ST

starcoder2-15b-instruct-v0.1

StarCoderGGUF by second-stateUpdated 2024-05-14
Download on Hugging Face
Parameters16B
Context16K
Layers40
Architecturestarcoder2
Smallest quant6.2 GB
Quick answer

starcoder2-15b-instruct-v0.1 needs about 11.0 GB of memory at Q4_K_M (a 9.9 GB file) with an 8K context window, so it fits on a 12 GB GPU. This repository has 13 quants, from 6.2 GB (Q2_K) to 31.9 GB (F16).

StarCoder2-15B-Instruct-v0.1 – a 15B code generation model

StarCoder2-15B-Instruct-v0.1 is a 15B parameter model fine‑tuned for single‑turn coding instructions. It is trained on a self‑aligned instruction dataset and released under the bigcode-openrail-m license. The model is optimized for Python code generation and can be run with llama.cpp.

Good at

  • Python code generation
  • Single‑turn coding instructions
  • Self‑aligned instruction response

Limits

  • No vision support
  • Limited to coding instructions
  • Requires 8k context for full performance

Best for

  • Writing helper functions
  • Debugging snippets
  • Generating code examples

Who it’s for. Developers and researchers needing a code‑centric LLM. The model is suitable for building coding assistants and educational tools where instruction‑based code generation is required.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
Q2_K6.2 GB 7.4 GB
Tight Get file →
Q3_K_S7.0 GB 8.2 GB
Too big Get file →
Q3_K_M8.0 GB 9.2 GB
Too big Get file →
Q3_K_L9.0 GB 10.1 GB
Too big Get file →
Q4_09.1 GB 10.2 GB
Too big Get file →
Q4_K_S9.2 GB 10.3 GB
Too big Get file →
Q4_K_M9.9 GB 11.0 GB
Too big Get file →
Q5_011.0 GB 12.2 GB
Too big Get file →
Q5_K_S11.0 GB 12.2 GB
Too big Get file →
Q5_K_M11.4 GB 12.6 GB
Too big Get file →
Q6_K13.1 GB 14.3 GB
Too big Get file →
Q8_017.0 GB 18.1 GB
Too big Get file →
F1631.9 GB 33.1 GB
Too big Get file →

For 8 GB VRAM, use Q4_K or Q3_K_S (est. 10.2 GB and 8.2 GB). For 12 GB VRAM, Q4_K_M or Q5_0 (est. 11 GB and 12.2 GB). For 16 GB VRAM, Q5_K_M or Q6_K (est. 12.6 GB and 14.3 GB).

What GPU do you need for starcoder2-15b-instruct-v0.1?

These figures add the file size, an fp16 KV cache at an 8K context window (worked out from this model’s own header) and about 0.5 GB for the runtime. They are estimates, so leave a little headroom, and expect to use less context or a smaller quant if you are right at the limit.

  • 4 GB VRAM: no quant fits at an 8K context.
  • 6 GB VRAM: no quant fits at an 8K context.
  • 8 GB VRAM: the best quant that fits is Q2_K (6.2 GB file, about 7.4 GB in use).
  • 12 GB VRAM: the best quant that fits is Q4_K_M (9.9 GB file, about 11.0 GB in use).
  • 16 GB VRAM: the best quant that fits is Q6_K (13.1 GB file, about 14.3 GB in use).
  • 24 GB VRAM: the best quant that fits is Q8_0 (17.0 GB file, about 18.1 GB in use).
  • 32 GB VRAM: the best quant that fits is Q8_0 (17.0 GB file, about 18.1 GB in use).
  • 48 GB VRAM: the best quant that fits is F16 (31.9 GB file, about 33.1 GB in use).

Which starcoder2-15b-instruct-v0.1 quant should you choose?

Q4_K_M is the usual starting point for starcoder2-15b-instruct-v0.1: it balances file size and quality well. Here it is a 9.9 GB file.

The Q8_0 file is 17.0 GB, about 1.7 times the size of Q4_K_M. It keeps the weights closer to the original, which only matters if you have the memory to spare.

If memory is tight, the smallest file is Q2_K at 6.2 GB, with the quality trade-offs that come with fewer bits. Use it only when memory is very tight. Expect a clear drop in quality, which is usually worse on small models and easier to live with on large ones.

New to quant names? Read GGUF quantization explained.

Specifications

Architecture from the GGUF header

Architecturestarcoder2
Parameters16B
Layers40
Trained context16,384 tokens
KV cache (fp16)80 KB per token

Source

Quantized bysecond-state
FamilyStarCoder
Downloads (HF)1,450
Last updated2024-05-14

Frequently asked questions about starcoder2-15b-instruct-v0.1

How much VRAM does starcoder2-15b-instruct-v0.1 need?

About 11.0 GB at Q4_K_M with an 8K context window. That is the 9.9 GB file, plus the KV cache and roughly 0.5 GB of runtime overhead. Smaller quants need less and longer context needs more.

Can I run starcoder2-15b-instruct-v0.1 on an 8 GB GPU?

Yes, with Q2_K: the 6.2 GB file needs about 7.4 GB at 8K context. Larger quants in this repository do not fit in 8 GB.

What is the file size of starcoder2-15b-instruct-v0.1 GGUF?

It depends on the quant: Q2_K is 6.2 GB, Q3_K_S is 7.0 GB, Q3_K_M is 8.0 GB, Q3_K_L is 9.0 GB, Q4_0 is 9.1 GB, Q4_K_S is 9.2 GB and 7 more.

What context length does starcoder2-15b-instruct-v0.1 support?

The GGUF header lists a trained context length of 16,384 tokens (16K). The KV cache grows with the context you use: about 0.7 GB at 8K and 1.3 GB at 16K.

Where can I download starcoder2-15b-instruct-v0.1 GGUF?

The files are on Hugging Face in the second-state/StarCoder2-15B-Instruct-v0.1-GGUF repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.

What license does starcoder2-15b-instruct-v0.1 use?

Hugging Face lists the license as bigcode-openrail-m. Licenses can differ between the quantized files and the original model, so check the repository and the base model before commercial use.

Can I run StarCoder2-15B-Instruct-v0.1 on 8 GB VRAM?

Yes, using the Q4_K or Q3_K_S quantized model will fit within 8 GB VRAM, with estimated memory usage of 10.2 GB and 8.2 GB respectively.

What context length does the model support?

The model was trained with a context length of 16384 tokens, but the GGUF quantized versions support up to 6144 tokens.

Is the model licensed for commercial use?

The model is released under the bigcode-openrail-m license, which allows commercial use with certain attribution and usage constraints.

Other GGUFs of starcoder2-15b-instruct-v0.1

Similar models

Browse all →

More from second-state

All by second-state →

More StarCoder models

All StarCoder →

Other medium models

All medium models →