MCPcopy Index your code
hub / github.com/algorithmicsuperintelligence/optillm / calc_stats

Function calc_stats

scripts/eval_aime_benchmark.py:522–552  ·  view source on GitHub ↗
(attempts)

Source from the content-addressed store, hash-verified

520
521 # Function to calculate statistics for a group of attempts
522 def calc_stats(attempts):
523 if not attempts:
524 return {
525 "count": 0,
526 "avg_thinking_tokens": 0,
527 "avg_thought_transitions": 0,
528 "transition_usage": {phrase: 0 for phrase in THOUGHT_TRANSITIONS},
529 "has_think_tags_pct": 0
530 }
531
532 thinking_tokens = [a['thought_analysis']['thinking_tokens'] for a in attempts]
533 thought_transitions = [a['thought_analysis']['thought_transitions'] for a in attempts]
534 has_think_tags = sum(1 for a in attempts if a['thought_analysis']['has_think_tags'])
535
536 # Count total transition usage
537 transition_usage = defaultdict(int)
538 for attempt in attempts:
539 for phrase, count in attempt['thought_analysis']['transition_counts'].items():
540 transition_usage[phrase] += count
541
542 return {
543 "count": len(attempts),
544 "avg_thinking_tokens": statistics.mean(thinking_tokens) if thinking_tokens else 0,
545 "median_thinking_tokens": statistics.median(thinking_tokens) if thinking_tokens else 0,
546 "min_thinking_tokens": min(thinking_tokens) if thinking_tokens else 0,
547 "max_thinking_tokens": max(thinking_tokens) if thinking_tokens else 0,
548 "avg_thought_transitions": statistics.mean(thought_transitions) if thought_transitions else 0,
549 "median_thought_transitions": statistics.median(thought_transitions) if thought_transitions else 0,
550 "transition_usage": dict(transition_usage),
551 "has_think_tags_pct": (has_think_tags / len(attempts)) * 100 if attempts else 0
552 }
553
554 # Calculate statistics
555 correct_stats = calc_stats(correct_attempts)

Callers 1

analyze_resultsFunction · 0.85

Calls

no outgoing calls

Tested by

no test coverage detected