In this tutorial we try to show by writing an example how to integrate the chatgpt api in a program written in VB.Net
First we start by creating a VB ChatGPT project
Once created,we will have the first form already inserted in our project
We will need the speech library, you go to the nuget package for installing the library
Once done, we begin to insert all our objects necessary to build our chatgpt, the first object is textEdit multiline where we will have the answer from the server, we give it the name txtAnswer
Then we insert the txtQuestion object where we write our questions, we insert the send button to send our request I try giving it the name btnSend
We insert the lblSpeech label which is used to receive part of the text each time we receive the answer.
Now we insert the randomness object associated with the temperature variable. Temperature is a hyperparameter used in some natural language processing models,including ChatGPT, to control the level of randomnes
After we insert the objet MaxTokens (TextEdit)
The models understand and process text by breaking it down into tokens.
Tokens can be words or just chunks of characters. For example, the word “hamburger” gets broken up into the tokens “ham”, “bur” and “ger”, while a short and common word like “pear” is a single token. Many tokens start with a whitespace, for example “ hello” and “ bye”.
The number of tokens processed in a given API request depends on the length of both your inputs and outputs. As a rough rule of thumb, 1 token is approximately 4 characters or 0.75 words for English text.
One limitation to keep in mind is that your text prompt and generated completion combined must be no more than the model's maximum context length (for most models this is 2048 tokens, or about 1500 words).
Check out the tokenizer tool to learn more about how text translates to tokens.
Then we insert a checkbox to put on pause listening and another to mute. We also use a combobox for the models.
The API is powered by a set of models with different capabilities and price points. the openai base GPT-3 models are called Davinci, Curie, Babbage and Ada. The Codex series is a descendant of GPT-3 that’s been trained on both natural language and code. To learn more, visit the models documentation of the openai website.
And another for the existing voices, which we must loadby sending a request to the server
Dim OPENAI_API_KEY As String = "" 'https://beta.openai.com/account/api-keys
Dim oSpeechRecognitionEngine As SpeechRecognitionEngine = Nothing
Dim oSpeechSynthesizer As System.Speech.Synthesis.SpeechSynthesizer = Nothing
We need oSpeechRecognitonEngine and oSpeechSynthesizer variables in the load procedure of the form we read the value of openai_api_key xml configuration file appi.config
Now we try to load the combo that contains the different voices that we can use.
Private Sub frmGPTChat_Load(sender As Object, e As EventArgs) Handles MyBase.Load
Dim oAppSettingsReader As New AppSettingsReader()
Dim sApiKey As String = oAppSettingsReader.GetValue("OPENAI_API_KEY", GetType(String)) & ""
If sApiKey = "" Then
MsgBox("Please enter your OpenAI API key in the App.config file.")
End
Else
OPENAI_API_KEY = sApiKey
End If
'SetModels() We can load all the models or load only the three models that I've noticed working
cbModel.Items.Clear()
cbModel.Items.Add("text-davinci-003")
cbModel.Items.Add("text-davinci-002")
cbModel.Items.Add("code-davinci-002")
cbModel.SelectedIndex = 0
cbVoice.Items.Clear()
Dim synth As New SpeechSynthesizer()
For Each voice In synth.GetInstalledVoices()
cbVoice.Items.Add(voice.VoiceInfo.Name)
Next
cbVoice.SelectedIndex = 0
End Sub
When you click on cklisten if selected, you see the lblSpeech label and we call the function speechtoText
Private Sub chkListen_CheckedChanged(sender As Object, e As EventArgs) Handles chkListen.CheckedChanged
If chkListen.Checked Then
lblSpeech.Text = ""
lblSpeech.Visible = True
SpeechToText()
Else
oSpeechRecognitionEngine.RecognizeAsyncStop()
lblSpeech.Visible = False
End If
End Sub
Private Sub chkMute_CheckedChanged(sender As Object, e As EventArgs) Handles chkMute.CheckedChanged
If chkMute.Checked Then
lblVoice.Visible = False
cbVoice.Visible = False
Else
lblVoice.Visible = True
cbVoice.Visible = True
End If
End Sub
If the oSpeechRecognitionEngine object is not initialized, we initialize it
Private Sub SpeechToText()
If oSpeechRecognitionEngine IsNot Nothing Then
oSpeechRecognitionEngine.RecognizeAsync(RecognizeMode.Multiple)
Exit Sub
End If
oSpeechRecognitionEngine = New SpeechRecognitionEngine(New System.Globalization.CultureInfo("en-US"))
oSpeechRecognitionEngine.LoadGrammar(New DictationGrammar())
AddHandler oSpeechRecognitionEngine.SpeechRecognized, AddressOf OnSpeechRecognized
AddHandler oSpeechRecognitionEngine.SpeechHypothesized, AddressOf OnSpeechHypothesized
oSpeechRecognitionEngine.SetInputToDefaultAudioDevice()
oSpeechRecognitionEngine.RecognizeAsync(RecognizeMode.Multiple)
End Sub
We also associate the OnSpeechRecognized event which allows us to prepare our question before sending it, we associate the OnSpeechHypoSynthesized event which allows us to visualize the text in the lblSpeech label.
