|
|
--- |
|
|
license: apache-2.0 |
|
|
--- |
|
|
|
|
|
<img src="https://cdn-uploads.huggingface.co/production/uploads/686c5c546abedce0f7ac048a/B7PeaDQCDnlgT3Tmf7fsb.png" width=250> |
|
|
|
|
|
# Instinct, the State-of-the-Art Open Next-Edit Model |
|
|
|
|
|
This repo contains the model weights for **Instinct**, [Continue](https://continue.dev)'s state-of-the-art open next-edit model. Robustly fine-tuned from Qwen2.5-Coder-7B, Instinct intelligently predicts your next move to keep you in flow. |
|
|
|
|
|
## Serving the model |
|
|
|
|
|
There are many ways to serve a local model with Continue. If you wish to serve the model using [SGLang](https://github.com/sgl-project/sglang), as we did internally, you can use the following command, optionally adding `--quantization fp8` if desired. |
|
|
|
|
|
`python3 -m sglang.launch_server --model-path continuedev/instinct --load-format safetensors` |
|
|
|
|
|
## Learn more |
|
|
|
|
|
For more information on the work behind Instinct, please refer to our blog. |
|
|
|
