Datasets:
id stringlengths 16 16 | domain stringclasses 5
values | source stringclasses 25
values | difficulty stringclasses 3
values | messages listlengths 3 3 | ground_truth stringlengths 0 26k | quality_score float64 4.5 10 | teacher_model stringclasses 1
value | completion_tokens int64 34 37.4k | generation_time stringlengths 32 32 |
|---|---|---|---|---|---|---|---|---|---|
743608c8b309e246 | math | NuminaMath-CoT | hard | [
{
"role": "system",
"content": "You are an expert mathematician. Solve the given problem with complete rigor.\n\nFormat your response EXACTLY as:\n<think>\n[Your complete step-by-step reasoning. Show all intermediate calculations.\nIf you make an error and catch it, show the correction.\nEnd with a clear st... | 9 | qwen3.8-max-preview | 482 | 2026-07-21T19:42:23.482750+00:00 | |
535134b5746e4088 | math | MetaMathQA | hard | [
{
"role": "system",
"content": "You are an expert mathematician. Solve the given problem with complete rigor.\n\nFormat your response EXACTLY as:\n<think>\n[Your complete step-by-step reasoning. Show all intermediate calculations.\nIf you make an error and catch it, show the correction.\nEnd with a clear st... | 4 | 10 | qwen3.8-max-preview | 347 | 2026-07-22T06:50:12.625415+00:00 |
a93720435fb46f34 | math | OrcaMath | easy | [
{
"role": "system",
"content": "You are an expert mathematician. Solve the given problem with complete rigor.\n\nFormat your response EXACTLY as:\n<think>\n[Your complete step-by-step reasoning. Show all intermediate calculations.\nIf you make an error and catch it, show the correction.\nEnd with a clear st... | Let's break down the information given:
1. Yoojung is standing on the far right.
2. Eunjeong is standing 5th from the far left.
3. There are 8 students standing between Eunjeong and Yoojung.
To find out the total number of students in line, we need to add up all the positions:
- The 4 students to the left of Eunjeon... | 9 | qwen3.8-max-preview | 438 | 2026-07-21T14:53:29.754950+00:00 |
9c240c26825b0195 | code | Evol-Code | hard | [
{
"role": "system",
"content": "You are an expert software engineer. Write complete, correct, runnable code.\n\nFormat your response EXACTLY as:\n<think>\n[Analyze the problem, consider edge cases, choose your approach,\nexplain your algorithm and data structure choices.]\n</think>\n\n```python\n[Complete, ... | Here is the rewritten code to meet the additional requirements:
```python
def find_max_subarray_sum(A):
if not A or not any(A):
return []
max_sums = []
for row in A:
if all(num < 0 for num in row):
max_sums.append(0)
elif all(num >= 0 for num in row):
max_su... | 9 | qwen3.8-max-preview | 1,968 | 2026-07-22T07:55:31.410226+00:00 |
db47c4070e7a8f2a | code | CodeAlpaca | medium | [
{
"role": "system",
"content": "You are an expert software engineer. Write complete, correct, runnable code.\n\nFormat your response EXACTLY as:\n<think>\n[Analyze the problem, consider edge cases, choose your approach,\nexplain your algorithm and data structure choices.]\n</think>\n\n```python\n[Complete, ... | List<Integer> list = Arrays.asList(5, 2, 8, 1);
Collections.sort(list); | 8.5 | qwen3.8-max-preview | 550 | 2026-07-21T12:58:59.786519+00:00 |
480f2ca42490a4d2 | math | NuminaMath-CoT | hard | [
{
"role": "system",
"content": "You are an expert mathematician. Solve the given problem with complete rigor.\n\nFormat your response EXACTLY as:\n<think>\n[Your complete step-by-step reasoning. Show all intermediate calculations.\nIf you make an error and catch it, show the correction.\nEnd with a clear st... | 7.5 | qwen3.8-max-preview | 1,689 | 2026-07-22T07:45:19.902508+00:00 | |
e51d4f50eb7efb44 | math | OrcaMath | easy | [
{
"role": "system",
"content": "You are an expert mathematician. Solve the given problem with complete rigor.\n\nFormat your response EXACTLY as:\n<think>\n[Your complete step-by-step reasoning. Show all intermediate calculations.\nIf you make an error and catch it, show the correction.\nEnd with a clear st... | First, let's convert all the measurements to milliliters (ml) to make it easier to compare.
1 liter = 1000 milliliters
So, the 1-liter water bottle contains:
1 l = 1000 ml
The 400-milliliter water bottle already is in milliliters.
The 2800-milliliter tank is also already in milliliters.
