fix: concatenate the trainer_callbacks and additional_callbacks in sft_trainer:train()

Signed-off-by: Vassilis Vassiliadis <vassilis.vassiliadis@ibm.com>

Author: VassilisVassiliadis

tuning/sft_trainer.py

@@ -150,7 +150,7 @@ def train(
 
     # Add any extra callback if passed by users
     if additional_callbacks is not None:
-        trainer_callbacks.append(additional_callbacks)
+        trainer_callbacks.extend(additional_callbacks)
 
     framework = AccelerationFrameworkConfig.from_dataclasses(
         quantized_lora_config, fusedops_kernels_config