Feat: rp mode

config/config.go

@@ -7,10 +7,13 @@ import (
 )
 
 type Config struct {
-	CurrentAPI    string `toml:"CurrentAPI"`
+	RemoteAPI      string `toml:"RemoteAPI"`
+	LocalAPI       string `toml:"LocalAPI"`
 	APIToken       string `toml:"APIToken"`
 	QuestionsPath  string `toml:"QuestionsPath"`
-	OutPath       string `toml:"OutPath"`
+	RPScenariosDir string `toml:"RPScenariosDir"`
+	OutDir         string `toml:"OutDir"`
+	MaxTurns       int    `toml:"MaxTurns"`
 }
 
 func LoadConfigOrDefault(fn string) *Config {
@@ -21,9 +24,12 @@ func LoadConfigOrDefault(fn string) *Config {
 	_, err := toml.DecodeFile(fn, &config)
 	if err != nil {
 		fmt.Println("failed to read config from file, loading default", "error", err)
-		config.CurrentAPI = "http://localhost:8080/completion"
+		config.RemoteAPI = "https://openrouter.ai/api/v1/completions"
+		config.LocalAPI = "http://localhost:8080/completion"
 		config.QuestionsPath = "data/questions.json"
-		config.OutPath = "data/out.json"
+		config.RPScenariosDir = "scenarios"
+		config.OutDir = "results"
+		config.MaxTurns = 10
 	}
 	return config
 }

main.go

@@ -1,13 +1,17 @@
 package main
 
 import (
+	"bytes"
 	"encoding/json"
 	"errors"
+	"flag"
 	"fmt"
 	"io"
 	"log/slog"
 	"net/http"
 	"os"
+	"path/filepath"
+	"strings"
 	"time"
 
 	"grailbench/config"
@@ -18,7 +22,6 @@ var (
 	logger     *slog.Logger
 	cfg        *config.Config
 	httpClient = &http.Client{}
-	currentModel = ""
 )
 
 func loadQuestions(fp string) ([]models.Question, error) {
@@ -35,7 +38,35 @@ func loadQuestions(fp string) ([]models.Question, error) {
 	return resp, nil
 }
 
-func chooseParser() RespParser {
+func loadRPScenarios(dir string) ([]*models.RPScenario, error) {
+	scenarios := []*models.RPScenario{}
+	files, err := os.ReadDir(dir)
+	if err != nil {
+		return nil, err
+	}
+	for _, file := range files {
+		if !file.IsDir() && strings.HasSuffix(file.Name(), ".json") {
+			fp := filepath.Join(dir, file.Name())
+			data, err := os.ReadFile(fp)
+			if err != nil {
+				logger.Error("failed to read scenario file", "error", err, "fp", fp)
+				continue // Skip this file
+			}
+			scenario := &models.RPScenario{}
+			if err := json.Unmarshal(data, &scenario); err != nil {
+				logger.Error("failed to unmarshal scenario file", "error", err, "fp", fp)
+				continue // Skip this file
+			}
+			scenarios = append(scenarios, scenario)
+		}
+	}
+	return scenarios, nil
+}
+
+func chooseParser(apiURL string) RespParser {
+	if apiURL == cfg.RemoteAPI {
+		return NewOpenRouterParser(logger)
+	}
 	return NewLCPRespParser(logger)
 }
 