Now, let's add up the total... | 9 | qwen3.8-max-preview | 517 | 2026-07-22T05:06:50.971772+00:00 |
af3c9ba600a32dee | reasoning | CommonsenseQA | medium | [
{
"role": "system",
"content": "You are an expert logical reasoner. Think through the problem carefully.\n\nFormat your response EXACTLY as:\n<think>\n[Break down the problem into components.\nConsider multiple hypotheses if applicable.\nEliminate incorrect paths explicitly.\nBuild toward the conclusion ste... | C | 9.5 | qwen3.8-max-preview | 483 | 2026-07-21T23:14:58.180986+00:00 |
d5bc20f841aba56c | code | Evol-Code | medium | [
{
"role": "system",
"content": "You are an expert software engineer. Write complete, correct, runnable code.\n\nFormat your response EXACTLY as:\n<think>\n[Analyze the problem, consider edge cases, choose your approach,\nexplain your algorithm and data structure choices.]\n</think>\n\n```python\n[Complete, ... | ```javascript
function compareNumbers(num1, num2, num3) {
if (typeof num1 !== "number" || typeof num2 !== "number" || typeof num3 !== "number") {
return "Error: All inputs must be numbers.";
}
if (num1 >= num2 && num1 >= num3) {
return num1;
} else if (num2 >= num1 && num2 >= num3) {
return num2;... | 8.5 | qwen3.8-max-preview | 708 | 2026-07-22T10:06:17.608724+00:00 |
9b8e6b7233eeae11 | math | NuminaMath-CoT | hard | [
{
"role": "system",
"content": "You are an expert mathematician. Solve the given problem with complete rigor.\n\nFormat your response EXACTLY as:\n<think>\n[Your complete step-by-step reasoning. Show all intermediate calculations.\nIf you make an error and catch it, show the correction.\nEnd with a clear st... | 9 | qwen3.8-max-preview | 769 | 2026-07-21T17:47:42.185975+00:00 | |
d50095471fd52a8a | instruction | tulu-3 | medium | [
{
"role": "system",
"content": "You are a helpful, precise assistant. Follow instructions exactly.\n\nFormat your response as:\n<think>\n[Understand what is being asked. Identify constraints and requirements.\nPlan your response structure.]\n</think>\n\n[Your complete, well-structured response following all... | 6 | qwen3.8-max-preview | 413 | 2026-07-22T03:17:58.795982+00:00 | |
317f6869a9d17cde | reasoning | QASC | hard | [
{
"role": "system",
"content": "You are an expert logical reasoner. Think through the problem carefully.\n\nFormat your response EXACTLY as:\n<think>\n[Break down the problem into components.\nConsider multiple hypotheses if applicable.\nEliminate incorrect paths explicitly.\nBuild toward the conclusion ste... | D | 9.5 | qwen3.8-max-preview | 674 | 2026-07-22T08:32:36.877004+00:00 |
49c883c50dc6e44d | reasoning | ARC-Easy | easy | [
{
"role": "system",
"content": "You are an expert logical reasoner. Think through the problem carefully.\n\nFormat your response EXACTLY as:\n<think>\n[Break down the problem into components.\nConsider multiple hypotheses if applicable.\nEliminate incorrect paths explicitly.\nBuild toward the conclusion ste... | D | 9.5 | qwen3.8-max-preview | 421 | 2026-07-22T04:03:37.263876+00:00 |
db817ca68529d910 | reasoning | SciQ | medium | [
{
"role": "system",
"content": "You are an expert logical reasoner. Think through the problem carefully.\n\nFormat your response EXACTLY as:\n<think>\n[Break down the problem into components.\nConsider multiple hypotheses if applicable.\nEliminate incorrect paths explicitly.\nBuild toward the conclusion ste... | C | 9.5 | qwen3.8-max-preview | 371 | 2026-07-21T13:10:36.777247+00:00 |
e4cf6588c13556b7 | math | NuminaMath-CoT | hard | [
{
"role": "system",
"content": "You are an expert mathematician. Solve the given problem with complete rigor.\n\nFormat your response EXACTLY as:\n<think>\n[Your complete step-by-step reasoning. Show all intermediate calculations.\nIf you make an error and catch it, show the correction.\nEnd with a clear st... | 9 | qwen3.8-max-preview | 508 | 2026-07-21T22:59:12.177072+00:00 | |
ed5ad7d7a3c53109 | math | OrcaMath | easy | [
{
"role": "system",
"content": "You are an expert mathematician. Solve the given problem with complete rigor.\n\nFormat your response EXACTLY as:\n<think>\n[Your complete step-by-step reasoning. Show all intermediate calculations.\nIf you make an error and catch it, show the correction.\nEnd with a clear st... | If you divided a number by 6 and got 9, the original number must have been 54 (because 54 / 6 = 9).
If you should have multiplied the number by 6 instead, you take the original number, which is 54, and multiply it by 6:
54 * 6 = 324
So, when you calculate correctly by multiplying the number by 6, you get 324. | 9 | qwen3.8-max-preview | 199 | 2026-07-21T21:14:04.282013+00:00 |
ffbd8b259f0136f4 | code | Evol-Code | medium | [
{
"role": "system",
"content": "You are an expert software engineer. Write complete, correct, runnable code.\n\nFormat your response EXACTLY as:\n<think>\n[Analyze the problem, consider edge cases, choose your approach,\nexplain your algorithm and data structure choices.]\n</think>\n\n```python\n[Complete, ... | The output of the program will be:
```
c = 11
d = 30
e = -1
f = 0
```
First, let's go through the program step by step.
1. The program defines a function `swap` that takes two pointers to integers and swaps their values. This function is used to swap the values of `a` and `b` in the `main` function.
