LL

Llama2 22B GPLATTY

LlamaGGUF by TheBlokeUpdated 2023-09-27
Download on Hugging Face
Parameters21.8B
Context4K
Layers40
Architecturellama
Smallest quant9.1 GB
Quick answer

Llama2 22B GPLATTY needs about 22.4 GB of memory at Q4_K_M (a 13.2 GB file) with an 8K context window, so it fits on a 24 GB GPU. This repository has 12 quants, from 9.1 GB (Q2_K) to 23.2 GB (Q8_0).

Llama2 22B GPLATTY is a 21.8B-parameter Llama model in GGUF format, quantized by TheBloke. Its trained context length is 4K tokens. The repository has 12 quant files, from 9.1 GB (Q2_K) to 23.2 GB (Q8_0).

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
Q2_K9.1 GB 13.9 GB
Too big Get file →
Q3_K_S9.5 GB 14.3 GB
Too big Get file →
Q3_K_M10.6 GB 15.5 GB
Too big Get file →
Q3_K_L11.6 GB 16.5 GB
Too big Get file →
Q4_012.3 GB 17.2 GB
Too big Get file →
Q4_K_S12.4 GB 17.3 GB
Too big Get file →
Q4_K_M13.2 GB 18.0 GB
Too big Get file →
Q5_015.0 GB 19.9 GB
Too big Get file →
Q5_K_S15.0 GB 19.9 GB
Too big Get file →
Q5_K_M15.5 GB 20.3 GB
Too big Get file →
Q6_K17.9 GB 22.8 GB
Too big Get file →
Q8_023.2 GB 28.1 GB
Too big Get file →

What GPU do you need for Llama2 22B GPLATTY?

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_K_M (13.2 GB file, about 22.4 GB in use).
  • 32 GB VRAM: the best quant that fits is Q6_K (17.9 GB file, about 27.1 GB in use).
  • 48 GB VRAM: the best quant that fits is Q8_0 (23.2 GB file, about 32.4 GB in use).

Which Llama2 22B GPLATTY quant should you choose?

Q4_K_M is the usual starting point for Llama2 22B GPLATTY: it balances file size and quality well. Here it is a 13.2 GB file.

The Q8_0 file is 23.2 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 9.1 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
Parameters21.8B
Layers40
Trained context4,096 tokens
KV cache (fp16)1040 KB per token

Source

Quantized byTheBloke
FamilyLlama
Licensellama2
Downloads (HF)345
Last updated2023-09-27

Frequently asked questions about Llama2 22B GPLATTY

How much VRAM does Llama2 22B GPLATTY need?

About 22.4 GB at Q4_K_M with an 8K context window. That is the 13.2 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 Llama2 22B GPLATTY on an 8 GB GPU?

Not at an 8K context. The smallest file needs about 18.3 GB, so you would need a larger GPU or CPU offloading.

What is the file size of Llama2 22B GPLATTY GGUF?

It depends on the quant: Q2_K is 9.1 GB, Q3_K_S is 9.5 GB, Q3_K_M is 10.6 GB, Q3_K_L is 11.6 GB, Q4_0 is 12.3 GB, Q4_K_S is 12.4 GB and 6 more.

What context length does Llama2 22B GPLATTY support?

The GGUF header lists a trained context length of 4,096 tokens (4K). The KV cache grows with the context you use: about 8.7 GB at 8K and 4.4 GB at 4K.

Where can I download Llama2 22B GPLATTY GGUF?

The files are on Hugging Face in the TheBloke/Llama2-22B-GPLATTY-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 Llama2 22B GPLATTY use?

Hugging Face lists the license as llama2. 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 TheBloke

All by TheBloke →

More Llama models

All Llama →

Other medium models

All medium models →