MI

chargoddard_mixtralnt-4x7b-test

MistralGGUF by TheBlokeUpdated 2023-12-13
Download on Hugging Face
Parameters24.2B
Context32K
Layers32
Architecturellama
Smallest quant8.1 GB
Quick answer

chargoddard_mixtralnt-4x7b-test needs about 15.2 GB of memory at Q4_K_M (a 13.6 GB file) with an 8K context window, so it fits on a 16 GB GPU. This repository has 8 quants, from 8.1 GB (Q2_K) to 25.7 GB (Q8_0). Longer context needs more: at 32K tokens the KV cache alone adds about 4.3 GB.

24.2B parameter Mixtral‑nt 4x7B test model in GGUF format

This model is a 24.2 B parameter mixture of experts architecture with a 32 768 token context window. It is released under a cc‑by‑nc‑4.0 license and is experimental, requiring a specific llama.cpp pull request to run.

Good at

  • Large context generation
  • Mixture of experts inference
  • GPU accelerated inference

Limits

  • Requires llama.cpp PR 4406
  • Not compatible with main or downstream clients
  • License cc‑by‑nc‑4.0 restricts commercial use

Best for

  • Research on MoE architectures
  • GPU accelerated inference testing
  • Educational experiments with GGUF

Who it’s for. Developers and researchers interested in experimental MoE models and GGUF format. Suitable for those who can build llama.cpp from the specified PR.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
Q2_K8.1 GB 9.6 GB
Too big Get file →
Q3_K_M10.5 GB 12.1 GB
Too big Get file →
Q4_013.6 GB 15.2 GB
Too big Get file →
Q4_K_M13.6 GB 15.2 GB
Too big Get file →
Q5_016.6 GB 18.2 GB
Too big Get file →
Q5_K_M16.6 GB 18.2 GB
Too big Get file →
Q6_K19.8 GB 21.4 GB
Too big Get file →
Q8_025.7 GB 27.2 GB
Too big Get file →

For 8 GB VRAM the smallest quant (Q2_K) still needs 9.6 GB memory, so it is not feasible. A 16 GB VRAM setup can run Q4_0 or Q4_K_M (13.62 GB file, 15.2 GB memory). With 32 GB VRAM, Q6_K or Q8_0 are viable options.

What GPU do you need for chargoddard_mixtralnt-4x7b-test?

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: the best quant that fits is Q2_K (8.1 GB file, about 9.6 GB in use).
  • 16 GB VRAM: the best quant that fits is Q4_K_M (13.6 GB file, about 15.2 GB in use).
  • 24 GB VRAM: the best quant that fits is Q6_K (19.8 GB file, about 21.4 GB in use).
  • 32 GB VRAM: the best quant that fits is Q8_0 (25.7 GB file, about 27.2 GB in use).

Which chargoddard_mixtralnt-4x7b-test quant should you choose?

Q4_K_M is the usual starting point for chargoddard_mixtralnt-4x7b-test: it balances file size and quality well. Here it is a 13.6 GB file.

The Q8_0 file is 25.7 GB, about 1.9 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 8.1 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
Parameters24.2B (approximate, mixture of experts)
Layers32
Trained context32,768 tokens
KV cache (fp16)128 KB per token

Source

Quantized byTheBloke
FamilyMistral
Downloads (HF)775
Last updated2023-12-13

Frequently asked questions about chargoddard_mixtralnt-4x7b-test

How much VRAM does chargoddard_mixtralnt-4x7b-test need?

About 15.2 GB at Q4_K_M with an 8K context window. That is the 13.6 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 chargoddard_mixtralnt-4x7b-test on an 8 GB GPU?

Not at an 8K context. The smallest file needs about 9.6 GB, so you would need a larger GPU or CPU offloading.

What is the file size of chargoddard_mixtralnt-4x7b-test GGUF?

It depends on the quant: Q2_K is 8.1 GB, Q3_K_M is 10.5 GB, Q4_0 is 13.6 GB, Q4_K_M is 13.6 GB, Q5_0 is 16.6 GB, Q5_K_M is 16.6 GB and 2 more.

What context length does chargoddard_mixtralnt-4x7b-test support?

The GGUF header lists a trained context length of 32,768 tokens (32K). The KV cache grows with the context you use: about 1.1 GB at 8K and 4.3 GB at 32K.

Where can I download chargoddard_mixtralnt-4x7b-test GGUF?

The files are on Hugging Face in the TheBloke/mixtralnt-4x7b-test-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 chargoddard_mixtralnt-4x7b-test use?

Hugging Face lists the license as cc-by-nc-4.0. 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 this model on 8 GB VRAM?

No, the smallest quant requires 9.6 GB memory at 8k context.

Does it support vision?

No vision support is indicated.

Is it compatible with LM Studio?

No, it requires a custom llama.cpp build.

Other GGUFs of mixtralnt-4x7b-test

Similar models

Browse all →

More from TheBloke

All by TheBloke →

More Mistral models

All Mistral →

Other medium models

All medium models →