mirror of
https://github.com/NousResearch/atropos.git
synced 2026-04-19 12:57:58 +00:00
535 lines
19 KiB
Python
535 lines
19 KiB
Python
"""
|
|
MATH-500 Evaluation Environment for Atropos (Generative Mode)
|
|
|
|
This environment evaluates models on MATH-500 - a subset of 500 problems from
|
|
the MATH benchmark that OpenAI created for their "Let's Verify Step by Step" paper.
|
|
|
|
Dataset: HuggingFaceH4/MATH-500
|
|
Paper: https://arxiv.org/abs/2305.20050
|
|
|
|
The evaluation follows a generative approach:
|
|
- Models receive challenging math problems
|
|
- Expected to provide step-by-step reasoning
|
|
- Final answer in \\boxed{} format
|
|
- Uses math_verify for robust answer verification
|
|
- Falls back to string normalization if symbolic comparison fails
|
|
|
|
Supports thinking mode with <think></think> tags for extended reasoning.
|
|
"""
|
|
|
|
import asyncio
|
|
import os
|
|
import random
|
|
import re
|
|
import time
|
|
from concurrent.futures import ProcessPoolExecutor
|
|
from typing import Dict, List, Optional, Tuple
|
|
|
|
import wandb
|
|
from datasets import load_dataset
|
|
from eval_helpers import (
|
|
THINK_CONTENT_AFTER_PATTERN,
|
|
create_system_content,
|
|
extract_boxed_answers,
|
|
extract_thinking_content,
|
|
format_math_answer_instruction,
|
|
get_default_thinking_prompt,
|
|
get_math_executor,
|
|
save_eval_results,
|
|
score_math_answer_async,
|
|
validate_thinking_format,
|
|
)
|
|
from pydantic import Field
|
|
from tqdm.asyncio import tqdm_asyncio
|
|
|
|
from atroposlib.envs.base import (
|
|
APIServerConfig,
|
|
BaseEnv,
|
|
BaseEnvConfig,
|
|
EvalHandlingEnum,
|
|
)
|
|
|
|
# Prompt template following lighteval's MATH-500 structure
|
|
MATH500_PROMPT_TEMPLATE = """Solve the following problem. The final line of your response MUST be of the following format:
|
|
"ANSWER: $ANSWER" (without quotes) where $ANSWER is the final answer. Think step by step before answering.
|
|
|
|
However, for reliable parsing, also put your final answer in \\boxed{{}} format.
|
|
|
|
{problem}"""
|
|
|
|
|
|
# Alternative prompt that just uses boxed (matches our standard)
|
|
MATH500_BOXED_PROMPT_TEMPLATE = """Solve the following math problem. {answer_instruction}
|
|
|
|
{problem}"""
|
|
|
|
|
|
class MATH500EvalConfig(BaseEnvConfig):
|
|
"""Configuration for MATH-500 evaluation environment."""
|
|
|
|
# Dataset configuration
|
|
dataset_name: str = Field(
|
|
default="HuggingFaceH4/MATH-500", description="HuggingFace dataset name"
|
|
)
|
|
subset: str = Field(default="default", description="Dataset subset")
|
|
eval_split: str = Field(default="test", description="Split to evaluate on")
|
|
shuffle_seed: int = Field(default=42, description="Random seed for shuffling")
|
|
|
|
# Generation parameters
|
|
eval_temperature: float = Field(
|
|
default=0.6, description="Temperature for evaluation generation"
|
|
)
|
|
eval_max_tokens: int = Field(
|
|
default=0, description="Max tokens for evaluation (0 = use model default)"
|
|
)
|
|
|
|
# System prompt configuration
|
|
custom_system_prompt: Optional[str] = Field(
|
|
default=None, description="Optional custom system prompt"
|
|
)
|
|
|
|
# Thinking mode configuration
|
|
thinking_mode: bool = Field(
|
|
default=True,
|
|
description="Whether to use thinking mode with <think></think> tags",
|
|
)
|
|
custom_thinking_prompt: Optional[str] = Field(
|
|
default=None, description="Optional custom thinking prompt"
|
|
)
|
|
|
|
# Math verification configuration
|
|
include_hope_suffix: bool = Field(
|
|
default=True,
|
|
description="Whether to include 'I hope it is correct' in answer instruction",
|
|
)
|
|
use_original_prompt: bool = Field(
|
|
default=False,
|
|
description="Use lighteval's original prompt format (ANSWER: format)",
|
|
)
|
|
max_math_workers: int = Field(
|
|
default=64,
|
|
description="Maximum workers for math verification ProcessPoolExecutor",
|
|
)
|
|
|
|
# Retry and debug configuration
|
|
max_retries: int = Field(
|
|
default=3, description="Maximum retries for failed API calls"
|
|
)
|
|
retry_delay: float = Field(
|
|
default=1.0, description="Delay between retries in seconds"
|
|
)
|
|
min_response_length: int = Field(
|
|
default=1, description="Minimum response length to consider valid"
|
|
)
|
|
full_debug: bool = Field(default=False, description="Enable full debug output")
|
|
|
|
# Override defaults
|
|
group_size: int = 1
|
|
max_num_workers: int = 1024
|
|
max_eval_workers: int = 256
|
|
max_num_workers_per_node: int = 128
|
|
use_wandb: bool = True
|
|
rollout_server_url: str = "http://localhost:8000"
|
|
total_steps: int = 1
|
|
wandb_name: str = "math500_eval"
|
|
steps_per_eval: int = 1
|
|
|
|
|
|
class MATH500EvalEnv(BaseEnv):
|
|
"""
|
|
MATH-500 Evaluation Environment.
