Fix Trackio init args

train_sft_qwen25_hf_jobs.py

@@ -105,12 +105,11 @@ def main() -> None:
         timestamp = datetime.utcnow().strftime("%Y-%m-%d_%H-%M-%S")
         run_name = f"sft_{args.model.split('/')[-1]}_{timestamp}"
 
-    trackio.init(
-        project=args.trackio_project,
-        run_name=run_name,
-        space_id=args.trackio_space_id,
-        group=args.trackio_group,
-        config={
+    trackio_kwargs = {
+        "project": args.trackio_project,
+        "run_name": run_name,
+        "space_id": args.trackio_space_id,
+        "config": {
             "model": args.model,
             "dataset": f"{args.dataset}:{args.split}",
             "max_length": args.max_length,
@@ -118,12 +117,17 @@ def main() -> None:
             "epochs": args.num_train_epochs,
             "max_train_samples": args.max_train_samples,
         },
-    )
+    }
+    if args.trackio_group:
+        trackio_kwargs["group"] = args.trackio_group
+    trackio.init(**trackio_kwargs)
 
     print(f"👤 HF user: {hf_user}")
     print(f"📦 Loading dataset: {args.dataset} [{args.split}]")
     dataset = load_dataset(args.dataset, split=args.split)
     print(f"✅ Dataset loaded: {len(dataset)} rows")
+    if len(dataset) < 2:
+        raise SystemExit("Dataset split must have at least 2 rows to create a train/eval split.")
 
     dataset = dataset.shuffle(seed=args.seed)
     eval_size = min(args.max_eval_samples, max(1, int(0.1 * len(dataset))))
@@ -199,4 +203,3 @@ def main() -> None:
 
 if __name__ == "__main__":
     main()
-