OpenHermes 2.5 Mistral 7B
OpenHermes 2.5 adds code-focused training to the Mistral-based OpenHermes family.
Training and model behavior
The training mix contains about one million examples. The card estimates that coding instructions account for roughly seven to fourteen percent of the data. It reports gains on some non-code benchmarks as well as coding, with a decline on another benchmark.OpenHermes 2.5 Mistral 7B model card
Working with this model
Practical guidance. Compare the release with OpenHermes 2 on your task mix. An aggregate gain can conceal a regression in a specific instruction or reasoning category.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Launch date | 2023 |
| Access | Downloadable |
| Model identifiers | teknium/OpenHermes-2.5-Mistral-7B |
| Context window | 32,768 tokens Configured position limit; practical serving limits depend on the runtime. OpenHermes 2.5 Mistral 7B configuration |
| Base model | mistralai/Mistral-7B-v0.1OpenHermes 2.5 Mistral 7B model card |
| Repository license label | apache-2.0OpenHermes 2.5 Mistral 7B model card |
Related models
- OpenHermes 2 Mistral 7B
Nous Research · OpenHermes
A Mistral instruction model published by Teknium.
Launch date October 2023Downloadable - OpenHermes 13B
Nous Research · OpenHermes
A Teknium instruction model trained on a public Hermes dataset.
Launch date 2023Downloadable - OpenHermes 7B
Nous Research · OpenHermes
A Teknium instruction model trained on a public Hermes dataset.
Launch date 2023Downloadable