bigcode-project/starcoder2

Health v2 · maintenance only

Flags: no_releases

Adoption not part of the score

2087 stars · 198 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

StarCoder2 is a family of open code generation language models (3B, 7B, 15B) trained on 600+ programming languages from The Stack v2, with training, inference, and evaluation code in this repository. The models are designed for code completion with a 16,384-token context window and are distributed via Hugging Face under Apache-2.0.

Use cases

• generate code completions from a function signature
• fine-tune a code LLM on my own repository data
• run a local code generation model on GPU
• evaluate a code model on programming benchmarks
• train a multilingual code model from scratch

When to choose

• you need an open-weights code completion model with permissive licensing
• you want to fine-tune or study a transparently trained code LLM
• you need code completion across many programming languages

When to avoid

• you need an instruction-following chat assistant rather than raw code completion
• you have no GPU or limited memory for large model inference
• you want a hosted API without managing model weights yourself

Facets

library · maturity active

llm-inference llm-training machine-learning deep-learning large-language-models artificial-intelligence developer-tools python cross-platform code-generation code-completion transformer-models huggingface open-weights gpu linux

Member repositories

For agents

markdown · JSON · MCP: product_card(name="bigcode-project/starcoder2")