|
|
|
|
Evaluates challenging math problem solving using the MATH-500 dataset.
|
|
Uses math_verify for robust answer verification with string fallback.
|
|
"""
|
|
|
|
name = "math500_eval"
|
|
|
|
def __init__(
|
|
self,
|
|
config: MATH500EvalConfig,
|
|
server_configs: List[APIServerConfig],
|
|
slurm_job_id: Optional[str] = None,
|
|
testing: bool = False,
|
|
):
|
|
super().__init__(config, server_configs, slurm_job_id, testing)
|
|
self.config: MATH500EvalConfig = config
|
|
self.eval_items: List[Dict] = []
|
|
self._dataset_loaded = False
|
|
self._math_executor: Optional[ProcessPoolExecutor] = None
|
|
|
|
@classmethod
|
|
def config_cls(cls) -> type:
|
|
return MATH500EvalConfig
|
|
|
|
async def setup(self) -> None:
|
|
"""Initialize the environment and load the dataset."""
|
|
await super().setup()
|
|
|
|
# Initialize math executor
|
|
self._math_executor = get_math_executor(self.config.max_math_workers)
|
|
|
|
if not self._dataset_loaded:
|
|
await self._load_dataset()
|
|
|
|
print(f"\nMATH-500 Evaluation Setup (Generative Mode):")
|
|
print(f" Dataset: {self.config.dataset_name}")
|
|
print(f" Evaluation split: {self.config.eval_split}")
|
|
print(f" Thinking mode: {self.config.thinking_mode}")
|
|
if self.config.thinking_mode:
|
|
thinking_prompt = get_default_thinking_prompt(
|
|
self.config.custom_thinking_prompt
|
|
)
|
|
print(f" Thinking prompt: {thinking_prompt[:80]}...")
|
|
print(f" Loaded {len(self.eval_items)} evaluation items")
|
|
|
|
async def _load_dataset(self) -> None:
|
|
"""Load and process the MATH-500 dataset."""
|
|
print(f"Loading MATH-500 dataset: {self.config.dataset_name}...")
|
|
|
|
try:
|
|
dataset = load_dataset(
|
|
self.config.dataset_name,
|
|
self.config.subset if self.config.subset != "default" else None,
|
|
trust_remote_code=True,
|
|
)
|
|
except Exception as e:
|
|
print(f"Error loading dataset: {e}")
|
|
raise
|
|
|
|
if self.config.eval_split not in dataset:
|
|
available_splits = list(dataset.keys())
|
|
raise ValueError(
|
|
f"Split '{self.config.eval_split}' not found. Available: {available_splits}"
|
|
)
|
|
|
|
split_data = dataset[self.config.eval_split]
|
|
|
|
# Process items
|
|
self.eval_items = []
|
|
for idx, item in enumerate(split_data):
|
|
problem = item.get("problem", "")
|
|
answer = item.get("answer", "") # MATH-500 has 'answer' field
|
|
solution = item.get("solution", "") # May also have solution
|
|
|
|
# Extract final answer
|
|
if answer:
|
|
final_answer = answer.strip()
|
|
elif solution:
|
|
# Try to extract from solution
|
|
boxed = extract_boxed_answers(solution)
|
|
final_answer = boxed[-1] if boxed else solution.strip()
|
|
else:
|
|
final_answer = ""
|
|
|
|
subject = item.get("subject", "unknown")
|
|
level = item.get("level", "")
|
|
unique_id = item.get("unique_id", str(idx))
|
|
|
|
self.eval_items.append(
|
|
{
|
|
"id": unique_id,
|
|
"problem": problem,
|
|
"answer": final_answer,
|
|
"solution": solution,
|
|
"subject": subject,
|
|
"level": level,
|
|
}
|
|
)
|
|
|
|
# Shuffle with seed
|
|
random.seed(self.config.shuffle_seed)
|
|
random.shuffle(self.eval_items)
|
|
|
|
self._dataset_loaded = True
|
|
print(f"Loaded {len(self.eval_items)} evaluation items")
|
|
|
|
def _format_prompt(self, item: Dict) -> str:
|
|
"""Format the problem into a prompt."""
