Dataset Viewer
Auto-converted to Parquet Duplicate
uuid
string
question
string
response_reasoning
string
response_content
string
correctness
bool
_uuid_index
int64
trajectory_source
string
_trajectory_matched
bool
raw_messages
list
qwen_text
string
num_qwen_tokens
int64
classified-0
The graph of $y = \frac{p(x)}{q(x)}$ is shown below, where $p(x)$ and $q(x)$ are quadratic. (Assume that the grid lines are at integers.) [asy] unitsize(0.6 cm); real func (real x) { return (2*(x - 1)/(x + 2)); } int i; for (i = -8; i <= 8; ++i) { draw((i,-8)--(i,8),gray(0.7)); draw((-8,i)--(8,i),gray(0.7));...
Okay, so I need to find p(3)/q(3) where p(x) and q(x) are quadratic polynomials, and the graph of y = p(x)/q(x) is given. The horizontal asymptote is y = 2, and the only vertical asymptote is x = -2. Let me start by recalling some things about rational functions. First, vertical asymptotes occur where the denominator ...
To solve the problem, we are given that the graph of the rational function $$ y = \frac{p(x)}{q(x)} $$ is shown, where $ p(x) $ and $ q(x) $ are both **quadratic** polynomials. The key features of the graph are: - A **horizontal asymptote** at $ y = 2 $. - A **vertical asymptote** at $ x = -2 $. - A **hole** at $ ...
false
0
E:\AReaL-zzy-add-trial\pipeline\work\train_classified.jsonl
false
[ { "content": "The graph of $y = \\frac{p(x)}{q(x)}$ is shown below, where $p(x)$ and $q(x)$ are quadratic. (Assume that the grid lines are at integers.)\n\n[asy]\nunitsize(0.6 cm);\n\nreal func (real x) {\n return (2*(x - 1)/(x + 2));\n}\n\nint i;\n\nfor (i = -8; i <= 8; ++i) {\n draw((i,-8)--(i,8),gray(0.7)...
<|im_start|>user The graph of $y = \frac{p(x)}{q(x)}$ is shown below, where $p(x)$ and $q(x)$ are quadratic. (Assume that the grid lines are at integers.) [asy] unitsize(0.6 cm); real func (real x) { return (2*(x - 1)/(x + 2)); } int i; for (i = -8; i <= 8; ++i) { draw((i,-8)--(i,8),gray(0.7)); draw((-8,i)--...
5,607
classified-1
157. Secret Code. Two conspirators had a secret code. Sometimes their correspondence contained simple arithmetic operations that looked completely innocent. However, in the code, each of the ten digits represented its own letter of the alphabet. Once, a sum was encountered which, after substituting the corresponding le...
Okay, so I need to solve this cryptarithmetic puzzle. Let me start by writing down the problem again to make sure I have it right. The sum is: ``` F L Y F O R Y O U R ---------- L I F E ``` And we know that I is 1 and O is 0. So, I can substitute those values in. Let me note that down: I = 1, O = 0. So, the let...
To solve the cryptarithmetic puzzle: $$ \begin{gathered} F\ L\ Y \\ F\ O\ R \\ Y\ O\ U\ R \\ \hline L\ I\ F\ E \\ \end{gathered} $$ we are told that: - $ I = 1 $ - $ O = 0 $ Each letter represents a unique digit from 0 to 9, and we must deduce the digit values for the other letters $ F, L, Y, R, U, E $ such that th...
false
1
E:\AReaL-zzy-add-trial\pipeline\work\train_classified.jsonl
false
[ { "content": "157. Secret Code. Two conspirators had a secret code. Sometimes their correspondence contained simple arithmetic operations that looked completely innocent. However, in the code, each of the ten digits represented its own letter of the alphabet. Once, a sum was encountered which, after substitutin...
<|im_start|>user 157. Secret Code. Two conspirators had a secret code. Sometimes their correspondence contained simple arithmetic operations that looked completely innocent. However, in the code, each of the ten digits represented its own letter of the alphabet. Once, a sum was encountered which, after substituting the...
5,818
classified-2
"A group with 7 young men and 7 young women was divided into pairs randomly. Find the probability th(...TRUNCATED)
"Okay, so I need to find the probability that when a group of 7 young men and 7 young women is divid(...TRUNCATED)
"To find the probability that **at least one pair consists of two women** when 7 young men and 7 you(...TRUNCATED)
false
2
E:\AReaL-zzy-add-trial\pipeline\work\train_classified.jsonl
false
[{"content":"A group with 7 young men and 7 young women was divided into pairs randomly. Find the pr(...TRUNCATED)
"<|im_start|>user\nA group with 7 young men and 7 young women was divided into pairs randomly. Find (...TRUNCATED)
6,008
classified-3
"Celine has 240 feet of fencing. She needs to enclose a rectangular area such that the area is eight(...TRUNCATED)
"Okay, so Celine has 240 feet of fencing, and she wants to enclose a rectangular area. The problem s(...TRUNCATED)
