Upload README_EXTRACT.txt with huggingface_hub

README_EXTRACT.txt

@@ -0,0 +1,50 @@
+# How to Extract Compressed Datasets
+
+Each dataset is compressed with zstd and split into <50GB chunks.
+
+## Extract a dataset:
+
+### Method 1: Extract in place
+```bash
+# Combine chunks, decompress, and extract
+cat scannetpp.tar.zst.* | zstd -d | tar -xf -
+
+# Or for specific dataset:
+cat processed_scannet_f.tar.zst.* | zstd -d | tar -xf -
+cat arkitscenes.tar.zst.* | zstd -d | tar -xf -
+```
+
+### Method 2: Extract to specific location
+```bash
+# Extract to /data/reason3r/data_compressed/
+cat scannetpp.tar.zst.* | zstd -d | tar -xf - -C /data/reason3r/data_compressed/
+```
+
+### Method 3: Extract with progress (if pv installed)
+```bash
+cat scannetpp.tar.zst.* | pv | zstd -d | tar -xf -
+```
+
+## Verify before extracting:
+```bash
+# List contents without extracting
+cat scannetpp.tar.zst.* | zstd -d | tar -tf - | head -20
+
+# Check if all chunks present
+ls -lh scannetpp.tar.zst.*
+# Should have .00, .01, .02, etc. in sequence
+```
+
+## Disk space needed:
+- scannetpp: ~44GB
+- processed_scannet_f: ~310GB  
+- arkitscenes: ~560GB
+
+## Transfer files:
+```bash
+# Copy to another machine (preserves chunks)
+rsync -avhP scannetpp.tar.zst.* user@remote:/destination/
+
+# Or use scp
+scp scannetpp.tar.zst.* user@remote:/destination/
+```