Datasets:

IFM
/

Math-Reasoning

text string	token_count int64
Student A: Okay, let’s crack this. The problem says “Let s = -1 - -5. Suppose 0 = v – s·v + 30. What are the prime factors of v?” Easy peasy. I’ll just plug in the numbers. Student B: Hold on. Before we dive in, let’s make sure we interpret everything correctly. That “- -5” could trip us up. What does s equal?...	900
Looking at this problem, I first notice… The three given numbers are all negative and seem to be getting more negative: \[ -36203,\qquad -73449,\qquad -110695. \] Because there are only three terms, there are many ways they could be generated. My first instinct is to check the simplest possibilities: an arith...	758
Student A: Okay, the problem says “Put –6, –47, 0 in increasing order.” I’m feeling pretty confident about this—let’s just list them from the smallest to the biggest. I think it’s –6, –47, 0. Student B: Hmm, I’m not sure about that. Remember, “increasing order” means we start with the smallest number and...	492
Interviewer: Let’s start at the beginning. How did you first interpret the problem? Interviewee: The problem asks for the probability of drawing exactly one u, one v, one m, and one h when we pick four letters without replacement from a multiset. The multiset is given by the counts of each letter: - ...	1,573
Interviewer: Thanks for joining us today. Let’s dive right into the problem. When you first saw it, what was your initial impression, and how did you decide to start tackling it? Interviewee: My first impression was that we’re dealing with a chain of algebraic steps: a polynomial evaluation, a small linear sys...	2,073
Teacher: Let’s look at the problem together. It says, “Let \(a(g)=3g+8\). Is \(a(8)\) a multiple of 11?” What does the notation \(a(g)\) tell you? Student: It’s a function named \(a\) that takes an input \(g\) and does something with it. I think it means we add 3 and \(g\) and then add 8. Teacher: Hmm, th...	1,167
Teacher: Let’s look at the problem: In base 13, what is \(9+(-219)\)? First, what does the notation “\(-219\) in base 13” tell us? Student: It means we have the base‑13 number 219, but with a minus sign in front of it. Teacher: Correct. So we’re adding a positive 9 to a negative number. Before we star...	1,256
Teacher: Today we have a little ordering puzzle. Here are four numbers: - (a) –1.5 - (b) –0.31 - (c) 10 - (d) –1⁄6 The question asks: Which is the fourth biggest value? Before we dive in, can you tell me in your own words what “fourth biggest” might mean when there are exactly four numbers? **Student...	973
Interviewer: Let’s start by hearing the problem in your own words. What are we being asked to solve? Subject Matter Expert: Sure. The problem gives us an equation, “‑10 – 270 = ‑10 × w,” and asks us to first determine the value of w. After we have w, we need to compute the greatest common factor (GCF) of w and...	1,353
Looking at this problem, I first notice the list of numbers I have to compare: \[ -168674,\; -1,\; 10,\; -5,\; -\tfrac23 . \] The question is “Which of these is closest to 1?”. My instinct tells me I need to measure how far each number is from 1. In other words, I should compute the absolute difference \(\|x-1\|\) fo...	1,061
Looking at this problem, I first notice that it’s a chain of simple algebraic definitions that I need to untangle one by one: - \(p = 9 + -7\) - \(z(g) = -4\,g^{2} + 5\,g^{p} + 0\,g^{2}\) - \(n(i) = -5\,i^{2}\) …and finally I’m asked to give \(n(z(s))\). So the task is basically “plug‑in‑the‑plug‑in” – but I should...	1,100
Interview Transcript – Solving a Probability Puzzle --- Interviewer: *Let’s start at the very beginning. When you first saw the problem—“What is the probability of picking two ‘v’s when two letters are drawn without replacement from a bag containing 4 ‘v’s and 10 ‘k’s?”—what was your initial reaction? How did...	1,459
Debater 1 (Place‑Value Advocate): I argue that the cleanest way to round –2769.8 to the nearest ten‑thousand is to apply the classic digit‑to‑the‑right rule. 1. Identify the target place: ten‑thousands = 10 000. In the number –2769.8 the ten‑thousands digit is 0 (there is no digit beyond the thousands pla...	905
Interviewer: How did you first approach this problem? Subject Matter Expert: I started by writing out the given terms with their positions, just to see the pattern more clearly: - Term 1: ‑1233 - Term 2: ‑1232 - Term 3: ‑1231 - Term 4: ‑1230 Seeing that each successive term is exactly +1 larger tha...	1,171
Looking at this problem, I first notice the numbers are huge and the numerator is negative while the denominator is positive, so I know the answer will be negative. I can ignore the sign for the moment and just work with \[ \frac{148\,864\,640}{4\,652\,020}. \] --- Hmm, if I try a quick estimate – 4.6 mil...	858
Interviewer: How did you first approach the problem of factoring \[ 3z^{3}-546z^{2}+26724z-395352\;? \] Subject‑Matter Expert: My first instinct was to look for any overall common factor among the coefficients. All four numbers—3, ‑546, 26724, and ‑395352—are divisible by 3, so I factored a 3 out right away...	2,107
Teacher: Today we have a small challenge. “Find the greatest common divisor of 1943 and 2668.” Do you remember what a greatest common divisor, or GCD, actually is? Student: It’s the biggest number that divides both numbers without leaving a remainder. Teacher: Exactly. And how do we usually find that nu...	930
Debater 1: I argue that the most reliable way to find the s‑th term of the sequence \[ -300\,229,\; -600\,456,\; -900\,683,\dots \] is to treat it as an arithmetic progression. The hallmark of an arithmetic sequence is a constant difference between consecutive terms. Computing the first difference we ge...	1,154
