agentlans/small-magpie

---
configs:
- config_name: Magpie-Pro-10K-GPT4o-mini
data_files:
- path:
- Magpie-Pro-10K-GPT4o-mini.jsonl.zst
split: train
- config_name: all
data_files:
- path:
- all.jsonl.zst
split: train
- config_name: magpie-ultra-v0.1
data_files:
- path:
- magpie-ultra-v0.1.jsonl.zst
split: train
- config_name: sample_k100
data_files:
- path:
- sample_k100.jsonl.zst
split: train
- config_name: sample_k1000
data_files:
- path:
- sample_k1000.jsonl.zst
split: train
- config_name: sample_k10000
data_files:
- path:
- sample_k10000.jsonl.zst
split: train
default: true
- config_name: sample_k200
data_files:
- path:
- sample_k200.jsonl.zst
split: train
- config_name: sample_k2000
data_files:
- path:
- sample_k2000.jsonl.zst
split: train
- config_name: sample_k20000
data_files:
- path:
- sample_k20000.jsonl.zst
split: train
- config_name: sample_k500
data_files:
- path:
- sample_k500.jsonl.zst
split: train
- config_name: sample_k5000
data_files:
- path:
- sample_k5000.jsonl.zst
split: train
- config_name: sample_k50000
data_files:
- path:
- sample_k50000.jsonl.zst
split: train
task_categories:
- text-generation
language:
- en
tags:
- magpie
- supervised-fine-tuning
---
# Smaller Magpie

A collection of smaller Magpie datasets compared to [agentlans/magpie](https://huggingface.co/datasets/agentlans/magpie).

For `argilla/magpie-ultra-v0.1`, only instructions rated as good or excellent were selected.
`output_quality` corresponds to the original dataset’s `score_difference`, which is the gap between instruct model and base model responses as evaluated by a reward model.
Please see the original dataset for details.

| Source | Rows |
|-----|------:|
| [argilla/magpie-ultra-v0.1](https://huggingface.co/datasets/argilla/magpie-ultra-v0.1) | 43923 |
| [Mxode/Magpie-Pro-10K-GPT4o-mini](https://huggingface.co/datasets/Mxode/Magpie-Pro-10K-GPT4o-mini) | 10000 |