MI

Mistral Crab DPO

MistralGGUF by bartowskiUpdated 2024-11-03
Download on Hugging Face
Parameters7.25B
Context32K
Layers32
Architecturellama
Smallest quant2.5 GB
Quick answer

Mistral Crab DPO needs about 5.9 GB of memory at Q4_K_M (a 4.4 GB file) with an 8K context window, so it fits on a 6 GB GPU. This repository has 24 quants, from 2.5 GB (IQ2_M) to 14.5 GB (F16). Longer context needs more: at 32K tokens the KV cache alone adds about 4.3 GB.

Mistral Crab DPO – 7.25B Llama‑style model fine‑tuned for complex instruction following

Mistral Crab DPO is a 7.25B parameter Llama‑based model trained on a constraint back‑translation dataset. It supports 32,768 token context length and is released under Apache‑2.0. The model is available in multiple GGUF quantizations for llama.cpp.

Good at

  • Complex instruction compliance
  • Constraint‑aware responses
  • Large context handling

Limits

  • No vision support
  • Requires 8,000 token context for memory estimates
  • No explicit language support list

Best for

  • Chatbots needing precise instruction adherence
  • Research on constraint‑based generation
  • Fine‑tuning for specialized instruction datasets

Who it’s for. Developers and researchers building applications that demand strict adherence to complex instructions. The model is suitable for users who can run llama.cpp or compatible inference engines and have access to GPUs or CPUs with sufficient memory.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
IQ2_M2.5 GB 4.1 GB
Fits Get file →
Q2_K2.7 GB 4.3 GB
Fits Get file →
Q2_K_L2.9 GB 4.4 GB
Fits Get file →
IQ3_XS3.0 GB 4.6 GB
Fits Get file →
Q3_K_S3.2 GB 4.7 GB
Fits Get file →
IQ3_M3.3 GB 4.9 GB
Fits Get file →
Q3_K_M3.5 GB 5.1 GB
Fits Get file →
Q3_K_L3.8 GB 5.4 GB
Fits Get file →
IQ4_XS3.9 GB 5.5 GB
Fits Get file →
Q3_K_XL3.9 GB 5.5 GB
Fits Get file →
Q4_04.1 GB 5.7 GB
Fits Get file →
Q4_04.1 GB 5.7 GB
Fits Get file →
Q4_04.1 GB 5.7 GB
Fits Get file →
Q4_04.1 GB 5.7 GB
Fits Get file →
Q4_K_S4.1 GB 5.7 GB
Fits Get file →
Q4_K_M4.4 GB 5.9 GB
Fits Get file →
Q4_K_L4.5 GB 6.0 GB
Fits Get file →
Q5_K_S5.0 GB 6.6 GB
Fits Get file →
Q5_K_M5.1 GB 6.7 GB
Fits Get file →
Q5_K_L5.2 GB 6.8 GB
Fits Get file →
Q6_K5.9 GB 7.5 GB
Tight Get file →
Q6_K_L6.0 GB 7.6 GB
Tight Get file →
Q8_07.7 GB 9.3 GB
Too big Get file →
F1614.5 GB 16.1 GB
Too big Get file →

For 8 GB VRAM, use Q4_0 (file size 4.11 GB, ~5.7 GB memory at 8k context). For 12 GB VRAM, Q4_K_M (4.37 GB, ~5.9 GB) or Q5_K_S (5 GB, ~6.6 GB). For 16 GB VRAM, Q6_K (5.95 GB, ~7.5 GB) or Q8_0 (7.7 GB, ~9.3 GB). Estimates are approximate.

What GPU do you need for Mistral Crab DPO?

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: the best quant that fits is Q4_K_L (4.5 GB file, about 6.0 GB in use).
  • 8 GB VRAM: the best quant that fits is Q6_K_L (6.0 GB file, about 7.6 GB in use).
  • 12 GB VRAM: the best quant that fits is Q8_0 (7.7 GB file, about 9.3 GB in use).
  • 16 GB VRAM: the best quant that fits is Q8_0 (7.7 GB file, about 9.3 GB in use).
  • 24 GB VRAM: the best quant that fits is F16 (14.5 GB file, about 16.1 GB in use).

Which Mistral Crab DPO quant should you choose?

Q4_K_M is the usual starting point for Mistral Crab DPO: it balances file size and quality well. Here it is a 4.4 GB file.

The Q8_0 file is 7.7 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 IQ2_M at 2.5 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
Parameters7.25B
Layers32
Trained context32,768 tokens
KV cache (fp16)128 KB per token

Source

Quantized bybartowski
FamilyMistral
Licenseapache-2.0
Downloads (HF)647
Last updated2024-11-03

Frequently asked questions about Mistral Crab DPO

How much VRAM does Mistral Crab DPO need?

About 5.9 GB at Q4_K_M with an 8K context window. That is the 4.4 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 Mistral Crab DPO on an 8 GB GPU?

Yes, with Q6_K_L: the 6.0 GB file needs about 7.6 GB at 8K context. Larger quants in this repository do not fit in 8 GB.

What is the file size of Mistral Crab DPO GGUF?

It depends on the quant: IQ2_M is 2.5 GB, Q2_K is 2.7 GB, Q2_K_L is 2.9 GB, IQ3_XS is 3.0 GB, Q3_K_S is 3.2 GB, IQ3_M is 3.3 GB and 18 more.

What context length does Mistral Crab DPO 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 Mistral Crab DPO GGUF?

The files are on Hugging Face in the bartowski/Mistral-Crab-DPO-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 Mistral Crab DPO 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 Mistral Crab DPO on 8 GB VRAM?

Yes, a Q4_0 quantization (4.11 GB file) will fit and use about 5.7 GB of VRAM at 8k context.

Does the model support non‑English languages?

The model card lists only English; no other languages are guaranteed.

Is there a vision capability?

No, the model does not include vision support.

Other GGUFs of Mistral-Crab-DPO

Similar models

Browse all →

More from bartowski

All by bartowski →

More Mistral models

All Mistral →

Other small models

All small models →