nbroad HF staff commited on
Commit
d02045d
·
1 Parent(s): fbef811

Update utils.py

Browse files
Files changed (1) hide show
  1. utils.py +8 -1
utils.py CHANGED
@@ -269,7 +269,7 @@ def batch_embed(
269
 
270
  username = api.whoami()["name"]
271
 
272
- if "/" in new_dataset_id:
273
  new_dataset_id = username + "/" + new_dataset_id
274
 
275
  repo = init_git_repo(new_dataset_id)
@@ -450,3 +450,10 @@ def push_to_repo(
450
  blocking=False,
451
  auto_lfs_prune=True,
452
  )
 
 
 
 
 
 
 
 
269
 
270
  username = api.whoami()["name"]
271
 
272
+ if "/" not in new_dataset_id:
273
  new_dataset_id = username + "/" + new_dataset_id
274
 
275
  repo = init_git_repo(new_dataset_id)
 
450
  blocking=False,
451
  auto_lfs_prune=True,
452
  )
453
+
454
+ most_recent_file = f"embeddings_{last_count}_{current_count}.parquet"
455
+
456
+ # Delete old files
457
+ for f in data_dir.glob("*.parquet"):
458
+ if f.name != most_recent_file:
459
+ f.unlink()