MI

Codestral-22B-v0.1-abliterated-v3

MistralGGUF by failspyUpdated 2024-06-03
Download on Hugging Face
Parameters22.2B
Context32K
Layers56
Architecturellama
Smallest quant10.8 GB
Quick answer

Codestral-22B-v0.1-abliterated-v3 needs about 18.1 GB of memory at Q5_K (a 15.7 GB file) with an 8K context window, so it fits on a 24 GB GPU. This repository has 6 quants, from 10.8 GB (Q3_K) to 44.5 GB. Longer context needs more: at 32K tokens the KV cache alone adds about 7.5 GB.

22B‑parameter Mistral model focused on coding tasks

Codestral‑22B‑v0.1‑abliterated‑v3 is a 22.2‑billion‑parameter Mistral architecture model. It is trained on a 32768‑token context window and is optimized for code generation. The model is released under the MNPL‑0.1 license.

Good at

  • Code generation
  • Programming assistance
  • Syntax‑aware responses

Limits

  • No vision support
  • License is other (MNPL‑0.1)
  • No mixture of experts

Best for

  • Writing scripts
  • Debugging code
  • Explaining programming concepts

Who it’s for. Developers and programmers looking for a large‑parameter code model. The model is suitable for environments that can handle its memory requirements and comply with the MNPL‑0.1 license.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
Q3_K10.8 GB 13.1 GB
Too big Get file →
Q4_K13.3 GB 15.7 GB
Too big Get file →
Q5_K15.7 GB 18.1 GB
Too big Get file →
Q6_K18.3 GB 20.6 GB
Too big Get file →
Q8_023.6 GB 26.0 GB
Too big Get file →
—44.5 GB 46.9 GB
Too big Get file →

For 8 GB VRAM, use the Q3_K quant (10.76 GB file, ~13.1 GB memory at 8k context). For 12 GB VRAM, Q4_K (13.34 GB file, ~15.7 GB memory). For 16 GB VRAM, Q5_K (15.72 GB file, ~18.1 GB memory).

What GPU do you need for Codestral-22B-v0.1-abliterated-v3?

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 Q4_K (13.3 GB file, about 15.7 GB in use).
  • 24 GB VRAM: the best quant that fits is Q6_K (18.3 GB file, about 20.6 GB in use).
  • 32 GB VRAM: the best quant that fits is Q8_0 (23.6 GB file, about 26.0 GB in use).
  • 48 GB VRAM: the best quant that fits is the main file (44.5 GB file, about 46.9 GB in use).

Which Codestral-22B-v0.1-abliterated-v3 quant should you choose?

Q5_K (15.7 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 Q3_K at 10.8 GB, with the quality trade-offs that come with fewer bits. A reasonable choice when a 4-bit file will not fit, mostly on larger models. Quality loss is noticeable but often acceptable.

New to quant names? Read GGUF quantization explained.

Specifications

Architecture from the GGUF header

Architecturellama
Parameters22.2B
Layers56
Trained context32,768 tokens
KV cache (fp16)224 KB per token

Source

Quantized byfailspy
FamilyMistral
Licenseother
Downloads (HF)158
Last updated2024-06-03

Frequently asked questions about Codestral-22B-v0.1-abliterated-v3

How much VRAM does Codestral-22B-v0.1-abliterated-v3 need?

About 18.1 GB at Q5_K with an 8K context window. That is the 15.7 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 Codestral-22B-v0.1-abliterated-v3 on an 8 GB GPU?

Not at an 8K context. The smallest file needs about 13.1 GB, so you would need a larger GPU or CPU offloading.

What is the file size of Codestral-22B-v0.1-abliterated-v3 GGUF?

It depends on the quant: Q3_K is 10.8 GB, Q4_K is 13.3 GB, Q5_K is 15.7 GB, Q6_K is 18.3 GB, Q8_0 is 23.6 GB, file is 44.5 GB.

What context length does Codestral-22B-v0.1-abliterated-v3 support?

The GGUF header lists a trained context length of 32,768 tokens (32K). The KV cache grows with the context you use: about 1.9 GB at 8K and 7.5 GB at 32K.

Where can I download Codestral-22B-v0.1-abliterated-v3 GGUF?

The files are on Hugging Face in the failspy/Codestral-22B-v0.1-abliterated-v3-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 Codestral-22B-v0.1-abliterated-v3 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 Codestral on 8 GB VRAM?

Use the Q3_K quant, which requires about 13.1 GB of VRAM at 8k context. It is the smallest quant available.

What license does this model use?

The model is released under the MNPL‑0.1 license, as indicated in the model card.

Does this model support image inputs?

No, the model does not have vision capabilities.

Similar models

Browse all →

More from failspy

All by failspy →

More Mistral models

All Mistral →

Other medium models

All medium models →