"To solve the problem, we are given that Celine has **240 feet of fencing** and wants to enclose a *(...TRUNCATED)
false
3
E:\AReaL-zzy-add-trial\pipeline\work\train_classified.jsonl
false
[{"content":"Celine has 240 feet of fencing. She needs to enclose a rectangular area such that the a(...TRUNCATED)
"<|im_start|>user\nCeline has 240 feet of fencing. She needs to enclose a rectangular area such that(...TRUNCATED)
7,433
classified-4
"8. Let the complex numbers be $z_{1}=-3-\\sqrt{3} \\mathrm{i}, z_{2}=\\sqrt{3}+\\mathrm{i}, z=\\sqr(...TRUNCATED)
"Okay, so I need to find the minimum value of |z - z₁| + |z - z₂| where z is given as √3 sinθ(...TRUNCATED)
"To determine the **minimum value of** \n$$\n|z - z_1| + |z - z_2|\n$$ \nwhere \n$$\nz_1 = -3 - \(...TRUNCATED)
false
4
E:\AReaL-zzy-add-trial\pipeline\work\train_classified.jsonl
false
[{"content":"8. Let the complex numbers be $z_{1}=-3-\\sqrt{3} \\mathrm{i}, z_{2}=\\sqrt{3}+\\mathrm(...TRUNCATED)
"<|im_start|>user\n8. Let the complex numbers be $z_{1}=-3-\\sqrt{3} \\mathrm{i}, z_{2}=\\sqrt{3}+\\(...TRUNCATED)
7,687
classified-5
"There are 5 houses, numbered 1 to 5 from left to right, as seen from across the street. Each house (...TRUNCATED)
"Okay, let's try to solve this logic puzzle. There are 5 houses with different people, each having u(...TRUNCATED)
"To determine the **CarModel** for **Peter**, we systematically analyzed the clues and built a consi(...TRUNCATED)
false
5
E:\AReaL-zzy-add-trial\pipeline\work\train_classified.jsonl
false
[{"content":"There are 5 houses, numbered 1 to 5 from left to right, as seen from across the street.(...TRUNCATED)
"<|im_start|>user\nThere are 5 houses, numbered 1 to 5 from left to right, as seen from across the s(...TRUNCATED)
7,221
classified-6
"17. At 6:20:25, a certain electronic watch displays 6:20:25. How many times within the hour from 5:(...TRUNCATED)
"<Parallel>\n<Outlines>\n<Trial>1: Fix the hour as 5 and impose the distinct-digit constraint agains(...TRUNCATED)
"To determine how many times between **5:00 and 6:00** an electronic watch displays a time with **fi(...TRUNCATED)
false
6
E:\AReaL-zzy-add-trial\pipeline\work\train_classified.jsonl
false
[{"content":"17. At 6:20:25, a certain electronic watch displays 6:20:25. How many times within the (...TRUNCATED)
"<|im_start|>user\n17. At 6:20:25, a certain electronic watch displays 6:20:25. How many times withi(...TRUNCATED)
10,529
classified-7
"What is the result of adding 0.45 to 52.7 and then subtracting 0.25? Let's think step by step and o(...TRUNCATED)
"Okay, let me try to figure out this math problem. So the question is, what is the result of adding (...TRUNCATED)
"To solve the problem of adding 0.45 to 52.7 and then subtracting 0.25, we can break it down into tw(...TRUNCATED)
false
7
E:\AReaL-zzy-add-trial\pipeline\work\train_classified.jsonl
false
[{"content":"What is the result of adding 0.45 to 52.7 and then subtracting 0.25? Let's think step b(...TRUNCATED)
"<|im_start|>user\nWhat is the result of adding 0.45 to 52.7 and then subtracting 0.25? Let's think (...TRUNCATED)
1,227
classified-8
"Two circles touch each other internally at point \\( M \\). Let \\( AB \\) be a chord of the larger(...TRUNCATED)
"Okay, so I need to prove that MT is the angle bisector of angle AMB. Let me start by visualizing th(...TRUNCATED)
"To prove that **$ MT $** is the **angle bisector** of **$\\angle AMB$**, we begin by analyzing the (...TRUNCATED)
false
8
E:\AReaL-zzy-add-trial\pipeline\work\train_classified.jsonl
false
[{"content":"Two circles touch each other internally at point \\( M \\). Let \\( AB \\) be a chord o(...TRUNCATED)
"<|im_start|>user\nTwo circles touch each other internally at point \\( M \\). Let \\( AB \\) be a c(...TRUNCATED)
12,575
classified-9
"Let the parabola \\( C: y^2 = 2x \\) have its directrix intersect the \\( x \\)-axis at point \\( A(...TRUNCATED)
"<Parallel>\n<Outlines>\n<Subtask>1: Compare the parabola to the standard right-opening form to dete(...TRUNCATED)
"To find the area of triangle $ \\triangle KMN $, we begin by analyzing the geometric setup involvin(...TRUNCATED)
false
9
E:\AReaL-zzy-add-trial\pipeline\work\train_classified.jsonl
false
[{"content":"Let the parabola \\( C: y^2 = 2x \\) have its directrix intersect the \\( x \\)-axis at(...TRUNCATED)
"<|im_start|>user\nLet the parabola \\( C: y^2 = 2x \\) have its directrix intersect the \\( x \\)-a(...TRUNCATED)
5,693
End of preview. Expand in Data Studio

