v0.1.16
This release adds support Mixtral and other models based on its Mixture of Experts (MoE) architecture. Note: this model requires at least 48GB of memory.
ollama run mixtral
New models
- Mixtral: A high-quality mixture of experts model with open weights.
- Dolphin Mixtral: An uncensored, fine-tuned model based on the Mixtral mixture of experts model that excels at coding tasks.
What's Changed
- Add support Mixtral and other models based on its Mixture of Experts (MoE) architecture
- Fixed issue where
load_duration
was not in the response for/api/generate
Full Changelog: v0.1.15...v0.1.16