Spaces:
Configuration error
Configuration error
## Using llama.cpp in the web UI | |
#### Pre-converted models | |
Simply place the model in the `models` folder, making sure that its name contains `ggml` somewhere and ends in `.bin`. | |
#### Convert LLaMA yourself | |
Follow the instructions in the llama.cpp README to generate the `ggml-model-q4_0.bin` file: https://github.com/ggerganov/llama.cpp#usage | |
## Performance | |
This was the performance of llama-7b int4 on my i5-12400F: | |
> Output generated in 33.07 seconds (6.05 tokens/s, 200 tokens, context 17) | |
You can change the number of threads with `--threads N`. | |