2. In the `main... | 7 | qwen3.8-max-preview | 830 | 2026-07-21T15:00:01.848676+00:00 |
1bf2558fca4498e5 | instruction | tulu-3 | medium | [
{
"role": "system",
"content": "You are a helpful, precise assistant. Follow instructions exactly.\n\nFormat your response as:\n<think>\n[Understand what is being asked. Identify constraints and requirements.\nPlan your response structure.]\n</think>\n\n[Your complete, well-structured response following all... | 7.5 | qwen3.8-max-preview | 1,122 | 2026-07-21T11:40:36.025067+00:00 | |
9b9003586b2f7ae4 | code | Evol-Code | hard | [
{
"role": "system",
"content": "You are an expert software engineer. Write complete, correct, runnable code.\n\nFormat your response EXACTLY as:\n<think>\n[Analyze the problem, consider edge cases, choose your approach,\nexplain your algorithm and data structure choices.]\n</think>\n\n```python\n[Complete, ... | SELECT c.name, c.email, c.phone
FROM customers c
JOIN purchases p ON c.id = p.customer_id
JOIN products pr ON p.product_id = pr.id
WHERE pr.category = 'specific_category'
GROUP BY c.name, c.email, c.phone
HAVING COUNT(p.purchase_id) >= 3
AND SUM(pr.price) > 100
AND SUM(pr.price) * COUNT(p.purchase_id) > 500; | 8.5 | qwen3.8-max-preview | 1,095 | 2026-07-21T16:44:21.691820+00:00 |
8189ae303247f51a | code | CodeAlpaca | medium | [
{
"role": "system",
"content": "You are an expert software engineer. Write complete, correct, runnable code.\n\nFormat your response EXACTLY as:\n<think>\n[Analyze the problem, consider edge cases, choose your approach,\nexplain your algorithm and data structure choices.]\n</think>\n\n```python\n[Complete, ... | if (num > 0):
print("The number is positive")
else:
print("The number is negative") | 9 | qwen3.8-max-preview | 267 | 2026-07-22T01:11:56.807080+00:00 |
bf8a43a05ec5c83a | math | MATH/algebra | hard | [
{
"role": "system",
"content": "You are an expert mathematician. Solve the given problem with complete rigor.\n\nFormat your response EXACTLY as:\n<think>\n[Your complete step-by-step reasoning. Show all intermediate calculations.\nIf you make an error and catch it, show the correction.\nEnd with a clear st... | Applying the midpoint formula gives us $$\left(\frac{-9+x}{2},\frac{1+y}{2}\right)=(3,-5).$$Solving $\frac{-9+x}{2} =3$ for $x$ and solving $\frac{1+y}{2}=-5$ for $y$, we find the coordinates $(x,y)$ to be $\boxed{(15,-11)}$. | 9 | qwen3.8-max-preview | 487 | 2026-07-22T03:43:42.011760+00:00 |
86bcbdc1ce774947 | reasoning | QASC | hard | [
{
"role": "system",
"content": "You are an expert logical reasoner. Think through the problem carefully.\n\nFormat your response EXACTLY as:\n<think>\n[Break down the problem into components.\nConsider multiple hypotheses if applicable.\nEliminate incorrect paths explicitly.\nBuild toward the conclusion ste... | E | 9.5 | qwen3.8-max-preview | 857 | 2026-07-22T10:24:36.385841+00:00 |
9bc5547dc289273c | math | MetaMathQA | easy | [
{
"role": "system",
"content": "You are an expert mathematician. Solve the given problem with complete rigor.\n\nFormat your response EXACTLY as:\n<think>\n[Your complete step-by-step reasoning. Show all intermediate calculations.\nIf you make an error and catch it, show the correction.\nEnd with a clear st... | 1000 | 10 | qwen3.8-max-preview | 285 | 2026-07-21T20:45:17.025212+00:00 |
a1a1f9e97a713261 | instruction | tulu-3 | medium | [
{
"role": "system",
"content": "You are a helpful, precise assistant. Follow instructions exactly.\n\nFormat your response as:\n<think>\n[Understand what is being asked. Identify constraints and requirements.\nPlan your response structure.]\n</think>\n\n[Your complete, well-structured response following all... | 6 | qwen3.8-max-preview | 241 | 2026-07-21T20:45:09.683143+00:00 | |
cfbd2b75ed5d569a | code | Evol-Code | medium | [
{
"role": "system",
"content": "You are an expert software engineer. Write complete, correct, runnable code.\n\nFormat your response EXACTLY as:\n<think>\n[Analyze the problem, consider edge cases, choose your approach,\nexplain your algorithm and data structure choices.]\n</think>\n\n```python\n[Complete, ... | Here is the corrected code:
```
import matplotlib.pyplot as plt
def bisection_method(f, a, b, tolerance):
while abs(a - b) > tolerance:
c = (a + b) / 2
if f(c) == 0:
return c
elif f(a) * f(c) < 0:
b = c
else:
a = c
return (a + b) / 2
def der... | 7.5 | qwen3.8-max-preview | 3,080 | 2026-07-21T20:25:42.481836+00:00 |
776d0cd1690d86b7 | code | CodeAlpaca | medium | [
{
"role": "system",
"content": "You are an expert software engineer. Write complete, correct, runnable code.\n\nFormat your response EXACTLY as:\n<think>\n[Analyze the problem, consider edge cases, choose your approach,\nexplain your algorithm and data structure choices.]\n</think>\n\n```python\n[Complete, ... | import json
with open('data.json') as f:
data = json.load(f)
max_population = 0
max_city_name = ""
for city in data['cities']:
if city['population'] > max_population:
max_population = city['population']
max_city_name = city['name']
print("The city with the highest population is", max_city_nam... | 9 | qwen3.8-max-preview | 569 | 2026-07-21T15:53:26.878139+00:00 |
49b66c515c2f277e | reasoning | SciQ | medium | [
{
"role": "system",
