Update app.py

app.py

@@ -10,7 +10,7 @@ from transformers import (
     AutoModelForSpeechSeq2Seq,
     TrainingArguments,
     Trainer,
-    DataCollatorForSeq2Seq,
+    DataCollatorSpeechSeq2Seq,  # ✅ Fix: Use correct data collator
 )
 
 # ================================
@@ -116,16 +116,15 @@ training_args = TrainingArguments(
     hub_token=HF_TOKEN,
 )
 
-# Data collator (for dynamic padding)
-data_collator = DataCollatorForSeq2Seq(processor, model=model)
+# ✅ FIX: Use correct Data Collator
+data_collator = DataCollatorSpeechSeq2Seq(processor=processor, model=model, return_tensors="pt")
 
-# Define Trainer (Fixed `processing_class` warning)
+# Define Trainer
 trainer = Trainer(
     model=model,
     args=training_args,
     train_dataset=train_dataset,
     eval_dataset=eval_dataset,
-    processing_class=processor,  # Fix: Replacing deprecated `tokenizer`
     data_collator=data_collator,
 )