Yuan2 M32 hf IMat needs about 23.2 GB of memory at Q3_K_L (a 21.1 GB file) with an 8K context window, so it fits on a 24 GB GPU. This repository has 17 quants, from 8.7 GB (IQ1_S) to 80.1 GB.
Yuan2 M32 hf IMat – 40B parameter MoE model quantized for llama.cpp
Yuan2 M32 hf IMat is a 40.2B parameter Mixture-of-Experts model with 32 experts, 2 active. It is quantized by legraphista for use with llama.cpp. The model supports a 4096 token context and is released under Apache‑2.0.
Good at
- Large‑scale language tasks
- Efficient inference with MoE
- High accuracy on coding and math benchmarks
Limits
- No vision support
- Requires 8K context for memory estimates
- License is Apache‑2.0
Best for
- Research prototyping
- Low‑latency inference on GPUs
- Fine‑tuning with llama.cpp
Who it’s for. Engineers and researchers who need a large MoE model that can run on local hardware via llama.cpp. The model is suitable for developers building language‑model applications that require efficient inference and can handle 4096 token contexts.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| IQ1_S | 8.7 GB | 10.0 GB | Too big | Get file → | |
| IQ2_XXS | 11.0 GB | 12.3 GB | Too big | Get file → | |
| IQ2_XS | 12.2 GB | 13.5 GB | Too big | Get file → | |
| Q2_K_S | 14.1 GB | 15.4 GB | Too big | Get file → | |
| Q2_K | 15.0 GB | 16.3 GB | Too big | Get file → | |
| IQ3_XXS | 15.9 GB | 17.2 GB | Too big | Get file → | |
| Q3_K_S | 17.7 GB | 19.0 GB | Too big | Get file → | |
| Q3_K | 19.5 GB | 20.8 GB | Too big | Get file → | |
| Q3_K_L | 21.1 GB | 22.4 GB | Too big | Get file → | |
| IQ4_NL | 23.0 GB | 24.3 GB | Too big | Get file → | |
| Q4_K_S | 23.2 GB | 24.5 GB | Too big | Get file → | |
| Q4_K | 24.7 GB | 26.0 GB | Too big | Get file → | |
| Q5_K_S | 28.0 GB | 29.3 GB | Too big | Get file → | |
| Q5_K | 28.8 GB | 30.1 GB | Too big | Get file → | |
| Q6_K | 33.2 GB | 34.5 GB | Too big | Get file → | |
| Q8_0 | 42.9 GB | 44.2 GB | Too big | Get file → | |
| — | 80.1 GB (4 parts) | 81.4 GB | Too big | Get file → |
For 8 GB VRAM, use IQ1_S (8.7 GB file, 10.8 GB memory at 8K context). For 12 GB VRAM, IQ2_XS (12.21 GB file, 14.3 GB memory). For 16 GB VRAM, Q4_K_S (23.19 GB file, 25.3 GB memory). Higher quants give better quality but need more memory.
What GPU do you need for Yuan2 M32 hf IMat?
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: the best quant that fits is IQ1_S (8.7 GB file, about 10.8 GB in use).
- 16 GB VRAM: the best quant that fits is IQ2_XS (12.2 GB file, about 14.3 GB in use).
- 24 GB VRAM: the best quant that fits is Q3_K_L (21.1 GB file, about 23.2 GB in use).
- 32 GB VRAM: the best quant that fits is Q5_K (28.8 GB file, about 30.9 GB in use).
- 48 GB VRAM: the best quant that fits is Q8_0 (42.9 GB file, about 45.0 GB in use).
Which Yuan2 M32 hf IMat quant should you choose?
Q3_K_L (21.1 GB) is the middle option in this repository. A reasonable choice when a 4-bit file will not fit, mostly on larger models. Quality loss is noticeable but often acceptable.
If memory is tight, the smallest file is IQ1_S at 8.7 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 Yuan2 M32 hf IMat
How much VRAM does Yuan2 M32 hf IMat need?
About 23.2 GB at Q3_K_L with an 8K context window. That is the 21.1 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 Yuan2 M32 hf IMat on an 8 GB GPU?
Not at an 8K context. The smallest file needs about 10.8 GB, so you would need a larger GPU or CPU offloading.
What is the file size of Yuan2 M32 hf IMat GGUF?
It depends on the quant: IQ1_S is 8.7 GB, IQ2_XXS is 11.0 GB, IQ2_XS is 12.2 GB, Q2_K_S is 14.1 GB, Q2_K is 15.0 GB, IQ3_XXS is 15.9 GB and 11 more.
What context length does Yuan2 M32 hf IMat support?
The GGUF header lists a trained context length of 4,096 tokens (4K). The KV cache grows with the context you use: about 1.6 GB at 8K and 0.8 GB at 4K.
Where can I download Yuan2 M32 hf IMat GGUF?
The files are on Hugging Face in the legraphista/Yuan2-M32-hf-IMat-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 Yuan2 M32 hf IMat 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 this model on 8 GB VRAM?
Yes, using IQ1_S quantization which requires about 10.8 GB memory at 8K context. The file size is 8.7 GB.
Does this model support vision tasks?
No, the model has no vision capabilities.
What license does this model use?
The model is released under the Apache‑2.0 license.