Mark-Experimental-V3-8B needs about 5.6 GB of memory at IQ4_NL (a 4.6 GB file) with an 8K context window, so it fits on a 6 GB GPU. Longer context needs more: at 32K tokens the KV cache alone adds about 2.2 GB.
Mark-Experimental-V3-8B: 8B sparse model with 32 layers and 32 routed experts
Mark-Experimental-V3-8B is an 8B-class sparse model with 32 layers, 32 routed experts, 2 shared experts, and full H-MLA attention. It contains 8,168,001,024 total parameters, 2,002,374,144 active per token. Quantized in IQ4_NL, the model fits within an 8 GB GPU.
Good at
- Sparse inference with 32 routed experts
- Full H-MLA attention for efficient KV cache
- Large context support up to 32,768 tokens
Limits
- Requires at least 8 GB VRAM
- License is agpl-3.0, restricts commercial use
- No vision or multimodal capabilities
Best for
- Low‑memory inference on consumer GPUs
- Research on sparse transformer architectures
- Generating long‑form text up to 32 768 tokens
Who it’s for. Developers and researchers needing a lightweight sparse transformer that can run on standard 8 GB GPUs. The model is suitable for applications requiring large context windows without high VRAM demands.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| IQ4_NL | 4.6 GB | 5.6 GB | Fits | Get file → |
The only available quant is IQ4_NL, 4.6 GB in size, estimating 5.6 GB memory usage at 8 k context. This fits comfortably on an 8 GB GPU, leaving space for KV cache. For tighter budgets, no other quant is provided.
What GPU do you need for Mark-Experimental-V3-8B?
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 (4.6 GB file, about 5.6 GB in use).
Which Mark-Experimental-V3-8B quant should you choose?
IQ4_NL (4.6 GB) is the middle option in this repository. The usual sweet spot between file size and quality, and the quant most people start with.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about Mark-Experimental-V3-8B
How much VRAM does Mark-Experimental-V3-8B need?
About 5.6 GB at IQ4_NL with an 8K context window. That is the 4.6 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 Mark-Experimental-V3-8B on an 8 GB GPU?
Yes, with IQ4_NL: the 4.6 GB file needs about 5.6 GB at 8K context. Larger quants in this repository do not fit in 8 GB.
What is the file size of Mark-Experimental-V3-8B GGUF?
It depends on the quant: IQ4_NL is 4.6 GB.
What context length does Mark-Experimental-V3-8B support?
The GGUF header lists a trained context length of 32,768 tokens (32K). The KV cache grows with the context you use: about 0.6 GB at 8K and 2.2 GB at 32K.
Where can I download Mark-Experimental-V3-8B GGUF?
The files are on Hugging Face in the miceworks/Mark-Experimental-V3-8B repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.
What license does Mark-Experimental-V3-8B use?
Hugging Face lists the license as agpl-3.0. Licenses can differ between the quantized files and the original model, so check the repository and the base model before commercial use.
What license does this model use?
The model is released under agpl-3.0, which requires source disclosure for derivative works.
Does the model support vision or multimodal inputs?
No, the model only supports text inputs and has no vision capabilities.