saillab/taco-maithili-33b
Updated
saillab/taco-nepali-33b
Updated • 1
datasets 272
saillab/vsfmed-v2
Updated • 66
saillab/psf-med
Updated • 48
saillab/psf-readout-alignment-results
Updated • 14
saillab/VinDR-CXR-PSF
Preview • Updated • 70
saillab/medical-vqa-robustness-analysis
Viewer • Updated • 1 • 38
saillab/alpaca-lao-cleaned
Viewer • Updated • 52k • 48
saillab/alpaca-tamil-cleaned
Viewer • Updated • 52k • 90 • 1
saillab/alpaca-urdu-cleaned
Viewer • Updated • 52k • 108
saillab/alpaca-sindhi-cleaned
Viewer • Updated • 52k • 20
saillab/alpaca-croatian-cleaned
Viewer • Updated • 52k • 67 • 1