refactor: sample dataset location

Author: Aaron Su

.gitattributes

@@ -2,6 +2,6 @@
 # Only track large binary files in sample_dataset to allow selective downloads
 
 # Track specific file types in sample_dataset only
-training/gr00t/sample_dataset/**/*.mp4 filter=lfs diff=lfs merge=lfs -text
-training/gr00t/sample_dataset/**/*.parquet filter=lfs diff=lfs merge=lfs -text
-training/gr00t/sample_dataset/**/*.jsonl filter=lfs diff=lfs merge=lfs -text
+training/sample_dataset/**/*.mp4 filter=lfs diff=lfs merge=lfs -text
+training/sample_dataset/**/*.parquet filter=lfs diff=lfs merge=lfs -text
+training/sample_dataset/**/*.jsonl filter=lfs diff=lfs merge=lfs -text

training/gr00t/Dockerfile

@@ -13,7 +13,7 @@ ENV DEBIAN_FRONTEND=noninteractive
 # System dependencies - consolidated for better layer caching
 RUN apt-get update && apt-get install -y --no-install-recommends \
     # Core utilities
-    wget curl ca-certificates \
+    wget curl ca-certificates unzip \
     # Git and version control
     git git-lfs \
     # Build essentials

training/gr00t/README.md

@@ -13,7 +13,7 @@
 # 1) DATASET_LOCAL_DIR (use dataset bundled or pre-mounted)
 # 2) DATASET_S3_URI (sync from S3 URI s3://bucket/path)
 # 3) HF_DATASET_ID (download from Hugging Face Datasets)
-# 4) Sample dataset (git clone with Git LFS): /workspace/sample-embodied-ai-platform/training/gr00t/sample_dataset
+# 4) Sample dataset (git clone with Git LFS): /workspace/sample-embodied-ai-platform/training/sample_dataset
 
 set -e  # Exit on any error
 
@@ -147,7 +147,7 @@ mkdir -p "$OUTPUT_DIR" || true
 
 # Resolve dataset source according to priority and ensure accessibility
 SAMPLE_REPO_DIR="/workspace/sample-embodied-ai-platform"
-DEFAULT_SAMPLE_DATASET_DIR="$SAMPLE_REPO_DIR/training/gr00t/sample_dataset"
+DEFAULT_SAMPLE_DATASET_DIR="$SAMPLE_REPO_DIR/training/sample_dataset"
 RESOLVED_DATASET_DIR=""
 
 echo "[Step] Resolve dataset source (priority: local -> s3 -> hf -> sample)"