llm jp 3.1 8x13b instruct4 needs about 52.6 GB of memory at Q4_K_M (a 45.4 GB file) with an 8K context window, which is more than a 48 GB GPU holds. This repository has 17 quants, from 20.4 GB (IQ2_XXS) to 45.4 GB (Q4_K_M).
llm jp 3.1 8x13b instruct4 is a 73.2B-parameter Llm model in GGUF format, quantized by mmnga. Its trained context length is 4K tokens. The repository has 17 quant files, from 20.4 GB (IQ2_XXS) to 45.4 GB (Q4_K_M).
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| IQ2_XXS | 20.4 GB | 24.3 GB | Too big | Get file → | |
| IQ2_XS | 22.5 GB | 26.3 GB | Too big | Get file → | |
| IQ2_S | 22.8 GB | 26.7 GB | Too big | Get file → | |
| IQ2_M | 24.9 GB | 28.7 GB | Too big | Get file → | |
| Q2_K | 28.5 GB | 32.3 GB | Too big | Get file → | |
| IQ3_XXS | 29.9 GB | 33.7 GB | Too big | Get file → | |
| IQ3_XS | 31.6 GB | 35.5 GB | Too big | Get file → | |
| Q3_K_S | 33.2 GB | 37.1 GB | Too big | Get file → | |
| IQ3_S | 33.2 GB | 37.1 GB | Too big | Get file → | |
| IQ3_M | 34.7 GB | 38.6 GB | Too big | Get file → | |
| Q3_K_M | 36.4 GB | 40.2 GB | Too big | Get file → | |
| Q3_K_L | 38.8 GB | 42.7 GB | Too big | Get file → | |
| IQ4_XS | 40.3 GB | 44.2 GB | Too big | Get file → | |
| Q4_0 | 42.3 GB | 46.2 GB | Too big | Get file → | |
| IQ4_NL | 42.5 GB | 46.3 GB | Too big | Get file → | |
| Q4_K_S | 42.8 GB | 46.7 GB | Too big | Get file → | |
| Q4_K_M | 45.4 GB | 49.3 GB | Too big | Get file → |
What GPU do you need for llm jp 3.1 8x13b instruct4?
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: no quant fits at an 8K context.
- 24 GB VRAM: no quant fits at an 8K context.
- 32 GB VRAM: the best quant that fits is IQ2_S (22.8 GB file, about 30.0 GB in use).
- 48 GB VRAM: the best quant that fits is IQ4_XS (40.3 GB file, about 47.5 GB in use).
Which llm jp 3.1 8x13b instruct4 quant should you choose?
Q4_K_M is the usual starting point for llm jp 3.1 8x13b instruct4: it balances file size and quality well. Here it is a 45.4 GB file.
If memory is tight, the smallest file is IQ2_XXS at 20.4 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
Source
Frequently asked questions about llm jp 3.1 8x13b instruct4
How much VRAM does llm jp 3.1 8x13b instruct4 need?
About 52.6 GB at Q4_K_M with an 8K context window. That is the 45.4 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 llm jp 3.1 8x13b instruct4 on an 8 GB GPU?
Not at an 8K context. The smallest file needs about 27.6 GB, so you would need a larger GPU or CPU offloading.
What is the file size of llm jp 3.1 8x13b instruct4 GGUF?
It depends on the quant: IQ2_XXS is 20.4 GB, IQ2_XS is 22.5 GB, IQ2_S is 22.8 GB, IQ2_M is 24.9 GB, Q2_K is 28.5 GB, IQ3_XXS is 29.9 GB and 11 more.
What context length does llm jp 3.1 8x13b instruct4 support?
The GGUF header lists a trained context length of 4,096 tokens (4K). The KV cache grows with the context you use: about 6.7 GB at 8K and 3.4 GB at 4K.
Where can I download llm jp 3.1 8x13b instruct4 GGUF?
The files are on Hugging Face in the mmnga/llm-jp-3.1-8x13b-instruct4-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 llm jp 3.1 8x13b instruct4 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.