@@ -48,6 +79,11 @@ func init() {
 }
 
 func main() {
+	mode := flag.String("mode", "questions", "Benchmark mode: 'questions' or 'rp'")
+	flag.Parse()
+
+	switch *mode {
+	case "questions":
 		questions, err := loadQuestions(cfg.QuestionsPath)
 		if err != nil {
 			panic(err)
@@ -60,97 +96,171 @@ func main() {
 		if err != nil {
 			panic(err)
 		}
-	if err := os.WriteFile(cfg.OutPath, data, 0666); err != nil {
+		if err := os.WriteFile(filepath.Join(cfg.OutDir, "results.json"), data, 0666); err != nil {
 			panic(err)
 		}
+	case "rp":
+		scenarios, err := loadRPScenarios(cfg.RPScenariosDir)
+		if err != nil {
+			panic(err)
+		}
+		for _, scenario := range scenarios {
+			logger.Info("running scenario", "name", scenario.Name)
+			conversation, err := runRPBench(scenario)
+			if err != nil {
+				logger.Error("failed to run rp bench", "error", err, "scenario", scenario.Name)
+				continue
+			}
+			if err := saveRPResult(scenario.Name, conversation); err != nil {
+				logger.Error("failed to save rp result", "error", err, "scenario", scenario.Name)
+			}
+		}
+	default:
+		fmt.Println("Unknown mode, please use 'questions' or 'rp'")
+	}
+}
+
+func saveRPResult(name string, conversation []models.RPMessage) error {
+	date := time.Now().Format("2006-01-02")
+	fn := fmt.Sprintf("%s_%s.txt", name, date)
+	fp := filepath.Join(cfg.OutDir, fn)
+
+	var b strings.Builder
+	for _, msg := range conversation {
+		b.WriteString(fmt.Sprintf("%s:\n%s\n\n", msg.Author, msg.Content))
+	}
+
+	return os.WriteFile(fp, []byte(b.String()), 0666)
+}
+
+func runRPBench(scenario *models.RPScenario) ([]models.RPMessage, error) {
+	conversation := []models.RPMessage{}
+	if scenario.FirstMsg != "" {
+		conversation = append(conversation, models.RPMessage{Author: "Scenario", Content: scenario.FirstMsg})
+	}
+
+	for i := 0; i < cfg.MaxTurns; i++ {
+		// User's turn (local model)
+		userPrompt := buildRPPrompt(scenario.UserSysPrompt, conversation, scenario.UserName)
+		userResp, err := callLLM(userPrompt, cfg.LocalAPI)
+		if err != nil {
+			return nil, fmt.Errorf("user turn failed: %w", err)
+		}
+		parser := chooseParser(cfg.LocalAPI)
+		userText, err := parser.ParseBytes(userResp)
+		if err != nil {
+			return nil, fmt.Errorf("user turn failed parsing response: %w", err)
+		}
+		conversation = append(conversation, models.RPMessage{Author: scenario.UserName, Content: userText})
+
+		// Character's turn (remote model)
+		charPrompt := buildRPPrompt(scenario.CharSysPrompt, conversation, scenario.CharName)
+		charResp, err := callLLM(charPrompt, cfg.RemoteAPI)
+		if err != nil {
+			return nil, fmt.Errorf("char turn failed: %w", err)
+		}
+		parser = chooseParser(cfg.RemoteAPI)
+		charText, err := parser.ParseBytes(charResp)
+		if err != nil {
+			return nil, fmt.Errorf("char turn failed parsing response: %w", err)
+		}
+		conversation = append(conversation, models.RPMessage{Author: scenario.CharName, Content: charText})
+	}
+	return conversation, nil
+}
+
+func buildRPPrompt(sysPrompt string, history []models.RPMessage, characterName string) string {
+	var b strings.Builder
+	b.WriteString(sysPrompt)
+	b.WriteString("\n\n")
+	for _, msg := range history {
+		b.WriteString(fmt.Sprintf("%s: %s\n", msg.Author, msg.Content))
+	}
+	b.WriteString(fmt.Sprintf("%s:", characterName))
+	return b.String()
 }
 
 func runBench(questions []models.Question) ([]models.Answer, error) {
 	answers := []models.Answer{}
 	for _, q := range questions {
-		resp, err := callLLM(buildPrompt(q.Question))
+		resp, err := callLLM(buildPrompt(q.Question), cfg.RemoteAPI) // Assuming remote for questions
 		if err != nil {
-			// log
-			panic(err)
+			logger.Error("failed to call llm", "error", err)
+			continue
 		}
-		parser := chooseParser()
+		parser := chooseParser(cfg.RemoteAPI)
 		respText, err := parser.ParseBytes(resp)
 		if err != nil {
-			panic(err)
+			logger.Error("failed to parse llm response", "error", err)
+			continue
 		}
-		a := models.Answer{Q: q, Answer: respText, Model: currentModel}
+		a := models.Answer{Q: q, Answer: respText, Model: "default"} // model name from resp?
 		answers = append(answers, a)
 	}
 	return answers, nil
 }
 