"content": "You are an expert logical reasoner. Think through the problem carefully.\n\nFormat your response EXACTLY as:\n<think>\n[Break down the problem into components.\nConsider multiple hypotheses if applicable.\nEliminate incorrect paths explicitly.\nBuild toward the conclusion ste... | D | 9.5 | qwen3.8-max-preview | 433 | 2026-07-21T23:09:45.278241+00:00 |
645e65747ae3e22a | code | Evol-Code | hard | [
{
"role": "system",
"content": "You are an expert software engineer. Write complete, correct, runnable code.\n\nFormat your response EXACTLY as:\n<think>\n[Analyze the problem, consider edge cases, choose your approach,\nexplain your algorithm and data structure choices.]\n</think>\n\n```python\n[Complete, ... | Assuming that the backup table already exists and has the same structure as mentioned in the question, you can use the following SQL query to drop multiple columns from the "Customers" table and update the "Customers_Backup" table with the dropped columns:
```sql
-- Step 1: Create a temporary table to store the droppe... | 8.5 | qwen3.8-max-preview | 1,427 | 2026-07-22T00:53:41.246372+00:00 |
e2dd4cda6956145b | instruction | tulu-3 | medium | [
{
"role": "system",
"content": "You are a helpful, precise assistant. Follow instructions exactly.\n\nFormat your response as:\n<think>\n[Understand what is being asked. Identify constraints and requirements.\nPlan your response structure.]\n</think>\n\n[Your complete, well-structured response following all... | 7.5 | qwen3.8-max-preview | 421 | 2026-07-21T15:51:32.768929+00:00 | |
aa8206faa7f07139 | math | MetaMathQA | hard | [
{
"role": "system",
"content": "You are an expert mathematician. Solve the given problem with complete rigor.\n\nFormat your response EXACTLY as:\n<think>\n[Your complete step-by-step reasoning. Show all intermediate calculations.\nIf you make an error and catch it, show the correction.\nEnd with a clear st... | 3 | 10 | qwen3.8-max-preview | 226 | 2026-07-22T08:36:35.423115+00:00 |
6290dc16d7d682c4 | reasoning | SciQ | medium | [
{
"role": "system",
"content": "You are an expert logical reasoner. Think through the problem carefully.\n\nFormat your response EXACTLY as:\n<think>\n[Break down the problem into components.\nConsider multiple hypotheses if applicable.\nEliminate incorrect paths explicitly.\nBuild toward the conclusion ste... | A | 9.5 | qwen3.8-max-preview | 350 | 2026-07-21T15:36:35.069570+00:00 |
504cb7bd0ff260ea | reasoning | QASC | hard | [
{
"role": "system",
"content": "You are an expert logical reasoner. Think through the problem carefully.\n\nFormat your response EXACTLY as:\n<think>\n[Break down the problem into components.\nConsider multiple hypotheses if applicable.\nEliminate incorrect paths explicitly.\nBuild toward the conclusion ste... | F | 9.5 | qwen3.8-max-preview | 865 | 2026-07-22T00:20:24.345659+00:00 |
c9ce9cbbdabd705f | reasoning | SciQ | medium | [
{
"role": "system",
"content": "You are an expert logical reasoner. Think through the problem carefully.\n\nFormat your response EXACTLY as:\n<think>\n[Break down the problem into components.\nConsider multiple hypotheses if applicable.\nEliminate incorrect paths explicitly.\nBuild toward the conclusion ste... | A | 9.5 | qwen3.8-max-preview | 509 | 2026-07-22T10:14:05.584082+00:00 |
bc8e94fc6cfe0a33 | reasoning | QASC | hard | [
{
"role": "system",
"content": "You are an expert logical reasoner. Think through the problem carefully.\n\nFormat your response EXACTLY as:\n<think>\n[Break down the problem into components.\nConsider multiple hypotheses if applicable.\nEliminate incorrect paths explicitly.\nBuild toward the conclusion ste... | D | 9.5 | qwen3.8-max-preview | 632 | 2026-07-21T18:53:17.626372+00:00 |
f90b894e9d9049b7 | code | CodeAlpaca | medium | [
{
"role": "system",
"content": "You are an expert software engineer. Write complete, correct, runnable code.\n\nFormat your response EXACTLY as:\n<think>\n[Analyze the problem, consider edge cases, choose your approach,\nexplain your algorithm and data structure choices.]\n</think>\n\n```python\n[Complete, ... | def convert_anagram(word):
result = {}
for c in word:
result.setdefault(c, 0)
result[c] += 1
return result | 9 | qwen3.8-max-preview | 474 | 2026-07-22T09:55:54.090556+00:00 |
cd42fdd58890d0f9 | code | Evol-Code | hard | [
{
"role": "system",
"content": "You are an expert software engineer. Write complete, correct, runnable code.\n\nFormat your response EXACTLY as:\n<think>\n[Analyze the problem, consider edge cases, choose your approach,\nexplain your algorithm and data structure choices.]\n</think>\n\n```python\n[Complete, ... | To generate a random prime number between 1 and 1000 that has at least three digits and is a palindrome in Python, you can use the following method:
```python
import random
def generate_random_palindrome_prime():
while True:
# Generate a random three-digit number
num = random.randint(100, 999)
... | 7.5 | qwen3.8-max-preview | 8,206 | 2026-07-21T10:14:58.121807+00:00 |
b2be5ded14f9edd5 | math | OrcaMath | easy | [
{
"role": "system",
"content": "You are an expert mathematician. Solve the given problem with complete rigor.\n\nFormat your response EXACTLY as:\n<think>\n[Your complete step-by-step reasoning. Show all intermediate calculations.\nIf you make an error and catch it, show the correction.\nEnd with a clear st... | To find the average number of trees planted by the three classes, we first need to find out how many trees each class planted.
Class (A) planted 35 trees.
