CoolFace
Datasetpublic

RiverRider/srt-cxr14-frozen-probe

Frozen general-purpose features beat a fine-tuned baseline on ChestX-ray14 A linear probe on frozen google/gemma-4-31B-it hidden states, on all 112,120 images of NIH ChestX-ray14, using the official test_list.txt. No fine-tuning, no radiology pretraining, no augmentation. One Linear(d, 14) under BCE, which is fourteen logistic regressions. ChestX-ray14, official split mean AUROC method Wang et al. 2017 (dataset authors) 0.7451 ResNet-50, fine-tuned end to end this… See the full description on the dataset page: https://huggingface.co/datasets/RiverRider/srt-cxr14-frozen-probe.

sourceHugging Faceapache-2.0updated 29d agoView on Hugging Face
2likes538downloads
discussions and pull requests

Conversations for this repository live on Hugging Face.

CoolFace shows imported repositories read-only. Posting into someone else’s repository from here would need an authorised integration and the account holder’s consent, so the link goes to the source instead.

Open discussions on Hugging Face
RiverRider/srt-cxr14-frozen-probe · CoolFace