-// openai vs completion
 func buildPrompt(q string) string {
-	// sure injection?
-	// completion
 	return fmt.Sprintf(`Q:\n%s\nA:\nSure,`, q)
 }
 
-func callLLM(prompt string) ([]byte, error) {
+func callLLM(prompt string, apiURL string) ([]byte, error) {
 	method := "POST"
-	// Generate the payload once as bytes
-	parser := chooseParser()
+	parser := chooseParser(apiURL)
 	payloadReader := parser.MakePayload(prompt)
+
+	var payloadBytes []byte
+	if payloadReader != nil {
+		var err error
+		payloadBytes, err = io.ReadAll(payloadReader)
+		if err != nil {
+			return nil, fmt.Errorf("failed to read payload: %w", err)
+		}
+	}
+
 	client := &http.Client{}
 	maxRetries := 6
-	baseDelay := 2 // seconds
+	baseDelay := 2 * time.Second
+
 	for attempt := 0; attempt < maxRetries; attempt++ {
-		// Create a new request for the attempt
-		req, err := http.NewRequest(method, cfg.CurrentAPI, payloadReader)
+		req, err := http.NewRequest(method, apiURL, bytes.NewReader(payloadBytes))
 		if err != nil {
-			if attempt == maxRetries-1 {
-				return nil, fmt.Errorf("LLM call failed after %d retries on request creation: %w", maxRetries, err)
-			}
-			logger.Error("failed to make new request; will retry", "error", err, "url", cfg.CurrentAPI, "attempt", attempt)
-			time.Sleep(time.Duration(baseDelay) * time.Second * time.Duration(attempt+1))
-			continue
+			return nil, fmt.Errorf("failed to create request: %w", err)
 		}
+
 		req.Header.Add("Content-Type", "application/json")
 		req.Header.Add("Accept", "application/json")
+		if apiURL == cfg.RemoteAPI {
 			req.Header.Add("Authorization", "Bearer "+cfg.APIToken)
+		}
+
 		resp, err := client.Do(req)
 		if err != nil {
 			if attempt == maxRetries-1 {
 				return nil, fmt.Errorf("LLM call failed after %d retries on client.Do: %w", maxRetries, err)
 			}
-			logger.Error("http request failed; will retry", "error", err, "url", cfg.CurrentAPI, "attempt", attempt)
-			delay := time.Duration(baseDelay*(attempt+1)) * time.Second
-			time.Sleep(delay)
+			logger.Error("http request failed; will retry", "error", err, "url", apiURL, "attempt", attempt)
+			time.Sleep(baseDelay * time.Duration(1<<attempt))
 			continue
 		}
+
 		body, err := io.ReadAll(resp.Body)
 		resp.Body.Close()
 		if err != nil {
-			if attempt == maxRetries-1 {
-				return nil, fmt.Errorf("LLM call failed after %d retries on reading body: %w", maxRetries, err)
+			return nil, fmt.Errorf("failed to read response body: %w", err)
 		}
-			logger.Error("failed to read response body; will retry", "error", err, "url", cfg.CurrentAPI, "attempt", attempt)
-			delay := time.Duration(baseDelay*(attempt+1)) * time.Second
-			time.Sleep(delay)
+
+		if resp.StatusCode >= 400 {
+			if attempt == maxRetries-1 {
+				return nil, fmt.Errorf("LLM call failed after %d retries, got status %d, body: %s", maxRetries, resp.StatusCode, string(body))
+			}
+			logger.Warn("retriable status code; will retry", "code", resp.StatusCode, "attempt", attempt, "body", string(body))
+			time.Sleep(baseDelay * time.Duration(1<<attempt))
 			continue
 		}
-		// Check status code
-		if resp.StatusCode >= 400 && resp.StatusCode < 600 {
-			if attempt == maxRetries-1 {
-				return nil, fmt.Errorf("LLM call failed after %d retries, got status %d", maxRetries, resp.StatusCode)
-			}
-			logger.Warn("retriable status code; will retry", "code", resp.StatusCode, "attempt", attempt)
-			delay := time.Duration((baseDelay * (1 << attempt))) * time.Second
-			time.Sleep(delay)
-			continue
-		}
-		if resp.StatusCode != http.StatusOK {
-			// For non-retriable errors, return immediately
-			return nil, fmt.Errorf("non-retriable status %d, body: %s", resp.StatusCode, string(body))
-		}
-		// Success
-		logger.Debug("llm resp", "body", string(body), "url", cfg.CurrentAPI, "attempt", attempt)
+
+		logger.Debug("llm resp", "body", string(body), "url", apiURL, "attempt", attempt)
 		return body, nil
 	}
-	return nil, errors.New("unknown error in retry loop")
+	return nil, errors.New("exceeded max retries")
 }

