|
|
--- |
|
|
license: apache-2.0 |
|
|
datasets: |
|
|
- continuedev/instinct-data |
|
|
--- |
|
|
|
|
|
<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 on our [dataset of real-world code edits](https://huggingface.co/datasets/continuedev/instinct-data), Instinct intelligently predicts your next move to keep you in flow. |
|
|
|
|
|
## Serving the model |
|
|
|
|
|
**Ollama**: We've released a [Q4_K_M GGUF quantization of Instinct](https://huggingface.co/continuedev/instinct-GGUF) for efficient local inference. Try it with [Continue's Ollama integration](https://docs.continue.dev/guides/ollama-guide). |
|
|
|
|
|
Besides Ollama, there are many ways to plug a local model into Continue; we internally used an endpoint served by [SGLang](https://github.com/sgl-project/sglang), which is one of the options below. Quantizing for faster inference is also an option that worked well for us. Serve the model using either of the below options, then [connect it with Continue](https://docs.continue.dev/guides/how-to-self-host-a-model). |
|
|
|
|
|
SGLang: `python3 -m sglang.launch_server --model-path continuedev/instinct --load-format safetensors` |
|
|
<br>vLLM : `vllm serve continuedev/instinct --served-model-name instinct --load-format safetensors` |
|
|
|
|
|
## Learn more |
|
|
|
|
|
For more information on the work behind Instinct, please refer to our blog. |