Class (B) planted 6 more trees than class (A), so class (B) planted 35 + 6 = 41 trees.
Class (C) planted 3 less trees than class (A), so class (C) planted 35 - 3 = ... | 9 | qwen3.8-max-preview | 291 | 2026-07-22T07:40:31.782176+00:00 |
e35ae80c80455904 | math | MetaMathQA | medium | [
{
"role": "system",
"content": "You are an expert mathematician. Solve the given problem with complete rigor.\n\nFormat your response EXACTLY as:\n<think>\n[Your complete step-by-step reasoning. Show all intermediate calculations.\nIf you make an error and catch it, show the correction.\nEnd with a clear st... | 88 | 10 | qwen3.8-max-preview | 256 | 2026-07-21T16:04:39.433840+00:00 |
115370ff7dd376ed | reasoning | ARC-Challenge | medium | [
{
"role": "system",
"content": "You are an expert logical reasoner. Think through the problem carefully.\n\nFormat your response EXACTLY as:\n<think>\n[Break down the problem into components.\nConsider multiple hypotheses if applicable.\nEliminate incorrect paths explicitly.\nBuild toward the conclusion ste... | B | 9.5 | qwen3.8-max-preview | 356 | 2026-07-22T05:25:19.653801+00:00 |
6c68935295449e74 | reasoning | SciQ | medium | [
{
"role": "system",
"content": "You are an expert logical reasoner. Think through the problem carefully.\n\nFormat your response EXACTLY as:\n<think>\n[Break down the problem into components.\nConsider multiple hypotheses if applicable.\nEliminate incorrect paths explicitly.\nBuild toward the conclusion ste... | D | 9.5 | qwen3.8-max-preview | 467 | 2026-07-21T13:01:24.219076+00:00 |
e1d539bde139deaf | reasoning | CommonsenseQA | medium | [
{
"role": "system",
"content": "You are an expert logical reasoner. Think through the problem carefully.\n\nFormat your response EXACTLY as:\n<think>\n[Break down the problem into components.\nConsider multiple hypotheses if applicable.\nEliminate incorrect paths explicitly.\nBuild toward the conclusion ste... | E | 9.5 | qwen3.8-max-preview | 559 | 2026-07-21T22:55:06.070741+00:00 |
b1edcc2ef665a598 | reasoning | QASC | hard | [
{
"role": "system",
"content": "You are an expert logical reasoner. Think through the problem carefully.\n\nFormat your response EXACTLY as:\n<think>\n[Break down the problem into components.\nConsider multiple hypotheses if applicable.\nEliminate incorrect paths explicitly.\nBuild toward the conclusion ste... | A | 9.5 | qwen3.8-max-preview | 667 | 2026-07-22T02:07:16.565933+00:00 |
cbff9e1f73c2402e | math | NuminaMath-CoT | hard | [
{
"role": "system",
"content": "You are an expert mathematician. Solve the given problem with complete rigor.\n\nFormat your response EXACTLY as:\n<think>\n[Your complete step-by-step reasoning. Show all intermediate calculations.\nIf you make an error and catch it, show the correction.\nEnd with a clear st... | 9 | qwen3.8-max-preview | 564 | 2026-07-22T05:57:43.175791+00:00 | |
7ec8910a2c8cb84c | reasoning | QASC | hard | [
{
"role": "system",
"content": "You are an expert logical reasoner. Think through the problem carefully.\n\nFormat your response EXACTLY as:\n<think>\n[Break down the problem into components.\nConsider multiple hypotheses if applicable.\nEliminate incorrect paths explicitly.\nBuild toward the conclusion ste... | G | 9.5 | qwen3.8-max-preview | 926 | 2026-07-21T16:43:11.534349+00:00 |
9078d49a6adbed1b | code | CodeAlpaca | medium | [
{
"role": "system",
"content": "You are an expert software engineer. Write complete, correct, runnable code.\n\nFormat your response EXACTLY as:\n<think>\n[Analyze the problem, consider edge cases, choose your approach,\nexplain your algorithm and data structure choices.]\n</think>\n\n```python\n[Complete, ... | The Set data structure can be used to remove duplicate characters from a string in JavaScript.
