SK

skywork_skywork-13b-base

SkyworkGGUF by TheBlokeUpdated 2023-10-31
Download on Hugging Face
Parameters13.9B
Context128K
Layers52
Architecturellama
Smallest quant5.8 GB
Quick answer

skywork_skywork-13b-base needs about 16.7 GB of memory at Q4_K_M (a 8.4 GB file) with an 8K context window, so it fits on a 24 GB GPU. This repository has 12 quants, from 5.8 GB (Q2_K) to 14.7 GB (Q8_0). Longer context needs more: at 32K tokens the KV cache alone adds about 31.4 GB.

Skywork-13B-Base, a 13.9B parameter Llama‑style model

Skywork-13B-Base is a 13.9B parameter Llama architecture model trained on 3.2TB of cleaned multilingual data, mainly Chinese and English, with 8% code. It supports a 131072 token context window and is available in multiple GGUF quantisations.

Good at

  • Large context handling
  • Multilingual text generation
  • Code generation

Limits

  • No vision support
  • License is other
  • No mixture of experts

Best for

  • Chat applications
  • Code assistance
  • Long‑form content creation

Who it’s for. Developers and researchers needing a large‑scale Llama‑style model with strong multilingual and code capabilities. Suitable for local inference with GGUF support.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
Q2_K5.8 GB 14.2 GB
Too big Get file →
Q3_K_S6.1 GB 14.4 GB
Too big Get file →
Q3_K_M6.8 GB 15.1 GB
Too big Get file →
Q3_K_L7.4 GB 15.8 GB
Too big Get file →
Q4_07.9 GB 16.2 GB
Too big Get file →
Q4_K_S7.9 GB 16.3 GB
Too big Get file →
Q4_K_M8.4 GB 16.7 GB
Too big Get file →
Q5_09.6 GB 17.9 GB
Too big Get file →
Q5_K_S9.6 GB 17.9 GB
Too big Get file →
Q5_K_M9.8 GB 18.2 GB
Too big Get file →
Q6_K11.4 GB 19.7 GB
Too big Get file →
Q8_014.7 GB 23.1 GB
Too big Get file →

For 8 GB VRAM, use Q4_K_M (8.4 GB file, ~16.7 GB memory at 8 k context). For 12 GB VRAM, Q5_K_M (9.84 GB file, ~18.2 GB memory). For 16 GB VRAM, Q6_K (11.37 GB file, ~19.7 GB memory).

What GPU do you need for skywork_skywork-13b-base?

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: no quant fits at an 8K context.
  • 12 GB VRAM: no quant fits at an 8K context.
  • 16 GB VRAM: the best quant that fits is Q3_K_L (7.4 GB file, about 15.8 GB in use).
  • 24 GB VRAM: the best quant that fits is Q8_0 (14.7 GB file, about 23.1 GB in use).

Which skywork_skywork-13b-base quant should you choose?

Q4_K_M is the usual starting point for skywork_skywork-13b-base: it balances file size and quality well. Here it is a 8.4 GB file.

The Q8_0 file is 14.7 GB, about 1.8 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 5.8 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

Architecturellama
Parameters13.9B
Layers52
Trained context131,072 tokens
KV cache (fp16)936 KB per token

Source

Quantized byTheBloke
FamilySkywork
Licenseother
Downloads (HF)1,849
Last updated2023-10-31

Frequently asked questions about skywork_skywork-13b-base

How much VRAM does skywork_skywork-13b-base need?

About 16.7 GB at Q4_K_M with an 8K context window. That is the 8.4 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 skywork_skywork-13b-base on an 8 GB GPU?

Not at an 8K context. The smallest file needs about 14.2 GB, so you would need a larger GPU or CPU offloading.

What is the file size of skywork_skywork-13b-base GGUF?

It depends on the quant: Q2_K is 5.8 GB, Q3_K_S is 6.1 GB, Q3_K_M is 6.8 GB, Q3_K_L is 7.4 GB, Q4_0 is 7.9 GB, Q4_K_S is 7.9 GB and 6 more.

What context length does skywork_skywork-13b-base support?

The GGUF header lists a trained context length of 131,072 tokens (128K). The KV cache grows with the context you use: about 7.9 GB at 8K and 31.4 GB at 32K.

Where can I download skywork_skywork-13b-base GGUF?

The files are on Hugging Face in the TheBloke/Skywork-13B-base-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 skywork_skywork-13b-base use?

Hugging Face lists the license as other. 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 Skywork-13B-Base on 8 GB VRAM?

Use the Q4_K_M quantised model; it requires about 16.7 GB VRAM at 8 k context, which exceeds 8 GB, so it is not feasible.

Does the model support Chinese and English?

Yes, the training data is 39.6% Chinese and 52.2% English, with 8% code.

Is there a vision capability?

No, the model does not support vision.

Similar models

Browse all →

More from TheBloke

All by TheBloke →

More Skywork models

All Skywork →

Other medium models

All medium models →