|
|
if self.config.use_original_prompt:
|
|
return MATH500_PROMPT_TEMPLATE.format(problem=item["problem"])
|
|
else:
|
|
answer_instruction = format_math_answer_instruction(
|
|
include_hope=self.config.include_hope_suffix
|
|
)
|
|
return MATH500_BOXED_PROMPT_TEMPLATE.format(
|
|
answer_instruction=answer_instruction, problem=item["problem"]
|
|
)
|
|
|
|
def _create_system_content(self) -> str:
|
|
"""Create system message content based on thinking mode."""
|
|
return (
|
|
create_system_content(
|
|
self.config.thinking_mode,
|
|
self.config.custom_thinking_prompt,
|
|
self.config.custom_system_prompt,
|
|
)
|
|
or ""
|
|
)
|
|
|
|
async def rollout_and_score_eval(
|
|
self,
|
|
item: Dict,
|
|
server: APIServerConfig,
|
|
) -> Optional[Dict]:
|
|
"""Run evaluation on a single item and return the result."""
|
|
prompt = self._format_prompt(item)
|
|
system_content = self._create_system_content()
|
|
|
|
messages = []
|
|
if system_content:
|
|
messages.append({"role": "system", "content": system_content})
|
|
messages.append({"role": "user", "content": prompt})
|
|
|
|
# Build API call parameters
|
|
kwargs = {
|
|
"model": server.model_name,
|
|
"messages": messages,
|
|
"temperature": self.config.eval_temperature,
|
|
}
|
|
if self.config.eval_max_tokens > 0:
|
|
kwargs["max_tokens"] = self.config.eval_max_tokens
|
|
|
|
response_text = ""
|
|
for attempt in range(self.config.max_retries):
|
|
try:
|
|
response = await self.server.chat_completion(**kwargs)
|
|
response_text = response.choices[0].message.content or ""
|
|
|
|
if len(response_text) >= self.config.min_response_length:
|
|
break
|
|
|
|
except Exception as e:
|
|
if self.config.full_debug:
|
|
print(f" API error (attempt {attempt + 1}): {e}")
|
|
if attempt < self.config.max_retries - 1:
|
|
await asyncio.sleep(self.config.retry_delay)
|
|
continue
|
|
|
|
if not response_text:
|
|
return None
|
|
|
|
# Validate thinking format
|
|
is_valid_format, content_for_extraction = validate_thinking_format(
|
|
response_text, self.config.thinking_mode
|
|
)
|
|
|
|
# Extract thinking content if present
|
|
thinking_content = (
|
|
extract_thinking_content(response_text)
|
|
if self.config.thinking_mode
|
|
else None
|
|
)
|
|
|
|
# Score using math_verify with string fallback
|
|
gold_answer = item["answer"]
|
|
is_correct, method, has_multiple_boxed = await score_math_answer_async(
|
|
gold=gold_answer,
|
|
response=response_text,
|
|
after_think=self.config.thinking_mode,
|
|
wrap_gold_boxed=True,
|
|
executor=self._math_executor,
|
|
debug=self.config.full_debug,
|
|
)
|
|
|
|
# Extract the boxed answer for logging
|
|
if self.config.thinking_mode:
|
|
match = THINK_CONTENT_AFTER_PATTERN.search(response_text)
|
|
score_content = match.group(1) if match else response_text
|
|
else:
|
|
score_content = response_text
|
|
|
|
boxed_answers = extract_boxed_answers(score_content)
|
|
extracted_answer = boxed_answers[0] if boxed_answers else None
|
|
|
|
if self.config.full_debug:
|
|
print(f"\n--- Item: {item['id']} ---")
|
|
print(
|
|
f"Subject: {item.get('subject', 'N/A')}, Level: {item.get('level', 'N/A')}"
|
|
)
|
|
print(f"Problem: {item['problem'][:100]}...")
