File size: 8,193 Bytes
b9f590a
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
"""
run_baseline.py β€” Run a Groq-powered agent against all 3 task tiers of the
Customer Support OpenEnv and record scores.

Usage:
    # Start the server first:
    #   uvicorn server.app:app --host 0.0.0.0 --port 7860
    #
    # Then run:
    #   python run_baseline.py

Environment variables:
    GROQ_API_KEY     β€” Required. Your Groq API key.
    ENV_BASE_URL     β€” Optional. Defaults to http://localhost:7860.
"""

import json
import os
import sys
from typing import Any, Dict, List

import httpx
from groq import Groq
from dotenv import load_dotenv

# Load variables from .env if present
load_dotenv()

# ------------------------------------------------------------------ #
#  Configuration
# ------------------------------------------------------------------ #

GROQ_API_KEY = os.environ.get("GROQ_API_KEY")
if not GROQ_API_KEY:
    print("ERROR: GROQ_API_KEY environment variable is not set.")
    sys.exit(1)

ENV_BASE_URL = os.environ.get("ENV_BASE_URL", "http://localhost:7860")
MODEL = "llama-3.1-8b-instant"
EPISODES_PER_TASK = 5

# ------------------------------------------------------------------ #
#  System prompt
# ------------------------------------------------------------------ #

SYSTEM_PROMPT = """\
You are a professional customer support agent. Your job is to help customers \
resolve their issues efficiently and politely.

For the EASY task: Read the customer message and reply with ONLY the category label.
  Valid categories are: refund, technical, shipping, billing, account

For the MEDIUM task: Write a single, complete, helpful reply that addresses the \
customer's issue.
  Include specific actions you are taking (e.g. "I have initiated a refund...").
  Keep it under 150 words.

For the HARD task (multi-turn):
  - Turn 1: Ask ONE clarifying question to better understand the issue.
  - Turn 2: Provide a concrete solution based on what the customer told you.
  - Turn 3: Close the conversation politely \
(e.g. "Happy to help! Is there anything else I can assist you with?")
"""

# ------------------------------------------------------------------ #
#  Groq client
# ------------------------------------------------------------------ #

ai_client = Groq(api_key=GROQ_API_KEY)


def get_agent_reply(conversation: List[str], task_name: str, turn: int) -> str:
    """Ask Groq for the next agent reply.

    Args:
        conversation: Full conversation history so far.
        task_name: Current task tier (easy, medium, hard).
        turn: Current turn number (1-indexed).

    Returns:
        The agent's text reply.
    """
    # Build the chat messages from conversation history
    messages = [{"role": "system", "content": SYSTEM_PROMPT}]
    
    for i, msg in enumerate(conversation):
        role = "user" if i % 2 == 0 else "assistant"
        messages.append({"role": role, "content": msg})

    # Add a turn-specific hint for hard tasks
    if task_name == "hard":
        hints = {
            1: "This is turn 1. Ask a clarifying question.",
            2: "This is turn 2. Provide a concrete solution.",
            3: "This is turn 3. Close the conversation politely.",
        }
        hint = hints.get(turn, "Continue the conversation appropriately.")
        messages.append({"role": "system", "content": f"[HINT FOR THIS TURN: {hint}]"})

    try:
        response = ai_client.chat.completions.create(
            model=MODEL,
            messages=messages,
            temperature=0.3,
            max_tokens=300,
        )
        return response.choices[0].message.content.strip()
    except Exception as e:
        print(f"  [Groq error] {e}")
        return "I apologize for the inconvenience. Let me help you with that."


