This is a script for tabular editor that will loop through all the measures in a model and get ChatGPT to write a description of the calculation.

AutoPopulateMeasureDescriptionsFromOpenAI.csx

#r "System.Net.Http"
using System.Net.Http;
using System.Text;
using Newtonsoft.Json.Linq;

// You need to signin to https://platform.openai.com/ and create an API key for your profile then paste that key 
// into the apiKey constant below
const string apiKey = "<YOUR API KEY HERE>";
const string uri = "https://api.openai.com/v1/completions";
const string question = "Explain the following calculation in a few sentences in simple business terms without using DAX function names:\n\n";

using (var client = new HttpClient()) {
    client.DefaultRequestHeaders.Clear();
    client.DefaultRequestHeaders.Add("Authorization", "Bearer " + apiKey);

    foreach (var t in Model.Tables)
    {
        foreach ( var m in t.Measures)
        {
            // Only uncomment the following when running from the command line or the script will 
            // show a popup after each measure
            //Info("Processing " + m.DaxObjectFullName) 
            //var body = new requestBody() { prompt = question + m.Expression   };
            var body = 
                "{ \"prompt\": " + JsonConvert.SerializeObject( question + m.Expression ) + 
                ",\"model\": \"gpt-3.5-turbo-instruct\" " +
                ",\"temperature\": 1 " +
                ",\"max_tokens\": 2048 " +
                ",\"stop\": \".\" }";

            var res = client.PostAsync(uri, new StringContent(body, Encoding.UTF8,"application/json"));
            res.Result.EnsureSuccessStatusCode();
            var result = res.Result.Content.ReadAsStringAsync().Result;
            var obj = JObject.Parse(result);
            var desc = obj["choices"][0]["text"].ToString().Trim();
            m.Description = desc + "\n=====\n" + m.Expression;
        }

    }
}

Any ideas here Darren?

Only 2 measures in the testing model...

No sorry I have no idea. That error is coming from the OpenAI servers. You'd have to ask them. A free account should be able to make 20 requests per minute.

I do have an updated version of the script here https://gist.github.com/dgosbell/4cc2c459214961a8d308ae4a8cab7822 which I posted in the follow up to my original blog post. It adds a delay to account for the API throttling, but that would only help when you have more measures in your model than the number of requests your account is allowed to issue per minute.

Ok thank you! You made it a model script not a “selected measures” script right?

If you have some measures selected the updated script will use those, otherwise it goes over all the measures in the model.

Super thanks again! Adam Sabourin

Hi Darren,
This was giving me 404 errors until I changed the model to the following:

","model": "gpt-3.5-turbo-instruct" " +

@Power-BI-guy Thanks. it looks like they deprecated the previous model https://platform.openai.com/docs/deprecations/2023-07-06-gpt-and-embeddings I've updated the script to use "gpt-3.5-turbo-instruct" too. I've also updated this more complicated sample which has some code to deal with the rate limiting https://gist.github.com/dgosbell/4cc2c459214961a8d308ae4a8cab7822