models/models.go

@@ -67,3 +67,21 @@ type LLMResp struct {
 	StoppingWord    string `json:"stopping_word"`
 	TokensCached    int    `json:"tokens_cached"`
 }
+
+type LCPResp struct {
+	Content string `json:"content"`
+}
+
+type RPScenario struct {
+	Name          string `json:"name"`
+	CharSysPrompt string `json:"char_sys_prompt"`
+	UserSysPrompt string `json:"user_sys_prompt"`
+	CharName      string `json:"char_name"`
+	UserName      string `json:"user_name"`
+	FirstMsg      string `json:"first_msg"`
+}
+
+type RPMessage struct {
+	Author  string `json:"author"`
+	Content string `json:"content"`
+}

parser.go

@@ -105,7 +105,7 @@ func NewOpenRouterParser(log *slog.Logger) *openRouterParser {
 
 func (p *openRouterParser) ParseBytes(body []byte) (string, error) {
 	// parsing logic here
-	resp := models.OpenRouterResp{}
+	resp := models.DSResp{}
 	if err := json.Unmarshal(body, &resp); err != nil {
 		p.log.Error("failed to unmarshal", "error", err)
 		return "", err
@@ -115,33 +115,25 @@ func (p *openRouterParser) ParseBytes(body []byte) (string, error) {
 		err := errors.New("empty choices in resp")
 		return "", err
 	}
-	text := resp.Choices[0].Message.Content
+	text := resp.Choices[0].Text
 	return text, nil
 }
 
 func (p *openRouterParser) MakePayload(prompt string) io.Reader {
 	// Models to rotate through
 	models := []string{
-		"google/gemini-2.0-flash-exp:free",
-		"deepseek/deepseek-chat-v3-0324:free",
-		"mistralai/mistral-small-3.2-24b-instruct:free",
-		"qwen/qwen3-14b:free",
-		"deepseek/deepseek-r1:free",
-		"google/gemma-3-27b-it:free",
-		"meta-llama/llama-3.3-70b-instruct:free",
+		"google/gemini-flash-1.5",
+		"deepseek/deepseek-coder",
+		"mistralai/mistral-7b-instruct",
+		"qwen/qwen-72b-chat",
+		"meta-llama/llama-3-8b-instruct",
 	}
 	// Get next model index using atomic addition for thread safety
 	p.modelIndex++
 	model := models[int(p.modelIndex)%len(models)]
 	strPayload := fmt.Sprintf(`{
 	"model": "%s",
-	"max_tokens": 300,
-	"messages": [
-		{
-		"role": "user",
-		"content": "%s"
-		}
-	]
+	"prompt": "%s"
 	}`, model, prompt)
 	p.log.Debug("made openrouter payload", "model", model, "payload", strPayload)
 	return strings.NewReader(strPayload)