For example:
[...new Set("string")].join(""); // returns "strng" | 7 | qwen3.8-max-preview | 761 | 2026-07-21T20:42:15.423096+00:00 |
f9dec7b1140ad76c | code | Evol-Code | medium | [
{
"role": "system",
"content": "You are an expert software engineer. Write complete, correct, runnable code.\n\nFormat your response EXACTLY as:\n<think>\n[Analyze the problem, consider edge cases, choose your approach,\nexplain your algorithm and data structure choices.]\n</think>\n\n```python\n[Complete, ... | Here is the code to accomplish the task:
```python
number = 13
count = 0
for i in range(1, 101):
if i % 3 == 0 and i % 2 == 0:
digit_sum = sum(int(digit) for digit in str(i))
if digit_sum > 10:
print(number, "*", i//3, "=", number * (i//3))
count += 1
if count == 10:
... | 5.5 | qwen3.8-max-preview | 11,396 | 2026-07-21T11:18:45.948898+00:00 |
9980ec1d2a1ca305 | math | MetaMathQA | medium | [
{
"role": "system",
"content": "You are an expert mathematician. Solve the given problem with complete rigor.\n\nFormat your response EXACTLY as:\n<think>\n[Your complete step-by-step reasoning. Show all intermediate calculations.\nIf you make an error and catch it, show the correction.\nEnd with a clear st... | 6 | 10 | qwen3.8-max-preview | 288 | 2026-07-22T06:50:05.942135+00:00 |
612f0fbc242a7470 | code | CodeAlpaca | medium | [
{
"role": "system",
"content": "You are an expert software engineer. Write complete, correct, runnable code.\n\nFormat your response EXACTLY as:\n<think>\n[Analyze the problem, consider edge cases, choose your approach,\nexplain your algorithm and data structure choices.]\n</think>\n\n```python\n[Complete, ... | SELECT *
FROM table
ORDER BY column_name DESC
LIMIT 3; | 7 | qwen3.8-max-preview | 609 | 2026-07-21T14:11:18.138194+00:00 |
40f4be250980aaf2 | instruction | tulu-3 | medium | [
{
"role": "system",
"content": "You are a helpful, precise assistant. Follow instructions exactly.\n\nFormat your response as:\n<think>\n[Understand what is being asked. Identify constraints and requirements.\nPlan your response structure.]\n</think>\n\n[Your complete, well-structured response following all... | 7.5 | qwen3.8-max-preview | 377 | 2026-07-21T21:21:22.840573+00:00 | |
67526dda2ac936fb | code | Evol-Code | hard | [
{
"role": "system",
"content": "You are an expert software engineer. Write complete, correct, runnable code.\n\nFormat your response EXACTLY as:\n<think>\n[Analyze the problem, consider edge cases, choose your approach,\nexplain your algorithm and data structure choices.]\n</think>\n\n```python\n[Complete, ... | num = 15
total = 0
def is_prime(num):
if num < 2:
return False
for i in range(2, int(num**0.5) + 1):
if num % i == 0:
return False
return True
for i in range(1, num+1):
if i % 2 == 0 and i % 3 == 0:
continue
elif is_prime(i):
total += i
print(total) | 9 | qwen3.8-max-preview | 725 | 2026-07-22T03:01:20.238061+00:00 |
996880f1259851e6 | reasoning | CommonsenseQA | medium | [
{
"role": "system",
"content": "You are an expert logical reasoner. Think through the problem carefully.\n\nFormat your response EXACTLY as:\n<think>\n[Break down the problem into components.\nConsider multiple hypotheses if applicable.\nEliminate incorrect paths explicitly.\nBuild toward the conclusion ste... | D | 9.5 | qwen3.8-max-preview | 556 | 2026-07-22T01:45:05.137781+00:00 |
40774d67735be21d | reasoning | SciQ | medium | [
{
"role": "system",
"content": "You are an expert logical reasoner. Think through the problem carefully.\n\nFormat your response EXACTLY as:\n<think>\n[Break down the problem into components.\nConsider multiple hypotheses if applicable.\nEliminate incorrect paths explicitly.\nBuild toward the conclusion ste... | D | 9.5 | qwen3.8-max-preview | 486 | 2026-07-22T03:48:07.448141+00:00 |
2e7f658452841dce | reasoning | CommonsenseQA | medium | [
{
"role": "system",
"content": "You are an expert logical reasoner. Think through the problem carefully.\n\nFormat your response EXACTLY as:\n<think>\n[Break down the problem into components.\nConsider multiple hypotheses if applicable.\nEliminate incorrect paths explicitly.\nBuild toward the conclusion ste... | D | 9.5 | qwen3.8-max-preview | 681 | 2026-07-21T13:49:37.835664+00:00 |
68c036bcf0bf09af | reasoning | SciQ | medium | [
{
"role": "system",
"content": "You are an expert logical reasoner. Think through the problem carefully.\n\nFormat your response EXACTLY as:\n<think>\n[Break down the problem into components.\nConsider multiple hypotheses if applicable.\nEliminate incorrect paths explicitly.\nBuild toward the conclusion ste... | C | 9.5 | qwen3.8-max-preview | 481 | 2026-07-22T09:06:28.308726+00:00 |
466328e16ac12448 | instruction | Dolly | medium | [
{
"role": "system",
"content": "You are a helpful, precise assistant. Follow instructions exactly.\n\nFormat your response as:\n<think>\n[Understand what is being asked. Identify constraints and requirements.\nPlan your response structure.]\n</think>\n\n[Your complete, well-structured response following all... | 1. Explain that there are other ways to communicate.
