UM

Umbra v2.1 MoE 4x10.7 i1

UmbraGGUF by mradermacherUpdated 2024-12-16
Download on Hugging Face
Parameters36B
Context32K
Layers48
Architecturellama
Smallest quant7.6 GB
Quick answer

Umbra v2.1 MoE 4x10.7 i1 needs about 24.1 GB of memory at Q4_K_M (a 22.0 GB file) with an 8K context window, so it fits on a 32 GB GPU. This repository has 21 quants, from 7.6 GB (IQ1_S) to 29.8 GB (Q6_K). Longer context needs more: at 32K tokens the KV cache alone adds about 6.4 GB.

Umbra v2.1 MoE 4x10.7 i1 – a 36B Mixture of Experts model with 32k context

Umbra v2.1 MoE 4x10.7 i1 is a 36B parameter Mixture of Experts model based on the llama architecture. It supports a 32768 token context window and is released under the Apache‑2.0 license. The model is optimized for general assistance, storytelling, and role‑playing scenarios.

Good at

  • General knowledge queries
  • Storytelling and RP/ERP
  • Long‑form context handling

Limits

  • No vision support
  • Requires large VRAM for higher‑quality quants
  • Limited to 32768 token context

Best for

  • Chatbots needing long context
  • Creative writing assistants
  • Interactive role‑play engines

Who it’s for. Developers and researchers building chat or creative AI applications that need a large context window and Mixture of Experts architecture. The model is suitable for users with sufficient GPU memory to run higher‑quality quantizations.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
IQ1_S7.6 GB 9.7 GB
Too big Get file →
IQ1_M8.4 GB 10.5 GB
Too big Get file →
IQ2_XXS9.7 GB 11.8 GB
Too big Get file →
IQ2_XS10.8 GB 12.9 GB
Too big Get file →
IQ2_S11.0 GB 13.2 GB
Too big Get file →
IQ2_M12.1 GB 14.2 GB
Too big Get file →
Q2_K13.3 GB 15.5 GB
Too big Get file →
IQ3_XXS14.1 GB 16.2 GB
Too big Get file →
IQ3_XS14.8 GB 16.9 GB
Too big Get file →
Q3_K_S15.7 GB 17.8 GB
Too big Get file →
IQ3_S15.7 GB 17.9 GB
Too big Get file →
IQ3_M16.0 GB 18.2 GB
Too big Get file →
Q3_K_M17.4 GB 19.6 GB
Too big Get file →
Q3_K_L18.9 GB 21.0 GB
Too big Get file →
IQ4_XS19.4 GB 21.5 GB
Too big Get file →
Q4_020.6 GB 22.7 GB
Too big Get file →
Q4_K_S20.7 GB 22.8 GB
Too big Get file →
Q4_K_M22.0 GB 24.1 GB
Too big Get file →
Q5_K_S25.0 GB 27.1 GB
Too big Get file →
Q5_K_M25.8 GB 27.9 GB
Too big Get file →
Q6_K29.8 GB 31.9 GB
Too big Get file →

For 8‑16 GB VRAM, use IQ1_S or IQ1_M. For 16‑24 GB VRAM, IQ2_XS or IQ2_S are suitable. For 24‑32 GB VRAM, Q3_K_M or Q4_K_M provide a good balance of size and quality. Memory estimates are approximate.

What GPU do you need for Umbra v2.1 MoE 4x10.7 i1?

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 IQ2_XXS (9.7 GB file, about 11.8 GB in use).
  • 16 GB VRAM: the best quant that fits is Q2_K (13.3 GB file, about 15.5 GB in use).
  • 24 GB VRAM: the best quant that fits is Q4_K_S (20.7 GB file, about 22.8 GB in use).
  • 32 GB VRAM: the best quant that fits is Q6_K (29.8 GB file, about 31.9 GB in use).

Which Umbra v2.1 MoE 4x10.7 i1 quant should you choose?

Q4_K_M is the usual starting point for Umbra v2.1 MoE 4x10.7 i1: it balances file size and quality well. Here it is a 22.0 GB file.

If memory is tight, the smallest file is IQ1_S at 7.6 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
Parameters36B (approximate, mixture of experts)
Layers48
Trained context32,768 tokens
KV cache (fp16)192 KB per token

Source

Quantized bymradermacher
FamilyUmbra
Licenseapache-2.0
Downloads (HF)2,009
Last updated2024-12-16

Frequently asked questions about Umbra v2.1 MoE 4x10.7 i1

How much VRAM does Umbra v2.1 MoE 4x10.7 i1 need?

About 24.1 GB at Q4_K_M with an 8K context window. That is the 22.0 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 Umbra v2.1 MoE 4x10.7 i1 on an 8 GB GPU?

Not at an 8K context. The smallest file needs about 9.7 GB, so you would need a larger GPU or CPU offloading.

What is the file size of Umbra v2.1 MoE 4x10.7 i1 GGUF?

It depends on the quant: IQ1_S is 7.6 GB, IQ1_M is 8.4 GB, IQ2_XXS is 9.7 GB, IQ2_XS is 10.8 GB, IQ2_S is 11.0 GB, IQ2_M is 12.1 GB and 15 more.

What context length does Umbra v2.1 MoE 4x10.7 i1 support?

The GGUF header lists a trained context length of 32,768 tokens (32K). The KV cache grows with the context you use: about 1.6 GB at 8K and 6.4 GB at 32K.

Where can I download Umbra v2.1 MoE 4x10.7 i1 GGUF?

The files are on Hugging Face in the mradermacher/Umbra-v2.1-MoE-4x10.7-i1-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 Umbra v2.1 MoE 4x10.7 i1 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?

Use the IQ1_S quant, which requires about 9.7 GB VRAM at 8k context. It is the smallest quant available.

Does this model support vision?

No, the model has no vision capabilities.

What license applies to this model?

The model is released under the Apache‑2.0 license.

Other GGUFs of Umbra-v2.1-MoE-4x10.7

Similar models

Browse all →

More from mradermacher

All by mradermacher →

More Umbra models

All Umbra →

Other large models

All large models →