Private Sub OnSpeechRecognized(sender As Object, e As SpeechRecognizedEventArgs)
lblSpeech.Text = "" 'Reset Hypothesized text
If txtQuestion.Text <> "" Then
txtQuestion.Text += vbCrLf
End If
Dim text As String = e.Result.Text
txtQuestion.Text += text
End Sub
Private Sub OnSpeechHypothesized(sender As Object, e As SpeechHypothesizedEventArgs)
Dim text As String = e.Result.Text
lblSpeech.Text = text
End Sub
Now we go to the send button which allows us to send the request thanks to the sendMsg function using the http protocol by invoking the openai API
Private Sub btnSend_Click(sender As Object, e As EventArgs) Handles btnSend.Click
Dim sQuestion As String = txtQuestion.Text
If sQuestion = "" Then
MsgBox("Type in your question!")
txtQuestion.Focus()
Exit Sub
End If
If txtAnswer.Text <> "" Then
txtAnswer.AppendText(vbCrLf)
End If
txtAnswer.AppendText("Me: " & sQuestion & vbCrLf)
txtQuestion.Text = ""
Try
Dim sAnswer As String = SendMsg(sQuestion)
txtAnswer.AppendText("Chat GPT: " & Replace(sAnswer, vbLf, vbCrLf))
SpeechToText(sAnswer)
Catch ex As Exception
txtAnswer.AppendText("Error: " & ex.Message)
End Try
End Sub
We have the endpoint api.openai.com/v1/completion and the request is done with post method with json content
Our request consists of maxtokens with value 2048, randomness 0.5, the user id for this example is 1 but we can associate the requests to differents users
We prepare the data variable with all the necessary parameters at the end we read our response in json format.
Function SendMsg(ByVal sQuestion As String)
System.Net.ServicePointManager.SecurityProtocol =
System.Net.SecurityProtocolType.Ssl3 Or
System.Net.SecurityProtocolType.Tls12 Or
System.Net.SecurityProtocolType.Tls11 Or
System.Net.SecurityProtocolType.Tls
Dim apiEndpoint As String = "https://api.openai.com/v1/completions"
Dim request As HttpWebRequest = WebRequest.Create(apiEndpoint)
request.Method = "POST"
request.ContentType = "application/json"
request.Headers.Add("Authorization", "Bearer " & OPENAI_API_KEY)
Dim iMaxTokens As Integer = txtMaxTokens.Text '2048
Dim dTemperature As Double = txtTemperature.Text '0.5
If dTemperature < 0 Or dTemperature > 1 Then
MsgBox("Randomness has to be between 0 and 1 with higher values resulting in more random text")
Return ""
End If
Dim sUserId As String = "1"
Dim sModel As String = cbModel.Text 'text-davinci-002, text-davinci-003
'https://beta.openai.com/docs/api-reference/completions/create
Dim data As String = "{"
data += " ""model"":""" & sModel & ""","
data += " ""prompt"": """ & PadQuotes(sQuestion) & ""","
data += " ""max_tokens"": " & iMaxTokens & ","
data += " ""user"": """ & sUserId & """, "
data += " ""temperature"": " & dTemperature & ", "
data += " ""frequency_penalty"": 0.0" & ", " 'Number between -2.0 and 2.0 Positive value decrease the model's likelihood to repeat the same line verbatim.
