self-hosted/ai
§01·model · /models

wizardlm2 7b

llmactiveApache-2.0

7B instruction-following LLM by Microsoft, trained with Evol-Instruct. Microsoft withdrew the original repository, so it is served from a community mirror — worth knowing before you build on it. One recipe covers it, on an 8 GB RTX 3060 Ti at 70.79 tokens/s in the shared 4-bit Ollama run, second only to Llama 2 7B on that card. Apache-2.0.

§02·GPUs that run this model
1 total

benchmarked·~ runs via recipe (not benchmarked)· untested·doesn't fit