Dataset · Open-Orca
Augmented FLAN for instruction tuning
Open-Orca/OpenOrca
A collection of augmented FLAN data aligned with the Orca distribution for instruction tuning.
What's inside
OpenOrca is a collection of augmented FLAN data closely aligned with the distribution from the Orca paper. The data was used to obtain strong model checkpoints; the card mentions Mistral-7B-OpenOrca. Size 1M–10M, language en, modality text.
What you can do with it
- SFT for instruction tuning
- Training LoRA adapter
- Evaluating instruction following
- Tasks
- Text classification, Token classification, Table question answering, Question answering, Zero shot classification, Summarization
- Languages
- English
- Size
- 1M–10M rows
- Downloads this month
- 11.6K
- Likes
- 1.6K
Models trained on it
- upstage/SOLAR-10.7B-Instruct-v1.0Text generation 43.7K
- uukuguy/speechless-code-mistral-7b-v1.0Text generation 9.2K
- TheBloke/Mistral-7B-OpenOrca-GGUFText generation 7.6K
- TheBloke/CausalLM-7B-GGUFText generation 6.6K
- TheBloke/CausalLM-14B-GGUFText generation 4.4K
- TheBloke/PuddleJumper-13B-GGUF 3.3K
- mradermacher/Aitana-7B-S-Instruct-i1-GGUFText generation 2.8K
- tastypear/CausalLM-7B-DPO-alpha-GGUFText generation 2.4K
How to load it
from datasets import load_dataset
ds = load_dataset("Open-Orca/OpenOrca")