chargoddard/WebInstructSub-prometheus

---
language:
- en
license: apache-2.0
size_categories:
- 1M<n<10M
dataset_info:
features:
- name: instruction
dtype: string
- name: generation
dtype: string
- name: model_name
dtype: string
- name: feedback
dtype: string
- name: result
dtype: int64
splits:
- name: train
num_bytes: 5133062799
num_examples: 2393650
download_size: 2513481071
dataset_size: 5133062799
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
tags:
- synthetic
- distilabel
- rlaif
---

<p align="left">
<a href="https://github.com/argilla-io/distilabel">
<img src="https://raw.githubusercontent.com/argilla-io/distilabel/main/docs/assets/distilabel-badge-light.png" alt="Built with Distilabel" width="200" height="32"/>
</a>
</p>

# Dataset Card for WebInstructSub-prometheus

This dataset has been created with [distilabel](https://distilabel.argilla.io/).

## Dataset Summary

[TIGER-Lab/WebInstructSub](https://huggingface.co/datasets/TIGER-Lab/WebInstructSub?row=1) evaluated for logical and effective reasoning using [prometheus-7b-v2.0](https://huggingface.co/prometheus-eval/prometheus-7b-v2.0).

This dataset contains a `pipeline.yaml` which can be used to reproduce the pipeline that generated it in distilabel using the `distilabel` CLI:

```console
distilabel pipeline run --config "https://huggingface.co/datasets/chargoddard/WebInstructSub-prometheus/raw/main/pipeline.yaml"
```

or explore the configuration:

```console
distilabel pipeline info --config "https://huggingface.co/datasets/chargoddard/WebInstructSub-prometheus/raw/main/pipeline.yaml"
```

## Dataset structure

The examples have the following structure per configuration:

<details><summary> Configuration: default </summary><hr>

```json
{
"feedback": "The response provided a logical explanation for the discrepancy between the two formulas, demonstrating a strong understanding of the subject matter. The respondent correctly identified that both formulas are mathematically equivalent and should yield the same result. Furthermore, the response effectively corrected the computation error, showing the ability to apply logical principles to resolve complex issues. The clear and coherent explanation, combined with the correct calculation, demonstrates advanced reasoning abilities that are in line with the score rubric\u0027s criteria. So the overall score is 5.",
"generation": "There\u0027s no discrepancy between the two formulas; both are mathematically equivalent and should give the same result. The error lies in the calculation. Let\u0027s correct the computation for the second formula:\n\nRt = (Ra * Rb) / (Ra + Rb) \nRt = (100 \u03a9 * 220 \u03a9) / (100 \u03a9 + 220 \u03a9) \nRt = 22000 \u03a9 / 320 \u03a9 \nRt \u2248 68.75 \u03a9 \n\nAs you can see, both methods correctly yield approximately 68.75 \u03a9. The issue was likely a miscalculation or incorrect input in your original attempt. Remember, the two formulas are indeed equivalent, and you can use either to calculate the total resistance in a parallel circuit with two resistors.",
"instruction": "When two resistors are connected in parallel, the total resistance, Rt, can be calculated using the formula: Rt = 1/G, where G is the sum of the reciprocals of the individual resistances (G = 1/Ra + 1/Rb). However, for two resistors specifically, my book also provides the formula: Rt = (Ra * Rb) / (Ra + Rb). I noticed that applying these formulas with Ra = 100 \u03a9 and Rb = 220 \u03a9 gives different results: 68.75 \u03a9 using the first formula and 66.67 \u03a9 using the second. Can you explain the discrepancy and correct the error if there is one?",
"model_name": "prometheus-eval/prometheus-7b-v2.0",
"result": 5
}
```

This subset can be loaded as:

```python
from datasets import load_dataset

ds = load_dataset("chargoddard/WebInstructSub-prometheus", "default")
```

Or simply as it follows, since there's only one configuration and is named `default`:

```python
from datasets import load_dataset

ds = load_dataset("chargoddard/WebInstructSub-prometheus")
```

</details>

原始信息汇总

数据集卡片 for WebInstructSub-prometheus

数据集概述

TIGER-Lab/WebInstructSub 用于评估逻辑和有效推理，使用 prometheus-7b-v2.0。

数据集结构

数据集包含以下特征：

• instruction: 字符串类型
• generation: 字符串类型
• model_name: 字符串类型
• feedback: 字符串类型
• result: 整数类型

数据集分为以下部分：

• train 部分：包含 2393650 个样本，总大小为 5133062799 字节

数据集的下载大小为 2513481071 字节，数据集总大小为 5133062799 字节。

数据集包含一个默认配置：

• default 配置：包含 train 部分，路径为 data/train-*

数据集标签包括：

• synthetic
• distilabel
• rlaif

数据集示例结构如下：

json { "feedback": "The response provided a logical explanation for the discrepancy between the two formulas, demonstrating a strong understanding of the subject matter. The respondent correctly identified that both formulas are mathematically equivalent and should yield the same result. Furthermore, the response effectively corrected the computation error, showing the ability to apply logical principles to resolve complex issues. The clear and coherent explanation, combined with the correct calculation, demonstrates advanced reasoning abilities that are in line with the score rubrics criteria. So the overall score is 5.", "generation": "Theres no discrepancy between the two formulas; both are mathematically equivalent and should give the same result. The error lies in the calculation. Lets correct the computation for the second formula:

Rt = (Ra * Rb) / (Ra + Rb)
Rt = (100 Ω * 220 Ω) / (100 Ω + 220 Ω)
Rt = 22000 Ω / 320 Ω
Rt ≈ 68.75 Ω

As you can see, both methods correctly yield approximately 68.75 Ω. The issue was likely a miscalculation or incorrect input in your original attempt. Remember, the two formulas are indeed equivalent, and you can use either to calculate the total resistance in a parallel circuit with two resistors.", "instruction": "When two resistors are connected in parallel, the total resistance, Rt, can be calculated using the formula: Rt = 1/G, where G is the sum of the reciprocals of the individual resistances (G = 1/Ra + 1/Rb). However, for two resistors specifically, my book also provides the formula: Rt = (Ra * Rb) / (Ra + Rb). I noticed that applying these formulas with Ra = 100 Ω and Rb = 220 Ω gives different results: 68.75 Ω using the first formula and 66.67 Ω using the second. Can you explain the discrepancy and correct the error if there is one?", "model_name": "prometheus-eval/prometheus-7b-v2.0", "result": 5 }

数据集可以通过以下方式加载：

python from datasets import load_dataset

ds = load_dataset("chargoddard/WebInstructSub-prometheus", "default")

或者简化为：

python from datasets import load_dataset

ds = load_dataset("chargoddard/WebInstructSub-prometheus")