Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →miosa-osa/self-consistency
Sample multiple paths, select most consistent - +17.9% on GSM8K
Passed 19 of 21 test cases with the skill, and 14 of 21 without it.
and 4 more in the full report
5 fixed15 passed either way2 still failing0 broke
Tokens, turns and time describe resource use, not a dollar price.
One of 4 test cases the skill fixed. Same model, same prompt, with and without the skill.
What percentage accuracy improvement was demonstrated when applying self-consistency to the AQuA multiple-choice algebra dataset over standard chain-of-thought?
| Checked | Without skill | With skill |
|---|---|---|
| Reports an accuracy improvement of approximately 12.2 percent. | Missed | Passed |
The instructions your model reads when the skill loads.
Sample multiple reasoning paths and select the most consistent answer. Achieved +17.9% on GSM8K, +12.2% on AQuA.
Use for:
Generate N (typically 5-10) reasoning paths with higher temperature.
Parse the conclusive answer from each path.
Select the most frequent answer across all paths.
def self_consistency(prompt, question, num_samples=5):
answers = []
for _ in range(num_samples):
response = reason_with_cot(prompt, question)
answer = extract_final_answer(response)
answers.append(answer)
# Majority vote
from collections import Counter
return Counter(answers).most_common(1)[0][0]
test cases passed, vs 14 / 21 without
More skills from the same source and nearby in the registry, with their headline benchmark lift.