2. Ignore your child's whining, step away if you can't.
3. Enforce discipline. | 6 | qwen3.8-max-preview | 573 | 2026-07-22T01:00:00.370039+00:00 |
d52ba11b40b14568 | math | NuminaMath-CoT | hard | [
{
"role": "system",
"content": "You are an expert mathematician. Solve the given problem with complete rigor.\n\nFormat your response EXACTLY as:\n<think>\n[Your complete step-by-step reasoning. Show all intermediate calculations.\nIf you make an error and catch it, show the correction.\nEnd with a clear st... | 9 | qwen3.8-max-preview | 690 | 2026-07-22T02:50:24.734639+00:00 | |
22c437a8fdf98d84 | code | Evol-Code | hard | [
{
"role": "system",
"content": "You are an expert software engineer. Write complete, correct, runnable code.\n\nFormat your response EXACTLY as:\n<think>\n[Analyze the problem, consider edge cases, choose your approach,\nexplain your algorithm and data structure choices.]\n</think>\n\n```python\n[Complete, ... | # Recursive implementation
def calculate_sum(index, current_sum):
term = 1 / (index ** 2)
new_sum = current_sum + term
if term < 1e-6: # Threshold condition
return new_sum, index
else:
return calculate_sum(index + 1, new_sum)
def main():
threshold = input("Enter the threshold... | 8 | qwen3.8-max-preview | 2,997 | 2026-07-21T12:39:39.344939+00:00 |
f83bc7bcfd42e06a | math | MetaMathQA | easy | [
{
"role": "system",
"content": "You are an expert mathematician. Solve the given problem with complete rigor.\n\nFormat your response EXACTLY as:\n<think>\n[Your complete step-by-step reasoning. Show all intermediate calculations.\nIf you make an error and catch it, show the correction.\nEnd with a clear st... | 100 | 9 | qwen3.8-max-preview | 762 | 2026-07-21T20:16:41.302747+00:00 |
1c3df660b0019cbc | instruction | tulu-3 | medium | [
{
"role": "system",
"content": "You are a helpful, precise assistant. Follow instructions exactly.\n\nFormat your response as:\n<think>\n[Understand what is being asked. Identify constraints and requirements.\nPlan your response structure.]\n</think>\n\n[Your complete, well-structured response following all... | 6 | qwen3.8-max-preview | 313 | 2026-07-22T04:11:56.283705+00:00 | |
5366e7c6cb3fac23 | math | NuminaMath-CoT | hard | [
{
"role": "system",
"content": "You are an expert mathematician. Solve the given problem with complete rigor.\n\nFormat your response EXACTLY as:\n<think>\n[Your complete step-by-step reasoning. Show all intermediate calculations.\nIf you make an error and catch it, show the correction.\nEnd with a clear st... | 9 | qwen3.8-max-preview | 415 | 2026-07-22T08:10:46.175706+00:00 | |
af48da2e49571489 | reasoning | QASC | hard | [
{
"role": "system",
"content": "You are an expert logical reasoner. Think through the problem carefully.\n\nFormat your response EXACTLY as:\n<think>\n[Break down the problem into components.\nConsider multiple hypotheses if applicable.\nEliminate incorrect paths explicitly.\nBuild toward the conclusion ste... | D | 9.5 | qwen3.8-max-preview | 866 | 2026-07-21T21:24:30.290184+00:00 |
4c3261e17e2d7ae6 | math | NuminaMath-CoT | hard | [
{
"role": "system",
"content": "You are an expert mathematician. Solve the given problem with complete rigor.\n\nFormat your response EXACTLY as:\n<think>\n[Your complete step-by-step reasoning. Show all intermediate calculations.\nIf you make an error and catch it, show the correction.\nEnd with a clear st... | 9 | qwen3.8-max-preview | 331 | 2026-07-22T09:40:01.761442+00:00 | |
96eba07a58829068 | code | CodeAlpaca | medium | [
{
"role": "system",
"content": "You are an expert software engineer. Write complete, correct, runnable code.\n\nFormat your response EXACTLY as:\n<think>\n[Analyze the problem, consider edge cases, choose your approach,\nexplain your algorithm and data structure choices.]\n</think>\n\n```python\n[Complete, ... | The number of elements in a list can be obtained using the len() method in Python.
For example, if the list is named my_list, the number of elements can be obtained as len(my_list). | 7.5 | qwen3.8-max-preview | 398 | 2026-07-22T05:05:35.567321+00:00 |
0fb3664f27dd0d18 | instruction | tulu-3 | medium | [
{
"role": "system",
"content": "You are a helpful, precise assistant. Follow instructions exactly.\n\nFormat your response as:\n<think>\n[Understand what is being asked. Identify constraints and requirements.\nPlan your response structure.]\n</think>\n\n[Your complete, well-structured response following all... | 6 | qwen3.8-max-preview | 165 | 2026-07-22T00:20:25.292174+00:00 | |
a1d51777fec30f81 | math | MetaMathQA | medium | [
{
"role": "system",
"content": "You are an expert mathematician. Solve the given problem with complete rigor.\n\nFormat your response EXACTLY as:\n<think>\n[Your complete step-by-step reasoning. Show all intermediate calculations.\nIf you make an error and catch it, show the correction.\nEnd with a clear st... | 3 | 10 | qwen3.8-max-preview | 311 | 2026-07-21T15:33:59.684550+00:00 |
974919879b49a8ca | code | CodeAlpaca | medium | [
{
"role": "system",
"content": "You are an expert software engineer. Write complete, correct, runnable code.\n\nFormat your response EXACTLY as:\n<think>\n[Analyze the problem, consider edge cases, choose your approach,\nexplain your algorithm and data structure choices.]\n</think>\n\n```python\n[Complete, ... | def reverse_string(string)
return string.reverse
end | 7 | qwen3.8-max-preview | 566 | 2026-07-21T15:12:57.171225+00:00 |
Qwen3.8-Max Distillation 50K
A curated dataset of 49,772 teacher-generated traces from qwen3.8-max-preview, prepared for supervised fine-tuning and off-policy knowledge distillation.
The teacher responses are preserved as returned by the API. Where the model emitted visible <think>...</think> blocks, those blocks remain in the assistant message. Some simpler prompts received direct answers without a thinking block.
Terms and provenance notice — not cleared for unrestricted commercial/model-training use. This dataset was generated through Alibaba Cloud Model Studio. The current Alibaba Cloud International Product Terms, §4.48(d)(v), restrict using Model Studio, its hosted models, or their outputs to train or develop products or services that compete with Alibaba Cloud or its affiliates unless expressly authorized. In addition, at least 12,106 rows (24.3%) originate from explicitly noncommercial sources: Evol-Instruct-Code (CC BY-NC-SA 4.0) and SciQ (CC BY-NC 3.0). Other sources have heterogeneous attribution and share-alike obligations. Public availability of this repository is not a grant of rights. Before training, redistribution, or commercial use, review the agreement applicable to your account/region/plan and every upstream source license. See
PROVENANCE.mdand the official Model Studio terms.
