https://github.com/alienkevin/llama2
https://github.com/alienkevin/llama2
Last synced: 11 months ago
JSON representation
- Host: GitHub
- URL: https://github.com/alienkevin/llama2
- Owner: AlienKevin
- Created: 2023-08-01T01:53:25.000Z (over 2 years ago)
- Default Branch: master
- Last Pushed: 2024-05-06T12:02:36.000Z (over 1 year ago)
- Last Synced: 2025-01-23T21:25:00.417Z (about 1 year ago)
- Language: JavaScript
- Size: 27 MB
- Stars: 1
- Watchers: 2
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Set up
1. Download the code llama model from: https://huggingface.co/AlienKevin/codellama-34b.Q5_K_M.v3.gguf/blob/main/codellama-34b.Q5_K_M.v3.gguf
Put the model under `models/`.
2. Compile llama.cpp:
```
cd llama.cpp
make main
```
3. Install node.js and copy the compiled `lsp.js` into the project root
4. Run model:
```
chmod +x run-code-34b.sh
./run-code-34b.sh
```
Specify the prompt on the first line of `run-code-34b.sh`:
```
python expand_prompt.py autoregressive list.filter_int
```
`expand_prompt.py` will expand the `autorgressive.template`, write the expanded form
to `autoregressive.prelude`, and then concatenate the query from `queries/list.filter_int.txt` to the prelude and write the whole prompt (prelude + query) to `autoregressive.prompt`.