MI

Miqu 1 120b

MiquGGUF by RichardErkhovUpdated 2024-10-26
Download on Hugging Face
Parameters120B
Context32K
Layers140
Architecturellama
Smallest quant44.2 GB
Quick answer

Miqu 1 120b needs about 77.3 GB of memory at Q4_K_M (a 72.1 GB file) with an 8K context window, which is more than a 48 GB GPU holds. This repository has 22 quants, from 44.2 GB (Q2_K) to 127.8 GB (Q8_0).

Miqu 1 120b is a 120B-parameter Miqu model in GGUF format, quantized by RichardErkhov. Its trained context length is 32K tokens. The repository has 22 quant files, from 44.2 GB (Q2_K) to 127.8 GB (Q8_0).

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
Q2_K44.2 GB (2 parts) 49.4 GB
Too big Get file →
IQ3_XS49.2 GB (2 parts) 54.4 GB
Too big Get file →
Q3_K_S51.8 GB (2 parts) 57.0 GB
Too big Get file →
IQ3_S52.0 GB (2 parts) 57.2 GB
Too big Get file →
IQ3_M53.7 GB (2 parts) 58.9 GB
Too big Get file →
Q3_K57.8 GB (2 parts) 63.0 GB
Too big Get file →
Q3_K_M57.8 GB (2 parts) 63.0 GB
Too big Get file →
Q3_K_L63.0 GB (2 parts) 68.2 GB
Too big Get file →
IQ4_XS64.8 GB (2 parts) 70.0 GB
Too big Get file →
Q4_067.8 GB (2 parts) 73.0 GB
Too big Get file →
Q4_K_S68.3 GB (2 parts) 73.5 GB
Too big Get file →
IQ4_NL68.4 GB (2 parts) 73.6 GB
Too big Get file →
Q4_K72.1 GB (2 parts) 77.3 GB
Too big Get file →
Q4_K_M72.1 GB (2 parts) 77.3 GB
Too big Get file →
Q4_175.3 GB (2 parts) 80.5 GB
Too big Get file →
Q5_082.8 GB (3 parts) 88.0 GB
Too big Get file →
Q5_K_S82.8 GB (3 parts) 88.0 GB
Too big Get file →
Q5_K85.0 GB (3 parts) 90.2 GB
Too big Get file →
Q5_K_M85.0 GB (3 parts) 90.2 GB
Too big Get file →
Q5_190.3 GB (3 parts) 95.5 GB
Too big Get file →
Q6_K98.7 GB (3 parts) 103.9 GB
Too big Get file →
Q8_0127.8 GB (4 parts) 133.0 GB
Too big Get file →

What GPU do you need for Miqu 1 120b?

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: no quant fits at an 8K context.
  • 48 GB VRAM: no quant fits at an 8K context.

Which Miqu 1 120b quant should you choose?

Q4_K_M is the usual starting point for Miqu 1 120b: it balances file size and quality well. Here it is a 72.1 GB file.

The Q8_0 file is 127.8 GB, about 1.8 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 44.2 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
Parameters120B
Layers140
Trained context32,764 tokens
KV cache (fp16)560 KB per token

Source

Quantized byRichardErkhov
FamilyMiqu
Licenseother
Downloads (HF)6
Last updated2024-10-26

Frequently asked questions about Miqu 1 120b

How much VRAM does Miqu 1 120b need?

About 77.3 GB at Q4_K_M with an 8K context window. That is the 72.1 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 Miqu 1 120b on an 8 GB GPU?

Not at an 8K context. The smallest file needs about 49.4 GB, so you would need a larger GPU or CPU offloading.

What is the file size of Miqu 1 120b GGUF?

It depends on the quant: Q2_K is 44.2 GB, IQ3_XS is 49.2 GB, Q3_K_S is 51.8 GB, IQ3_S is 52.0 GB, IQ3_M is 53.7 GB, Q3_K is 57.8 GB and 16 more.

What context length does Miqu 1 120b support?

The GGUF header lists a trained context length of 32,764 tokens (32K). The KV cache grows with the context you use: about 4.7 GB at 8K and 18.8 GB at 32K.

Where can I download Miqu 1 120b GGUF?

The files are on Hugging Face in the RichardErkhov/wolfram_-_miqu-1-120b-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 Miqu 1 120b use?

Hugging Face lists the license as other. Licenses can differ between the quantized files and the original model, so check the repository and the base model before commercial use.

Similar models

Browse all →

More from RichardErkhov

All by RichardErkhov →

More Miqu models

All Miqu →

Other large models

All large models →