OSError when converting cdialgpt to onnx format

[qiuxia-alone]

Describe the bug i download the whole file from CDial-GPT_LCCC-large , then add "model_type": "openai-gpt" to config.json, meet error as converting: OSError: Couldn't reach server at 'CDial-GPT_LCCC-large/config.json' to download configuration file or configuration file is not a valid JSON file. Please check network or file content here: CDial-GPT_LCCC-large/config.json.

Urgency none

System information

• Platform: Darwin-20.5.0-x86_64-i386-64bit
• Python version: 3.7.11
• PyTorch Version:1.9.0
• Transformers Version:4.9.2
• Onnxruntime Version:1.9.0

To Reproduce code: python -m onnxruntime.transformers.convert_to_onnx -m CDial-GPT_LCCC-large/ --output cdialgpt.onnx -p fp32 --optimize_onnx

[pranavsharma]

If you add a comma at the end of the line "model_type": "openai-gpt", it'll fix the issue you've identified. However, you'll run into another one AttributeError: 'OpenAIGPTConfig' object has no attribute 'n_inner'. This is because we don't support OpenAIGPTConfig today. cc @tianleiwu @wangyems

[tianleiwu]

@qiuxia-alone, the convert_to_onnx script only supports gpt2 model (Example configuration: https://huggingface.co/gpt2/blob/main/config.json).

I think you can modify it a little to export other GPT model family (like openai-gpi, megatron-gpt or gpt-neo).