QW

Qwen1.5 22B Chat Merge

QwenGGUF by DisOOMUpdated 2024-03-20
Download on Hugging Face
Parameters22B
Context32K
Layers65
Architectureqwen2
Smallest quant12.6 GB
Quick answer

Qwen1.5 22B Chat Merge needs about 26.7 GB of memory at Q5_0 (a 15.3 GB file) with an 8K context window, so it fits on a 32 GB GPU. This repository has 3 quants, from 12.6 GB (Q4_0) to 23.4 GB (Q8_0). Longer context needs more: at 32K tokens the KV cache alone adds about 43.6 GB.

Qwen1.5 22B Chat Merge is a 22B-parameter Qwen model in GGUF format, quantized by DisOOM. Its trained context length is 32K tokens. The repository has 3 quant files, from 12.6 GB (Q4_0) to 23.4 GB (Q8_0).

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
Q4_012.6 GB 24.0 GB
Too big Get file →
Q5_015.3 GB 26.7 GB
Too big Get file →
Q8_023.4 GB 34.8 GB
Too big Get file →

What GPU do you need for Qwen1.5 22B Chat Merge?

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: the best quant that fits is Q4_0 (12.6 GB file, about 24.0 GB in use).
  • 32 GB VRAM: the best quant that fits is Q5_0 (15.3 GB file, about 26.7 GB in use).
  • 48 GB VRAM: the best quant that fits is Q8_0 (23.4 GB file, about 34.8 GB in use).

Which Qwen1.5 22B Chat Merge quant should you choose?

Q5_0 (15.3 GB) is the middle option in this repository. Closer to the original model than 4-bit, for a moderate increase in size. A good pick when you have memory to spare.

If memory is tight, the smallest file is Q4_0 at 12.6 GB, with the quality trade-offs that come with fewer bits. The usual sweet spot between file size and quality, and the quant most people start with.

New to quant names? Read GGUF quantization explained.

Specifications

Architecture from the GGUF header

Architectureqwen2
Parameters22B
Layers65
Trained context32,768 tokens
KV cache (fp16)1300 KB per token

Source

Quantized byDisOOM
FamilyQwen
Licenseother
Downloads (HF)77
Last updated2024-03-20

Frequently asked questions about Qwen1.5 22B Chat Merge

How much VRAM does Qwen1.5 22B Chat Merge need?

About 26.7 GB at Q5_0 with an 8K context window. That is the 15.3 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 Qwen1.5 22B Chat Merge on an 8 GB GPU?

Not at an 8K context. The smallest file needs about 24.0 GB, so you would need a larger GPU or CPU offloading.

What is the file size of Qwen1.5 22B Chat Merge GGUF?

It depends on the quant: Q4_0 is 12.6 GB, Q5_0 is 15.3 GB, Q8_0 is 23.4 GB.

What context length does Qwen1.5 22B Chat Merge support?

The GGUF header lists a trained context length of 32,768 tokens (32K). The KV cache grows with the context you use: about 10.9 GB at 8K and 43.6 GB at 32K.

Where can I download Qwen1.5 22B Chat Merge GGUF?

The files are on Hugging Face in the DisOOM/Qwen1.5-22B-Chat-Merge-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 Qwen1.5 22B Chat Merge 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.

Similar models

Browse all →

More from DisOOM

All by DisOOM →

More Qwen models

All Qwen →

Other medium models

All medium models →