kshitijthakkar commited on
Commit
ecc1140
·
1 Parent(s): 6b05c3d

fix: Revert to non-streaming dataset loading for better UX

Browse files

Changed default dataset loading from streaming to full download to fix
first-load data display issues.

**Problem:**
- With streaming=True, data loads asynchronously
- Users saw empty screens on first app load
- Data appeared only after navigation (once streaming completed)

**Solution:**
- Changed use_streaming default: True → False
- App now pre-loads full dataset before creating Gradio interface
- Users see data immediately on first access
- No more empty screen issues

**Performance:**
- Slightly longer initial app startup (~5-10 seconds)
- Much better user experience - no confusion
- Data is cached after first load, so subsequent access is instant

Users can still enable streaming via USE_STREAMING=true environment variable
if they prefer faster startup over immediate data availability.

Files changed (1) hide show
  1. data_loader.py +2 -2
data_loader.py CHANGED
@@ -33,7 +33,7 @@ class DataLoader:
33
  json_data_path: Optional[str] = None,
34
  leaderboard_dataset: Optional[str] = None,
35
  hf_token: Optional[str] = None,
36
- use_streaming: bool = True
37
  ):
38
  self.data_source = data_source
39
  self.json_data_path = Path(json_data_path or os.getenv("JSON_DATA_PATH", "./sample_data"))
@@ -459,7 +459,7 @@ def create_data_loader_from_env() -> DataLoader:
459
  Configured DataLoader instance
460
  """
461
  data_source = os.getenv("DATA_SOURCE", "both")
462
- use_streaming = os.getenv("USE_STREAMING", "true").lower() == "true"
463
 
464
  return DataLoader(
465
  data_source=data_source,
 
33
  json_data_path: Optional[str] = None,
34
  leaderboard_dataset: Optional[str] = None,
35
  hf_token: Optional[str] = None,
36
+ use_streaming: bool = False
37
  ):
38
  self.data_source = data_source
39
  self.json_data_path = Path(json_data_path or os.getenv("JSON_DATA_PATH", "./sample_data"))
 
459
  Configured DataLoader instance
460
  """
461
  data_source = os.getenv("DATA_SOURCE", "both")
462
+ use_streaming = os.getenv("USE_STREAMING", "false").lower() == "true"
463
 
464
  return DataLoader(
465
  data_source=data_source,