Adding data source to a GPT model deployment

Question

When adding a new data source to an existing GPT model deployment, I notice that it requires me to link the model to a blob storage. I have also read on the Azure OpenAI blog that it does not support .xlsx files. How can I transform my xlsx table into some data that can be fed into the model via blob storage. Also is there any preferred format for storing, like the model works the best with a specific .extension ? Does it work the same with all file types?

Thanks!

Answer 1

Ionut Dutescu Greetings & Welcome to Microsoft Q&A forum!

When adding a new data source to an existing GPT model deployment, I notice that it requires me to link the model to a blob storage. I have also read on the Azure OpenAI blog that it does not support .xlsx files.

Yes, depending on the data source you have chosen, you need to create them earlier to link .

I have also read on the Azure OpenAI blog that it does not support .xlsx files. How can I transform my xlsx table into some data that can be fed into the model via blob storage. Also is there any preferred format for storing, like the model works the best with a specific .extension ? Does it work the same with all file types?

Please note that, Azure OpenAI On Your Data supports the following file types:

• .txt
• .md
• .html
• .docx
• .pptx
• .pdf

You can convert your xlsx table into supported formats as mentioned above and that can be fed into the model via blob storage.

While converting data from an unsupported format into a supported format, make sure the conversion:

• Doesn't lead to significant data loss.
• Doesn't add unexpected noise to your data.

Please see data preparation script. The script chunks data so that the model's responses are more accurate.

I hope this helps. Do let me know if you have any further queries.

If the response helped, please do click Accept Answer and Yes for was this answer helpful.

Answer 2

Hello,

I was wondering if you were able to check my question?

Thanks,

Have a nice day!

Answer 3

Hello,

I was wondering, how is the data accessed by the model. I have seen that in the playground chat, it shows me the referenced files from the blob storage. However I only have one file in the storage, and in the references it splits the file into multiple files. Why is this so? It seems like it cannot read the whole file and it is missing details from that part of the file. How can I avoid this split?

Thanks!