Merge pull request #36 from snehalchennuru/main

add support for function calling

gpt3.go

@@ -32,6 +32,7 @@ type EmbeddingEngine string
 const (
 	GPT3Dot5Turbo             = "gpt-3.5-turbo"
 	GPT3Dot5Turbo0301         = "gpt-3.5-turbo-0301"
+	GPT3Dot5Turbo0613         = "gpt-3.5-turbo-0613"
 	TextSimilarityAda001      = "text-similarity-ada-001"
 	TextSimilarityBabbage001  = "text-similarity-babbage-001"
 	TextSimilarityCurie001    = "text-similarity-curie-001"
@@ -180,8 +181,13 @@ func (c *client) Engine(ctx context.Context, engine string) (*EngineObject, erro
 
 func (c *client) ChatCompletion(ctx context.Context, request ChatCompletionRequest) (*ChatCompletionResponse, error) {
 	if request.Model == "" {
-		request.Model = GPT3Dot5Turbo
+		if request.Functions == nil {
+			request.Model = GPT3Dot5Turbo
+		} else {
+			request.Model = GPT3Dot5Turbo0613
+		}		
 	}
+
 	request.Stream = false
 
 	req, err := c.newRequest(ctx, "POST", "/chat/completions", request)

gpt3_test.go

@@ -213,6 +213,32 @@ func TestResponses(t *testing.T) {
 				},
 			},
 		},
+		{
+			"ChatCompletionWithFunctionCall",
+			func() (interface{}, error) {
+				return client.ChatCompletion(ctx, gpt3.ChatCompletionRequest{})
+			},
+			&gpt3.ChatCompletionResponse{
+				ID:      "chatcmpl-123",
+				Object:  "messages",
+				Created: 123456789,
+				Model:   "gpt-3.5-turbo-0613",
+				Choices: []gpt3.ChatCompletionResponseChoice{
+					{
+						Index:        0,
+						FinishReason: "function_call",
+						Message: gpt3.ChatCompletionResponseMessage{
+							Role:    "assistant",
+							Content: "",
+							FunctionCall: &gpt3.Function{
+								Name: "get_current_weather",
+								Arguments: `"{"location": "Boston, MA"}"`,
+							},
+						},
+					},
+				},
+			},
+		},
 		{
 			"Completion",
 			func() (interface{}, error) {

models.go

@@ -39,6 +39,40 @@ type ChatCompletionRequestMessage struct {
 
 	// Content is the content of the message
 	Content string `json:"content"`
+
+	// FunctionCall is the name and arguments of a function that should be called, as generated by the model.
+	FunctionCall *Function `json:"function_call,omitempty"`
+
+	// Name is the the name of the author of this message. `name` is required if role is `function`, and it should be the name of the function whose response is in the `content`.
+	Name string `json:"name,omitempty"`
+}
+
+// Function represents a function with a name and arguments.
+type Function struct {
+	Name      string `json:"name"`
+	Arguments string `json:"arguments"`
+}
+
+// ChatCompletionFunctions represents the functions the model may generate JSON inputs for.
+type ChatCompletionFunctions struct {
+	Name        string                           `json:"name"`
+	Description string                           `json:"description,omitempty"`
+	Parameters  ChatCompletionFunctionParameters `json:"parameters"`
+}
+
+// ChatCompletionFunctionParameters captures the metadata of the function parameter.
+type ChatCompletionFunctionParameters struct {
+	Type        string                                       `json:"type"`
+	Description string                                       `json:"description,omitempty"`
+	Properties  map[string]FunctionParameterPropertyMetadata `json:"properties"`
+	Required    []string                                     `json:"required"`
+}
+
+// FunctionParameterPropertyMetadata represents the metadata of the function parameter property.
+type FunctionParameterPropertyMetadata struct {
+	Type        string   `json:"type"`
+	Description string   `json:"description,omitempty"`
+	Enum        []string `json:"enum,omitempty"`
 }
 
 // ChatCompletionRequest is a request for the chat completion API
@@ -49,6 +83,9 @@ type ChatCompletionRequest struct {
 	// Messages is a list of messages to use as the context for the chat completion.
 	Messages []ChatCompletionRequestMessage `json:"messages"`
 
+	// Functions is a list of functions the model may generate JSON inputs for.
+	Functions []ChatCompletionFunctions `json:"functions"`
+
 	// What sampling temperature to use, between 0 and 2. Higher values like 0.8 will make the output more random, while lower values like 0.2 will make it more focused and deterministic
 	Temperature *float32 `json:"temperature,omitempty"`
 
@@ -152,8 +189,9 @@ type LogprobResult struct {
 
 // ChatCompletionResponseMessage is a message returned in the response to the Chat Completions API
 type ChatCompletionResponseMessage struct {
-	Role    string `json:"role"`
-	Content string `json:"content"`
+	Role         string   `json:"role"`
+	Content      string   `json:"content"`
+	FunctionCall *Function `json:"function_call,omitempty"`
 }
 
 // ChatCompletionResponseChoice is one of the choices returned in the response to the Chat Completions API