Dataset · teknium
OpenHermes 2.5 instruction mixture
teknium/OpenHermes-2.5
Mixture of open and synthetic instructions for training chat models.
What's inside
An English text dataset used to train OpenHermes 2/2.5 and Nous Hermes 2 models. It is a compilation of many open datasets and specially created synthetic data, with a size ranging from 1M to 10M examples.
What you can do with it
- SFT of chat models on instructions
- Train LoRA for a conversational assistant
- Evaluate instruction-following
- Languages
- English
- Size
- 1M–10M rows
- Downloads this month
- 16.9K
- Likes
- 934
Models trained on it
- dphn/dolphin-2.9.1-yi-1.5-34bText generation 4.8M
- NousResearch/Hermes-2-Pro-Mistral-7B-GGUF 71.2K
- HuggingFaceTB/SmolLM-135M-InstructText generation 23.9K
- NousResearch/Nous-Hermes-2-Mixtral-8x7B-DPOText generation 20.7K
- bartowski/dolphin-2.9-llama3-8b-GGUFText generation 19.8K
- NousResearch/Hermes-2-Theta-Llama-3-8BText generation 12.6K
- NousResearch/Nous-Hermes-2-SOLAR-10.7BText generation 11.2K
- HuggingFaceTB/SmolLM-1.7B-InstructText generation 11K
How to load it
from datasets import load_dataset
ds = load_dataset("teknium/OpenHermes-2.5")