YAML Metadata Warning:empty or missing yaml metadata in repo card

Check out the documentation for more information.

Parason: Subtask- and Trial-Parallel Reasoning Data

Training data for Parason, which turns long sequential reasoning traces into executable parallel programs. Three subsets, one per training stage.

Subfolder Stage Contents
sft-8x SFT (main) Parallel trajectories with <Subtask> / <Trial> branches, Trial stages expanded 8x — extra branches sampled with Qwen3-8B, their <Trial> goals written by Gemini-3-Flash
sft-1x SFT (baseline) The unexpanded parallel trajectories: 964 annotated Qwen3-8B traces released by ThreadWeaver, each parallel stage labeled Subtask or Trial by Gemini-3-Flash
rl PA-GRPO RL prompt set — Polaris-53k (~53,000 complex reasoning problems), used unmodified

Each subfolder holds a single train.parquet; there is no validation split. The two SFT sets are matched in effective token budget (8 epochs over sft-1x ≈ 1 epoch over sft-8x).

Trajectory format

The SFT trajectories mark two distinct forms of parallelism:

  • <Subtask> — AND-branches. The problem splits into independent sub-goals, each producing a required result. Shortens the critical path.
  • <Trial> — OR-branches. Competing speculative attempts test different hypotheses; all outputs are concatenated back into the reasoning history. Widens search.
<think>
[Initial problem analysis and setup]

<Parallel>
<Outlines>
<Trial> 1: [First hypothesis to test]
<Trial> 2: [Second hypothesis to test]
<Subtask> 3: [An independent sub-goal whose result is required]
</Outlines>

<Thread>
1: [Detailed reasoning for the first branch...]
</Thread>
<Thread>
2: [Detailed reasoning for the second branch...]
</Thread>
<Thread>
3: [Detailed reasoning for the third branch...]
</Thread>
</Parallel>
</think>

[Final answer in \boxed{} format]

Reserved tags, added as special tokens before training:

<Think> </Think>  <Parallel> </Parallel>  <Outlines> </Outlines>  <Outline> </Outline>
<Trial> </Trial>  <Subtask> </Subtask>    <Thread> </Thread>      <Conclusion> </Conclusion>

Schema

sft-8x, sft-1x

Field Type Description
uuid string Source-trace identifier
question string Problem statement
response_reasoning string Trajectory body — the <think> contents, containing the <Parallel> blocks
response_content string Post-</think> final answer, ending in \boxed{...}
correctness bool Whether the source trace was graded correct
raw_messages list[dict] [{"role": "user", ...}, {"role": "assistant", ...}]
qwen_text string The training field. Full trace with the Qwen3 chat template applied
num_qwen_tokens int Token count of qwen_text under the Qwen3-8B tokenizer

rl — standard veRL prompt format: prompt (list[dict]), data_source, ability, reward_model ({"style": "rule", "ground_truth": ...}), extra_info.

Downloads last month
30