Dataset summary
| Split | Rows |
|---|---|
| Train | 44,796 |
| Validation | 2,488 |
| Test | 2,488 |
| Total | 49,772 |
Domain distribution
| Domain | Rows | Share |
|---|---|---|
| Math | 16,597 | 33.3% |
| Code | 14,057 | 28.2% |
| Reasoning | 11,690 | 23.5% |
| Instruction following | 7,302 | 14.7% |
| Tool use | 126 | 0.3% |
Source distribution
The prompt mixture includes:
- Math: MetaMathQA, NuminaMath-CoT, OrcaMath, GSM8K, and all seven MATH subject configurations
- Code: Evol-Instruct-Code, CodeAlpaca, MBPP, and HumanEval
- Reasoning: SciQ, CommonsenseQA, QASC, ARC-Challenge, ARC-Easy, and OpenBookQA
- Instruction following: Tulu 3, Dolly, and IFEval
- Tool use: synthetic multi-tool scenarios
The largest individual sources are Evol-Code (8,138), MetaMathQA (6,406), CodeAlpaca (5,604), Tulu 3 (5,397), and NuminaMath-CoT (4,247).
Generation
- Teacher:
qwen3.8-max-preview - API: Alibaba Cloud OpenAI-compatible endpoint
- Temperature: 0.7
- Top-p: 0.95
- Maximum completion: 8,192 tokens
- Prompting: domain-specific system prompts for structured reasoning, final answers, code, and tool calls
- Reasoning budget: a provider-supported thinking budget of 2,048 tokens was used for most of the generation run; early traces were generated before that cap was introduced
- Raw output policy: teacher outputs were not rewritten or normalized before curation
Curation
Raw generation produced 53,693 records. The curation pipeline:
- Required system/user/assistant message structure.
- Removed incomplete generations (
finish_reason != stop). - Removed empty, very short, and low-substance responses.
- Applied domain-specific checks, including boxed-answer checks for math and code-presence checks for code tasks.
- Removed tool-use refusals that did not attempt the task.
- Assigned a heuristic quality score.
- Removed exact response duplicates and duplicate prompt prefixes.
- Randomly created 90%/5%/5% train, validation, and test splits.
Curated quality indicators
| Indicator | Value |
|---|---|
| Mean heuristic quality score | 8.59 / 10 |
| Score range | 4.5–10.0 |
Responses containing <think> |
82.6% |
Math responses containing \\boxed{...} |
99.4% |
| Code responses containing fenced code | 99.4% |
| Average completion length | 957 tokens |
The quality score is a heuristic, not a factual-correctness judgment.
Record schema
Each row contains:
{
"id": "stable 16-character trace identifier",
"domain": "math | code | reasoning | instruction | tool_use",
"source": "source dataset name",
"difficulty": "easy | medium | hard | unknown",
"messages": [
{"role": "system", "content": "..."},
{"role": "user", "content": "..."},
{"role": "assistant", "content": "..."},
],
"ground_truth": "reference answer when available",
"quality_score": 8.5,
"teacher_model": "qwen3.8-max-preview",
"completion_tokens": 957,
"generation_time": "ISO-8601 timestamp",
}
ground_truth is empty for open-ended sources and for some source datasets that did not expose a directly extractable answer. Approximately 19.6% of curated records have no ground-truth string.
Usage
from datasets import load_dataset
ds = load_dataset("r0b0tlab/qwen3.8-max-distillation-50k")
print(ds)
print(ds["train"][0]["messages"])
For response-only SFT, mask the system and user tokens and compute loss only over the assistant response.
Intended uses
- Supervised fine-tuning of chat or reasoning models
- Off-policy teacher-to-student distillation
- Ablations on reasoning-format and direct-answer mixtures
- Research on math, code, scientific reasoning, and instruction following
Important limitations
- Generated text can be wrong. The responses were not independently verified problem-by-problem.
<think>content is model-emitted visible reasoning prompted by the collection pipeline. It should not be interpreted as privileged access to hidden model state.- The dataset includes prompts originating from standard benchmark splits, including GSM8K, MATH, HumanEval, MBPP, ARC, IFEval, and related sets. A model trained on this dataset must not be fairly evaluated on overlapping benchmark items without contamination controls.
- The source mixture is predominantly English but includes some multilingual instruction data.
- Tool-use data is only 126 rows after strict curation and is not sufficient by itself for robust function-calling training.
- Heuristic filtering favors structured responses and may introduce style bias.
- The random train/validation/test split is intended for pipeline development, not uncontaminated capability evaluation.
Licensing and attribution
This is a multi-source derived dataset. The repository does not claim to replace or override the licenses, terms, or attribution requirements of the underlying prompt datasets or the teacher-model provider. Users are responsible for reviewing and complying with the terms of every upstream source relevant to their use case.
Because the source mixture has heterogeneous upstream terms, the Hub metadata uses license: other rather than applying one blanket license to all records.
Citation
@misc{r0b0tlab2026qwen38distillation50k,
title = {Qwen3.8-Max Distillation 50K},
author = {r0b0tlab},
year = {2026},
publisher = {Hugging Face},
howpublished = {\url{https://huggingface.co/datasets/r0b0tlab/qwen3.8-max-distillation-50k}}
}
- Downloads last month
- 663