From 4a032ce5785fac07072c191a93dcb1b919962c9d Mon Sep 17 00:00:00 2001
From: Ali Nowroozi <59235031+ali-n98@users.noreply.github.com>
Date: Thu, 12 Feb 2026 10:47:07 -0800
Subject: [PATCH] Improved embedding extraction in README for smoother workflow

The comments in the bash required manual deletion which may have caused problems (e.g., extra spaces which caused the script not to work). Added the comment contents to "Key points" section.
---
 README.md | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/README.md b/README.md
index 434333f..8abf129 100644
--- a/README.md
+++ b/README.md
@@ -84,10 +84,10 @@ To extract vision embeddings using the CT models, please refer to the example me
 ```bash
 uv run rate-extract \
     --model pillar0 \
-    --dataset rve_chest_ct \ # rve_abd_ct, rve_brain_ct, rve_chest_ct 
+    --dataset rve_chest_ct \ 
     --split train \
     --batch-size 4 \
-    --model-repo-id YalaLab/Pillar0-ChestCT \ # YalaLab/Pillar0-AbdomenCT, YalaLab/Pillar0-BrainCT, YalaLab/Pillar0-ChestCT 
+    --model-repo-id YalaLab/Pillar0-ChestCT \
     --ct-window-type all \
     --output-dir /path/to/cache \
     data.train_json=/path/to/json \
@@ -101,6 +101,8 @@ uv run rate-extract \
    - `train.json`: JSON file with sample metadata (e.g., `{"sample_name": "EXAMPLE_ACCESSION", "nii_path": null, "report_metadata": "FINDINGS: ..."}`)
    - `manifest.csv`: CSV mapping samples to cached volumes (columns: `sample_name`, `image_cache_path`)
    - Volume directories: Each volume should be in a directory with `volume.mp4` and `metadata.json`
+3. **Dataset options include**: `rve_abd_ct`, `rve_brain_ct`, `rve_chest_ct`
+4. **Model repo options include**: `YalaLab/Pillar0-AbdomenCT`, `YalaLab/Pillar0-BrainCT`, `YalaLab/Pillar0-ChestCT`
 
 ## Troubleshooting
 ### Common Issues