Debater 1: I argue that the best approach to find the \(n\)th term of the sequence \[ 1805,\;7093,\;15925,\;28301 \] is the finite‑difference method. By computing successive differences we can quickly detect the polynomial degree of the sequence without solving any systems of equations. 1. **First dif...	1,122
Teacher: Let's look at this equation together: \(-b + 6 = 31b - 26\). What are we trying to find? Student: We need the value of \(b\) that makes the equation true. Teacher: Exactly. When you have a variable on both sides, what’s a useful first step? Student: I guess we should try to get all the \(b\)...	878
Student A: Alright, let’s tackle 8 274 205 ÷ 31. I’m pretty sure we just do long division straight away. Student B: Yep, long division works. Start with the leftmost digits that are at least as big as 31. Student A: The first digit is 8, which is smaller than 31, so we take the first two digits: 82. **St...	915
Student A: Hey, we’ve got this JavaScript function we need to translate to Python: ```js function isValidNumber(num) { return !isNaN(num) && num > 0; } ``` But the twist is that the Python version should also accept arrays (i.e., lists) and handle extremely large numbers or strings that represent huge num...	1,475
Student A: Okay, let’s tackle this one. We have \[ -300h^3 + 1444h^3 + 64h^3 + 300h^3. \] All the terms have the same \(h^3\), so it’s just a matter of adding the numbers in front, right? Easy peasy. Student B: Yeah, they’re like terms. But remember to keep an eye on the signs—especially that first \(-300\...	761
Teacher: Let’s look at the problem: What is the remainder when 364 is divided by 9? How would you start tackling it? Student: I guess I could just do the division: 364 ÷ 9. Teacher: That’s a solid approach. Before you crunch the numbers, can you think of any shortcut that might help, especially with a...	932
Student A: Okay, let’s tackle this one. The expression is \(0 + 2\;(\sqrt{98}+1)\;4\). First off, that zero at the very beginning—yeah, adding zero does nothing, right? So we can just drop it and work with \(2(\sqrt{98}+1)4\). Student B: Hold on a second. It’s true that \(0 + x = x\), but make sure we keep...	950
Debate: Solving the Quintic \( \displaystyle \frac{3}{2}r^{5}-\frac{9351}{2}r^{4}+\frac{1038285}{2}r^{3}-\frac{33894705}{2}r^{2}+126054396\,r-109621512=0\) --- ### Debater 1 – “The Rational‑Root‑Hunter” **I argue that the most reliable and transparent way to crack this equation is a straight‑line applicati...	3,142
Looking at this problem, I first notice that everything is wrapped in a big outer exponent \(10/17\). So my ultimate goal is to simplify the thing inside the parentheses, then raise that result to \(10/17\). The inside looks like a mess of fractions and powers of \(q\): \[ \Bigl(\,\frac{q\cdot q}{\,q\cdot q^{3}}\;\B...	1,768
Teacher: Let’s look at this sequence together: \[ -815,\; -6881,\; -23421,\; -55673,\; -108875,\; -188265,\; -299081,\; -446561,\; \dots \] What do you notice at first glance? Any pattern jump out at you? Student: It’s all negative and the numbers seem to be getting bigger in magnitude. Maybe it’s multiply...	1,989
Student A: Okay, we have the function \(a(l) = -198l - 284\). To check if \(a(-13)\) is even we just plug in \(-13\). So it should be \(a(-13) = -198\cdot(-13) - 284\). Student B: Right, that’s the right substitution. Just remember the minus signs—two negatives make a positive for the product, then we still...	714
Teacher: Let’s look at the equation \[ 4v^{2}-3996v+7976=0. \] What type of equation is this, and what are the usual ways we can solve it? Student: It’s a quadratic equation. We could try factoring it, use the quadratic formula, or maybe complete the square. Teacher: Good. Before we decide which metho...	1,591
Teacher: Today we’ll differentiate the expression \[ a\,t^{4}-18\,a\,t^{2}-298160\,a+705\,t^{4} \] with respect to the variable \(t\). First, can you tell me what the variable is here and what should be treated as constants? Student: The variable is \(t\), so everything that isn’t \(t\) – that means \(...	1,320
Debater 1: I argue that the best approach is to eliminate the decimals, treat the numbers as whole integers, multiply, and then re‑insert the decimal point based on the total number of fractional places. - Write \(0.335784 = 335784 \times 10^{-6}\) and \(-1.7 = -17 \times 10^{-1}\). - Multiply the integer par...	1,252
Teacher: Today we have a little comparison problem. Which number is larger: \[ -157\qquad\text{or}\qquad -\frac{4357}{28}\;? \] Student: Hmmm… I’m not sure. My gut says ‑157 looks “bigger” because the other one is a fraction. Teacher: Let’s pause a moment. When both numbers are negative, what does...	1,288
Teacher: Let’s look at the statement we need to verify: > “For any angle α, the equation > \[ > 1 = (\sqrt{\alpha}\,\cdot\alpha)^{2}+\frac{1}{\cos\alpha} > \] > is always satisfied.” What do you think the first thing we should do with the left‑hand side of that equality? --- Student: Well, the left‑hand...	2,008
Student A: Okay, let’s crack this. We need the j‑th term of the sequence \[ -263,\;502,\;1299,\;2134,\;3013,\;3942. \] I’m pretty sure it’s some kind of polynomial pattern. Student B: Right, the first thing we should do is look at the differences. If the first differences are constant, it’s arithmetic; if ...	1,906
Interviewer: How did you first approach this problem? Subject Matter Expert: The first thing I did was recognize the nature of the task: we have a linear function defined by \(u(a)=3981a+79656\) and we need its value at a specific input, \(a=-20\). So the natural approach is direct substitution—replace \(a\)...	1,473

