MI

llava v1.6 mistral 7b

MistralGGUF by mradermacherUpdated 2024-05-06
Download on Hugging Face
Parameters7.24B
Context32K
Layers32
Architecturellama
Smallest quant2.9 GB
Quick answer

llava v1.6 mistral 7b needs about 6.1 GB of memory at Q4_K_M (a 4.5 GB file) with an 8K context window, so it fits on a 8 GB GPU. This repository has 16 quants, from 2.9 GB (Q2_K) to 7.8 GB (Q8_0). Longer context needs more: at 32K tokens the KV cache alone adds about 4.3 GB.

LLaVA v1.6 Mistral 7B – multimodal instruction-following model

LLaVA v1.6 Mistral 7B is a multimodal transformer based on Mistral‑7B‑Instruct‑v0.2. It was trained on 558K image‑text pairs, 158K GPT‑generated instruction data, 500K academic VQA, 50K GPT‑4V, and 40K ShareGPT. The model supports vision and text generation up to 32768 tokens.

Good at

  • Vision‑text instruction following
  • Image captioning
  • Multimodal Q&A

Limits

  • No weighted/imatrix quants available
  • Requires 8k context for memory estimates
  • Apache‑2.0 license

Best for

  • Research on multimodal LLMs
  • Chatbot prototyping
  • Vision‑based instruction tasks

Who it’s for. Researchers and hobbyists in computer vision, NLP, machine learning, and AI who need a multimodal LLM for experimentation and prototyping. The model is open‑source under Apache‑2.0.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
Q2_K2.9 GB 4.4 GB
Fits Get file →
IQ3_XS3.2 GB 4.7 GB
Fits Get file →
Q3_K_S3.3 GB 4.9 GB
Fits Get file →
IQ3_S3.3 GB 4.9 GB
Fits Get file →
IQ3_M3.4 GB 5.0 GB
Fits Get file →
Q3_K_M3.7 GB 5.2 GB
Fits Get file →
Q3_K_L4.0 GB 5.6 GB
Fits Get file →
IQ4_XS4.1 GB 5.7 GB
Fits Get file →
Q4_04.3 GB 5.8 GB
Fits Get file →
Q4_K_S4.3 GB 5.9 GB
Fits Get file →
IQ4_NL4.3 GB 5.9 GB
Fits Get file →
Q4_K_M4.5 GB 6.1 GB
Fits Get file →
Q5_K_S5.2 GB 6.7 GB
Fits Get file →
Q5_K_M5.3 GB 6.9 GB
Fits Get file →
Q6_K6.1 GB 7.7 GB
Tight Get file →
Q8_07.8 GB 9.4 GB
Too big Get file →

For 8 GB VRAM, use Q4_0 (file size 4.26 GB, estimated 5.8 GB memory). For 12 GB VRAM, Q5_K_M (5.29 GB, 6.9 GB memory) or Q6_K (6.1 GB, 7.7 GB memory). Larger VRAM allows Q8_0 (7.82 GB, 9.4 GB memory).

What GPU do you need for llava v1.6 mistral 7b?

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 IQ4_NL (4.3 GB file, about 5.9 GB in use).
  • 8 GB VRAM: the best quant that fits is Q6_K (6.1 GB file, about 7.7 GB in use).
  • 12 GB VRAM: the best quant that fits is Q8_0 (7.8 GB file, about 9.4 GB in use).

Which llava v1.6 mistral 7b quant should you choose?

Q4_K_M is the usual starting point for llava v1.6 mistral 7b: it balances file size and quality well. Here it is a 4.5 GB file.

The Q8_0 file is 7.8 GB, about 1.7 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 2.9 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
Parameters7.24B
Layers32
Trained context32,768 tokens
KV cache (fp16)128 KB per token

Source

Quantized bymradermacher
FamilyMistral
Licenseapache-2.0
Downloads (HF)3,368
Last updated2024-05-06

Frequently asked questions about llava v1.6 mistral 7b

How much VRAM does llava v1.6 mistral 7b need?

About 6.1 GB at Q4_K_M with an 8K context window. That is the 4.5 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 llava v1.6 mistral 7b on an 8 GB GPU?

Yes, with Q6_K: the 6.1 GB file needs about 7.7 GB at 8K context. Larger quants in this repository do not fit in 8 GB.

What is the file size of llava v1.6 mistral 7b GGUF?

It depends on the quant: Q2_K is 2.9 GB, IQ3_XS is 3.2 GB, Q3_K_S is 3.3 GB, IQ3_S is 3.3 GB, IQ3_M is 3.4 GB, Q3_K_M is 3.7 GB and 10 more.

What context length does llava v1.6 mistral 7b 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 llava v1.6 mistral 7b GGUF?

The files are on Hugging Face in the mradermacher/llava-v1.6-mistral-7b-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 llava v1.6 mistral 7b 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 LLaVA v1.6 Mistral 7B on 8 GB VRAM?

Yes, using the Q4_0 quant which estimates 5.8 GB memory at 8k context.

Does the model support weighted or imatrix quants?

No, only static quants are provided; weighted/imatrix quants are not available.

What license does the model use?

The model is licensed under Apache‑2.0.

Similar models

Browse all →

More from mradermacher

All by mradermacher →

More Mistral models

All Mistral →

Other small models

All small models →