|
|
print(f"Gold answer: {gold_answer}")
|
|
print(f"Extracted: {extracted_answer}")
|
|
print(f"Correct: {is_correct} (method: {method})")
|
|
|
|
return {
|
|
"item_id": item["id"],
|
|
"subject": item.get("subject", "unknown"),
|
|
"level": item.get("level", ""),
|
|
"problem": item["problem"][:200],
|
|
"gold_answer": gold_answer,
|
|
"extracted_answer": extracted_answer,
|
|
"verification_method": method,
|
|
"is_correct": is_correct if is_correct is not None else False,
|
|
"has_multiple_boxed": has_multiple_boxed,
|
|
"format_valid": is_valid_format,
|
|
"response": response_text,
|
|
"thinking_content": thinking_content,
|
|
"has_thinking": thinking_content is not None,
|
|
}
|
|
|
|
async def evaluate(self, *args, **kwargs) -> Dict:
|
|
"""Run the full MATH-500 evaluation."""
|
|
print(f"\n{'='*60}")
|
|
print("Starting MATH-500 Evaluation (Generative Mode)")
|
|
print(f"{'='*60}")
|
|
print(f" Total questions: {len(self.eval_items)}")
|
|
print(f" Thinking mode: {self.config.thinking_mode}")
|
|
print(f"{'='*60}\n")
|
|
|
|
# Create evaluation tasks
|
|
async def eval_task(item):
|
|
return await self.rollout_and_score_eval(item, self.server_configs[0])
|
|
|
|
tasks = [eval_task(item) for item in self.eval_items]
|
|
|
|
# Run with progress bar
|
|
results = await tqdm_asyncio.gather(*tasks, desc="Evaluating MATH-500")
|
|
|
|
# Filter out failed results
|
|
valid_results = [r for r in results if r is not None]
|
|
|
|
if not valid_results:
|
|
print("Warning: No valid evaluation results obtained")
|
|
return {"error": "No valid results", "accuracy": 0.0}
|
|
|
|
# Calculate overall metrics
|
|
total = len(valid_results)
|
|
correct = sum(1 for r in valid_results if r["is_correct"])
|
|
accuracy = correct / total if total > 0 else 0.0
|
|
|
|
# Calculate per-subject metrics
|
|
subject_metrics = {}
|
|
for r in valid_results:
|
|
subject = r.get("subject", "unknown")
|
|
if subject not in subject_metrics:
|
|
subject_metrics[subject] = {"total": 0, "correct": 0}
|
|
subject_metrics[subject]["total"] += 1
|
|
if r["is_correct"]:
|
|
subject_metrics[subject]["correct"] += 1
|
|
|
|
for subject in subject_metrics:
|
|
s_total = subject_metrics[subject]["total"]
|
|
s_correct = subject_metrics[subject]["correct"]
|
|
subject_metrics[subject]["accuracy"] = (
|
|
s_correct / s_total if s_total > 0 else 0.0
|
|
)
|
|
|
|
# Calculate per-level metrics
|
|
level_metrics = {}
|
|
for r in valid_results:
|
|
level = r.get("level", "unknown") or "unknown"
|
|
if level not in level_metrics:
|
|
level_metrics[level] = {"total": 0, "correct": 0}
|
|
level_metrics[level]["total"] += 1
|
|
if r["is_correct"]:
|
|
level_metrics[level]["correct"] += 1
|
|
|
|
for level in level_metrics:
|
|
l_total = level_metrics[level]["total"]
|
|
l_correct = level_metrics[level]["correct"]
|
|
level_metrics[level]["accuracy"] = (
|
|
l_correct / l_total if l_total > 0 else 0.0
|
|
)
|
|
|
|
# Count verification methods and other stats
|
|
method_counts = {}
|
|
for r in valid_results:
|
|
method = r.get("verification_method", "unknown")
|
|
method_counts[method] = method_counts.get(method, 0) + 1
|
|
|
|
multiple_boxed = sum(
|
|
1 for r in valid_results if r.get("has_multiple_boxed", False)
|
|
)
|
|
format_valid = sum(1 for r in valid_results if r.get("format_valid", True))
