Why Is the Answer of the Fine-Tuned Pangu Model Truncated Abnormally?

After you deploy a fine-tuned model and ask a question that belongs to the same target task, the model's answer is incomplete and is abnormally truncated. Locate the fault as follows:

• Inference parameter settings: Check the setting of the max_token parameter. Increase its value to increase the maximum length of the model's answer and avoid abnormal truncation. Note that the value of this parameter has an upper limit. Adjust it based on the actual requirements of the target task and the maximum length supported by the model.
• Model specifications: The maximum length of model outputs varies by model specifications. If the length required by the target task exceeds the maximum supported, you can choose another model.
• Data quality: Check whether the training data contains abnormal truncation. If it does, you can cleanse the data using rules.