# ------------------------------------------------------------------ #
#  Environment API helpers
# ------------------------------------------------------------------ #

def env_reset(client: httpx.Client, task_name: str, seed: int) -> Dict[str, Any]:
    """POST /reset β€” start a new episode."""
    response = client.post(
        f"{ENV_BASE_URL}/reset",
        json={"task_name": task_name, "seed": seed},
    )
    response.raise_for_status()
    return response.json()


def env_step(client: httpx.Client, session_id: str, message: str, intent: str = None) -> Dict[str, Any]:
    """POST /step β€” submit an agent action."""
    payload = {"session_id": session_id, "message": message}
    if intent:
        payload["intent"] = intent

    response = client.post(f"{ENV_BASE_URL}/step", json=payload)
    response.raise_for_status()
    return response.json()


# ------------------------------------------------------------------ #
#  Run episodes
# ------------------------------------------------------------------ #

def run_task(client: httpx.Client, task_name: str) -> List[float]:
    """Run EPISODES_PER_TASK episodes for a given task tier."""
    rewards = []

    for ep in range(EPISODES_PER_TASK):
        try:
            reset_data = env_reset(client, task_name, seed=ep)
            session_id = reset_data["session_id"]
            obs = reset_data.get("observation", {})

            done = obs.get("done", False)
            reward = obs.get("reward", None)
            turn = 0

            while not done:
                turn += 1
                conversation = obs.get("conversation", [])

                # Get the agent's reply from Groq
                agent_reply = get_agent_reply(conversation, task_name, turn)

                if task_name == "easy":
                    intent = "classify"
                elif task_name == "medium":
                    intent = "respond"
                else:
                    intent_map = {1: "clarify", 2: "respond", 3: "close"}
                    intent = intent_map.get(turn, "respond")

                step_data = env_step(client, session_id, agent_reply, intent)
                obs = step_data.get("observation", {})
                done = obs.get("done", False)
                reward = obs.get("reward", None)

                if turn >= 15:
                    print(f"  [Warning] Episode {ep + 1} exceeded 15 turns, breaking.")
                    break

            episode_reward = reward if reward is not None else 0.0
            rewards.append(episode_reward)
            print(f"  Episode {ep + 1}/{EPISODES_PER_TASK}: reward = {episode_reward:.2f}")

        except Exception as e:
            print(f"  Episode {ep + 1}/{EPISODES_PER_TASK}: ERROR β€” {e}")
            rewards.append(0.0)

    return rewards


# ------------------------------------------------------------------ #
#  Main
# ------------------------------------------------------------------ #

def main():
    print("=" * 60)
    print("  Customer Support OpenEnv β€” Baseline Evaluation")
    print(f"  Model: {MODEL}")
    print(f"  Server: {ENV_BASE_URL}")
    print(f"  Episodes per task: {EPISODES_PER_TASK}")
    print("=" * 60)

    results = {}

    with httpx.Client(timeout=60.0) as client:
        for task_name in ["easy", "medium", "hard"]:
            print(f"\n{'─' * 40}")
            print(f"  Task: {task_name.upper()}")
            print(f"{'─' * 40}")

            rewards = run_task(client, task_name)
            avg_reward = sum(rewards) / len(rewards) if rewards else 0.0
            results[task_name] = {
                "average_score": round(avg_reward, 4),
                "scores": [round(r, 4) for r in rewards],
                "episodes": len(rewards),
            }

    print(f"\n{'=' * 60}")
    print("  RESULTS SUMMARY")
    print(f"{'=' * 60}")
    print(f"  {'Task':<12} {'Avg Score':<12} {'Episodes':<10} {'Scores'}")
    print(f"  {'─' * 50}")
    for task_name in ["easy", "medium", "hard"]:
        r = results[task_name]
        scores_str = ", ".join(f"{s:.2f}" for s in r["scores"])
        print(f"  {task_name:<12} {r['average_score']:<12.4f} {r['episodes']:<10} [{scores_str}]")
    print(f"{'=' * 60}\n")

    output_path = os.path.join(os.path.dirname(__file__), "baseline_scores.json")
    with open(output_path, "w", encoding="utf-8") as f:
        json.dump(results, f, indent=2)
    print(f"  Results saved to: {output_path}")


if __name__ == "__main__":
    main()