|
|
has_thinking = sum(1 for r in valid_results if r.get("has_thinking", False))
|
|
has_boxed = sum(
|
|
1 for r in valid_results if r.get("extracted_answer") is not None
|
|
)
|
|
|
|
metrics = {
|
|
"accuracy": accuracy,
|
|
"total_evaluated": total,
|
|
"total_correct": correct,
|
|
"num_subjects": len(subject_metrics),
|
|
"has_boxed_rate": has_boxed / total if total > 0 else 0.0,
|
|
"multiple_boxed_rate": multiple_boxed / total if total > 0 else 0.0,
|
|
"format_compliance_rate": format_valid / total if total > 0 else 0.0,
|
|
"thinking_utilization_rate": has_thinking / total if total > 0 else 0.0,
|
|
"subject_metrics": subject_metrics,
|
|
"level_metrics": level_metrics,
|
|
"verification_methods": method_counts,
|
|
}
|
|
|
|
print(f"\n{'='*60}")
|
|
print("MATH-500 Evaluation Results")
|
|
print(f"{'='*60}")
|
|
print(f" Overall Accuracy: {accuracy:.2%} ({correct}/{total})")
|
|
print(f" Has \\boxed{{}} Rate: {has_boxed / total:.2%}")
|
|
print(f" Format Compliance: {format_valid / total:.2%}")
|
|
if self.config.thinking_mode:
|
|
print(f" Thinking Utilization: {has_thinking / total:.2%}")
|
|
if subject_metrics and len(subject_metrics) > 1:
|
|
print(f"\n Per-Subject Breakdown:")
|
|
for subject, data in sorted(
|
|
subject_metrics.items(), key=lambda x: -x[1]["accuracy"]
|
|
):
|
|
print(
|
|
f" {subject}: {data['accuracy']:.2%} ({data['correct']}/{data['total']})"
|
|
)
|
|
if level_metrics and len(level_metrics) > 1:
|
|
print(f"\n Per-Level Breakdown:")
|
|
for level, data in sorted(level_metrics.items()):
|
|
print(
|
|
f" Level {level}: {data['accuracy']:.2%} ({data['correct']}/{data['total']})"
|
|
)
|
|
print(f"{'='*60}\n")
|
|
|
|
# Save results
|
|
if self.config.data_dir_to_save_evals:
|
|
self._save_results(metrics, valid_results)
|
|
|
|
return metrics
|
|
|
|
def _save_results(self, metrics: Dict, results: List[Dict]) -> None:
|
|
"""Save evaluation results to disk."""
|
|
save_eval_results(self.config.data_dir_to_save_evals, metrics, results)
|
|
|
|
async def wandb_log(self, metrics: Dict, step: int = 0) -> None:
|
|
"""Log metrics to Weights & Biases."""
|
|
if not self.config.use_wandb:
|
|
return
|
|
|
|
log_dict = {
|
|
"math500/accuracy": metrics.get("accuracy", 0),
|
|
"math500/total_evaluated": metrics.get("total_evaluated", 0),
|
|
"math500/has_boxed_rate": metrics.get("has_boxed_rate", 0),
|
|
"math500/format_compliance_rate": metrics.get("format_compliance_rate", 0),
|
|
"math500/thinking_utilization_rate": metrics.get(
|
|
"thinking_utilization_rate", 0
|
|
),
|
|
}
|
|
|
|
# Log per-subject accuracies
|
|
for subject, data in metrics.get("subject_metrics", {}).items():
|
|
safe_name = subject.replace(" ", "_")[:30]
|
|
log_dict[f"math500/accuracy_{safe_name}"] = data.get("accuracy", 0)
|
|
|
|
wandb.log(log_dict, step=step)
|
|
|
|
# Required abstract method implementations
|
|
async def get_next_item(self) -> Optional[Dict]:
|
|
"""Not used in evaluation mode."""
|
|
return None
|
|
|
|
async def collect_trajectories(self, item) -> List:
|
|
"""Not used in evaluation mode."""
|
|
return []
|
|
|
|
async def score(self, rollout_group_data) -> Optional[List]:
|
|
"""Not used in evaluation mode."""
|
|
return None
|
|
|
|
|
|
if __name__ == "__main__":
|
|
MATH500EvalEnv.cli()
|