End of preview. Expand in Data Studio

Math-Reasoning

Dataset Description

Mathematical problem-solving, rewriting, and dialogue data for reasoning-oriented language-model training. This repository is part of the K2 Horizon collection.

The repository is organized into multiple subsets. Every subset has a train split backed by Parquet shards, which supports Dataset Viewer inspection and streaming access.

K2 Horizon Dataset Series

Dataset repository	Focus	Subsets
IFM/TxT360-v2	Web and question-answering text	3
IFM/Code-Reasoning	Code reasoning and task synthesis	7
IFM/Math-Reasoning	Mathematical reasoning and dialogue	5
IFM/SFT-Reasoning	Instruction following and SFT-style data	2
IFM/Pretrain-Behaviors	Behavior-focused pretraining data	7

Dataset Subsets

Subset	Data files
`math-thinking-qwen`	`math-thinking-qwen/*.parquet`
`math-thinking-oss`	`math-thinking-oss/*.parquet`
`math-rewrite`	`math-rewrite/*.parquet`
`math-dialogue`	`math-dialogue/*.parquet`
`socratic-math-dialogue`	`socratic-math-dialogue/*.parquet`

Repository Structure

README.md
math-thinking-qwen/
  <source-file>-<stable-id>-00000.parquet
  <source-file>-<stable-id>-00001.parquet
math-thinking-oss/
  <source-file>-<stable-id>-00000.parquet
  <source-file>-<stable-id>-00001.parquet
math-rewrite/
  <source-file>-<stable-id>-00000.parquet
  <source-file>-<stable-id>-00001.parquet
math-dialogue/
  <source-file>-<stable-id>-00000.parquet
  <source-file>-<stable-id>-00001.parquet
socratic-math-dialogue/
  <source-file>-<stable-id>-00000.parquet
  <source-file>-<stable-id>-00001.parquet

The shard prefix is derived from the source JSONL filename and a stable identifier. Updating one source JSONL file replaces only that file's Parquet shards.

Data Fields

Records originate as JSON objects and are converted to Parquet for release. Field names and nested structures can differ by subset. Inspect features before building a processing pipeline:

from datasets import load_dataset

dataset = load_dataset(
    "IFM/Math-Reasoning",
    "math-thinking-qwen",
    split="train",
    streaming=True,
)
print(dataset.features)
print(next(iter(dataset)))

Data Provenance and Processing

Individual subsets may have undergone source-specific filtering, cleaning, deduplication, quality scoring, or synthetic-data generation. Users should evaluate each subset for their target use case and inspect the available provenance metadata.

Intended Use

This dataset is intended for language-model training and research. The subsets can be streamed independently, combined with user-defined sampling weights, or inspected through the Hugging Face Dataset Viewer.

Limitations and Responsible Use

Large-scale training data can contain factual errors, duplicated material, sensitive topics, stereotypes, unsafe content, and other artifacts. Dataset users are responsible for performing evaluations, risk assessment, and filtering appropriate to their application.

License and Terms of Use

This dataset is licensed under the Apache License 2.0 available at https://www.apache.org/licenses/LICENSE-2.0.

Downloads last month: 549

Collection including IFM/Math-Reasoning

K2 Horizon

Collection

K2 Horizon models, datasets, and supporting resources • 22 items • Updated about 9 hours ago • 76