data += " ""presence_penalty"": 0.0" & ", " ' Number between -2.0 and 2.0. Positive values increase the model's likelihood to talk about new topics.
data += " ""stop"": [""#"", "";""]" 'Up to 4 sequences where the API will stop generating further tokens. The returned text will not contain the stop sequence.
data += "}"
Using streamWriter As New StreamWriter(request.GetRequestStream())
streamWriter.Write(data)
streamWriter.Flush()
streamWriter.Close()
End Using
Dim response As HttpWebResponse = request.GetResponse()
Dim streamReader As New StreamReader(response.GetResponseStream())
Dim sJson As String = streamReader.ReadToEnd()
'Return sJson
Dim oJavaScriptSerializer As New System.Web.Script.Serialization.JavaScriptSerializer
Dim oJson As Hashtable = oJavaScriptSerializer.Deserialize(Of Hashtable)(sJson)
Dim sResponse As String = oJson("choices")(0)("text")
Return sResponse
End Function
The setmodels function allows to initialize the combo with the possible models, for the moment I noted that only 3 models that I can use the others don't work, maybe because my registration in the openai site is free
Private Sub SetModels()
'https://beta.openai.com/docs/models/gpt-3
System.Net.ServicePointManager.SecurityProtocol =
System.Net.SecurityProtocolType.Ssl3 Or
System.Net.SecurityProtocolType.Tls12 Or
System.Net.SecurityProtocolType.Tls11 Or
System.Net.SecurityProtocolType.Tls
Dim apiEndpoint As String = "https://api.openai.com/v1/models"
Dim request As HttpWebRequest = WebRequest.Create(apiEndpoint)
request.Method = "GET"
request.ContentType = "application/json"
request.Headers.Add("Authorization", "Bearer " & OPENAI_API_KEY)
Dim response As HttpWebResponse = request.GetResponse()
Dim streamReader As New StreamReader(response.GetResponseStream())
Dim sJson As String = streamReader.ReadToEnd()
'Return sJson
cbModel.Items.Clear()
Dim oSortedList As SortedList = New SortedList()
Dim oJavaScriptSerializer As New System.Web.Script.Serialization.JavaScriptSerializer
Dim oJson As Hashtable = oJavaScriptSerializer.Deserialize(Of Hashtable)(sJson)
Dim oList As Object() = oJson("data")
For i As Integer = 0 To oList.Length - 1
Dim sId As String = oList(i)("id")
oSortedList.Add(sId, sId)
Next
'Text-davinci - 3
'Text-davinci - 2
'code-davinci - 2
For Each oItem As DictionaryEntry In oSortedList
cbModel.Items.Add(oItem.Key)
Next
End Sub
Let's not forget to replace the different characters not allowed using a separate function called PdQuotes before sending the message
Private Function PadQuotes(ByVal s As String) As String
If s.IndexOf("\") <> -1 Then
s = Replace(s, "\", "\\")
End If
If s.IndexOf(vbCrLf) <> -1 Then
s = Replace(s, vbCrLf, "\n")
End If
If s.IndexOf(vbCr) <> -1 Then
s = Replace(s, vbCr, "\r")
End If
If s.IndexOf(vbLf) <> -1 Then
s = Replace(s, vbLf, "\f")
End If
If s.IndexOf(vbTab) <> -1 Then
s = Replace(s, vbTab, "\t")
End If
If s.IndexOf("""") = -1 Then
Return s
Else
Return Replace(s, """", "\""")
End If
End Function
End Class
Once the code part is finished, we will look for the key, we must make a registration in the website openai.com, it's free and we paste the key in our xml file api.config
, either we can initialize it with the code or in the project configuration file app.config.
