Local LLM model page

StarCoder 2 (15B)

Transparent open code LLM. 600+ languages. Trained on The Stack v2. 1.3M downloads.

Parameters
15B
Minimum RAM
12 GB
Model size
9 GB
Quantization
Q5_K_M

Can StarCoder 2 (15B) run locally?

StarCoder 2 (15B) is best suited for mainstream Macs and PCs with 16 GB RAM. LocalClaw recommends Q5_K_M as the default quantization, with at least 12 GB RAM.

Search term for LM Studio or compatible runtimes: starcoder2-15b

Hugging Face repository: bartowski/starcoder2-15b-GGUF

codepower

Strengths

  • Transparent open code LLM. 600+ languages. Trained on The Stack v2. 1.3M downloads.

Limitations

  • Performance depends heavily on quantization, RAM bandwidth and runtime support.

Best use cases

  • code
  • power

Benchmarks

Speed: 6/10

Quality: 6/10

Coding: 8/10

Reasoning: 5/10

Technical details

Developer: starcoder

License: See model repository

Context window: Unknown tokens

Architecture: See model card

Released: 2024-02