Dataset · argilla
Instructions and responses from MagPie Ultra v0.1
argilla/magpie-ultra-v0.1
50k unfiltered instruction-response pairs in English for training language models.
What's inside
The dataset contains instruction-response pairs; the first unfiltered version includes 50k examples. It was collected using distilabel according to the MagPie recipe. The language is English, and data is presented in tabular and text form. In v1.0, the same recipe is expanded to 1 million rows, adding multi-turn and more diversity.
What you can do with it
- SFT of instructional LLM
- Train LoRA for instruction following
- Compare with magpie-ultra-v1.0
- Tasks
- Text generation
- Languages
- English
- Size
- 10K–100K rows
- Downloads this month
- 2.6K
- Likes
- 230
Models trained on it
- tokyotech-llm/Llama-3.1-Swallow-8B-Instruct-v0.3Text generation 5.1K
- mradermacher/Llama-3.2-SUN-2.5B-chat-i1-GGUF 2.3K
- mradermacher/Llama-3.2-SUN-HDIC-1B-Instruct-i1-GGUF 1.4K
- mradermacher/Gemma-2-Llama-Swallow-9b-it-v0.1-Heretic-i1-GGUF 1.4K
- QuantFactory/Llama-3.2-3B-Agent007-Coder-GGUF 1.3K
- mradermacher/Llama-3.2-SUN-1B-chat-i1-GGUF 1.2K
- tokyotech-llm/Gemma-2-Llama-Swallow-9b-it-v0.1Text generation 1.2K
- meditsolutions/Llama-3.2-SUN-2.5B-chatText generation 933
How to load it
from datasets import load_dataset
ds = load_dataset("argilla/magpie-ultra-v0.1")