WI

WizardLM-2-8x22B

WizardlmGGUF by backyardaiUpdated 2024-05-22
Download on Hugging Face
Parameters141B
Context64K
Layers56
Architecturellama
Smallest quant29.6 GB
Quick answer

WizardLM-2-8x22B needs about 88.0 GB of memory at Q4_K_M (a 85.6 GB file) with an 8K context window, which is more than a 48 GB GPU holds. This repository has 22 quants, from 29.6 GB (IQ1_S) to 281.2 GB (F16). Longer context needs more: at 32K tokens the KV cache alone adds about 7.5 GB.

WizardLM-2-8x22B – 141B parameter LLM with 65536 token context

WizardLM-2-8x22B is a 141B parameter model from Microsoft, trained on 65536 tokens. It is available in GGUF format and supports large context lengths. The model is licensed under apache-2.0.

Good at

  • Large context reasoning
  • Complex chat dialogue
  • High parameter density

Limits

  • No vision support
  • Requires large VRAM for full precision
  • License is apache-2.0

Best for

  • Enterprise chatbots
  • Research prototyping
  • Long-form content generation

Who it’s for. Developers and researchers needing a high‑capacity LLM for local deployment. The model is suitable for users with GPUs that can handle large memory footprints and who prefer an open‑source license.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
IQ1_S29.6 GB 32.0 GB
Too big Get file →
IQ2_XXS37.9 GB 40.3 GB
Too big Get file →
IQ2_XS42.0 GB 44.4 GB
Too big Get file →
IQ2_S42.6 GB 45.0 GB
Too big Get file →
IQ2_M46.7 GB 49.1 GB
Too big Get file →
Q2_K_S48.1 GB 50.5 GB
Too big Get file →
Q2_K52.1 GB (2 parts) 54.5 GB
Too big Get file →
IQ3_XXS54.9 GB (2 parts) 57.3 GB
Too big Get file →
IQ3_XS58.2 GB (2 parts) 60.6 GB
Too big Get file →
IQ3_S61.5 GB (2 parts) 63.9 GB
Too big Get file →
Q3_K_S61.5 GB (2 parts) 63.9 GB
Too big Get file →
IQ3_M64.5 GB (2 parts) 66.9 GB
Too big Get file →
Q3_K_M67.8 GB (2 parts) 70.2 GB
Too big Get file →
Q3_K_L72.6 GB (2 parts) 75.0 GB
Too big Get file →
IQ4_XS75.5 GB (2 parts) 77.9 GB
Too big Get file →
Q4_K_S80.5 GB (2 parts) 82.9 GB
Too big Get file →
Q4_K_M85.6 GB (2 parts) 88.0 GB
Too big Get file →
Q5_K_S97.0 GB (2 parts) 99.4 GB
Too big Get file →
Q5_K_M100.0 GB (2 parts) 102.3 GB
Too big Get file →
Q6_K115.5 GB (3 parts) 117.9 GB
Too big Get file →
Q8_0149.4 GB (3 parts) 151.8 GB
Too big Get file →
F16281.2 GB (6 parts) 283.6 GB
Too big Get file →

For 8 GB VRAM, use IQ1_S (29.64 GB file, 32 GB memory at 8k context). For 12 GB VRAM, IQ2_XXS (37.88 GB file, 40.3 GB memory). For 16 GB VRAM, Q2_K (52.1 GB file, 54.5 GB memory).

What GPU do you need for WizardLM-2-8x22B?

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: no quant fits at an 8K context.
  • 24 GB VRAM: no quant fits at an 8K context.
  • 32 GB VRAM: the best quant that fits is IQ1_S (29.6 GB file, about 32.0 GB in use).
  • 48 GB VRAM: the best quant that fits is IQ2_S (42.6 GB file, about 45.0 GB in use).

Which WizardLM-2-8x22B quant should you choose?

Q4_K_M is the usual starting point for WizardLM-2-8x22B: it balances file size and quality well. Here it is a 85.6 GB file.

The Q8_0 file is 149.4 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 IQ1_S at 29.6 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
Parameters141B (approximate, mixture of experts)
Layers56
Trained context65,536 tokens
KV cache (fp16)224 KB per token

Source

Quantized bybackyardai
FamilyWizardlm
Licenseapache-2.0
Downloads (HF)2,058
Last updated2024-05-22

Frequently asked questions about WizardLM-2-8x22B

How much VRAM does WizardLM-2-8x22B need?

About 88.0 GB at Q4_K_M with an 8K context window. That is the 85.6 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 WizardLM-2-8x22B on an 8 GB GPU?

Not at an 8K context. The smallest file needs about 32.0 GB, so you would need a larger GPU or CPU offloading.

What is the file size of WizardLM-2-8x22B GGUF?

It depends on the quant: IQ1_S is 29.6 GB, IQ2_XXS is 37.9 GB, IQ2_XS is 42.0 GB, IQ2_S is 42.6 GB, IQ2_M is 46.7 GB, Q2_K_S is 48.1 GB and 16 more.

What context length does WizardLM-2-8x22B support?

The GGUF header lists a trained context length of 65,536 tokens (64K). The KV cache grows with the context you use: about 1.9 GB at 8K and 7.5 GB at 32K.

Where can I download WizardLM-2-8x22B GGUF?

The files are on Hugging Face in the backyardai/WizardLM-2-8x22B-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 WizardLM-2-8x22B use?

Hugging Face lists the license as apache-2.0. 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 WizardLM-2-8x22B on 8 GB VRAM?

Use the IQ1_S quantization; it requires about 32 GB memory at 8k context, which is beyond 8 GB VRAM, so it is not feasible.

What is the maximum context length supported?

The model supports up to 65536 tokens of context.

Is the model licensed for commercial use?

It is licensed under apache-2.0, which allows commercial use with attribution.

Other GGUFs of WizardLM-2-8x22B

Similar models

Browse all →

More from backyardai

All by backyardai →

More Wizardlm models

All Wizardlm →

Other large models

All large models →