YI

Yi 34Bx2 MoE 60B

YiGGUF by RichardErkhovUpdated 2024-10-11
Download on Hugging Face
Parameters60.8B
Context195K
Layers60
Architecturellama
Smallest quant2.7 GB
Quick answer

Yi 34Bx2 MoE 60B needs about 39.2 GB of memory at Q4_K_M (a 36.7 GB file) with an 8K context window, so it fits on a 48 GB GPU. This repository has 22 quants, from 2.7 GB (IQ3_S) to 64.6 GB (Q8_0). Longer context needs more: at 32K tokens the KV cache alone adds about 8.1 GB.

Yi 34Bx2 MoE 60B is a 60.8B-parameter Yi model in GGUF format, quantized by RichardErkhov. Its trained context length is 195K tokens. The repository has 22 quant files, from 2.7 GB (IQ3_S) to 64.6 GB (Q8_0).

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
IQ3_S2.7 GB 5.2 GB
Fits Get file →
Q4_K_S13.6 GB 16.1 GB
Too big Get file →
Q2_K22.4 GB 24.9 GB
Too big Get file →
IQ3_XS25.0 GB 27.5 GB
Too big Get file →
Q3_K_S26.3 GB 28.8 GB
Too big Get file →
IQ4_NL26.6 GB 29.1 GB
Too big Get file →
IQ3_M27.1 GB 29.6 GB
Too big Get file →
Q3_K29.2 GB 31.7 GB
Too big Get file →
Q3_K_M29.2 GB 31.7 GB
Too big Get file →
Q3_K_L31.8 GB 34.3 GB
Too big Get file →
IQ4_XS32.8 GB 35.3 GB
Too big Get file →
Q4_034.3 GB 36.8 GB
Too big Get file →
Q4_K36.7 GB 39.2 GB
Too big Get file →
Q4_K_M36.7 GB 39.2 GB
Too big Get file →
Q4_138.1 GB 40.6 GB
Too big Get file →
Q5_041.9 GB (2 parts) 44.4 GB
Too big Get file →
Q5_K_S41.9 GB (2 parts) 44.4 GB
Too big Get file →
Q5_K43.1 GB (2 parts) 45.6 GB
Too big Get file →
Q5_K_M43.1 GB (2 parts) 45.6 GB
Too big Get file →
Q5_145.6 GB (2 parts) 48.2 GB
Too big Get file →
Q6_K49.9 GB (2 parts) 52.4 GB
Too big Get file →
Q8_064.6 GB (2 parts) 67.1 GB
Too big Get file →

What GPU do you need for Yi 34Bx2 MoE 60B?

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 IQ3_S (2.7 GB file, about 5.2 GB in use).
  • 8 GB VRAM: the best quant that fits is IQ3_S (2.7 GB file, about 5.2 GB in use).
  • 12 GB VRAM: the best quant that fits is IQ3_S (2.7 GB file, about 5.2 GB in use).
  • 16 GB VRAM: the best quant that fits is IQ3_S (2.7 GB file, about 5.2 GB in use).
  • 24 GB VRAM: the best quant that fits is Q4_K_S (13.6 GB file, about 16.1 GB in use).
  • 32 GB VRAM: the best quant that fits is Q3_K (29.2 GB file, about 31.7 GB in use).
  • 48 GB VRAM: the best quant that fits is Q5_K (43.1 GB file, about 45.6 GB in use).

Which Yi 34Bx2 MoE 60B quant should you choose?

Q4_K_M is the usual starting point for Yi 34Bx2 MoE 60B: it balances file size and quality well. Here it is a 36.7 GB file.

The Q8_0 file is 64.6 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 IQ3_S at 2.7 GB, with the quality trade-offs that come with fewer bits. A reasonable choice when a 4-bit file will not fit, mostly on larger models. Quality loss is noticeable but often acceptable.

New to quant names? Read GGUF quantization explained.

Specifications

Architecture from the GGUF header

Architecturellama
Parameters60.8B (approximate, mixture of experts)
Layers60
Trained context200,000 tokens
KV cache (fp16)240 KB per token

Source

Quantized byRichardErkhov
FamilyYi
Licenseapache-2.0
Downloads (HF)189
Last updated2024-10-11

Frequently asked questions about Yi 34Bx2 MoE 60B

How much VRAM does Yi 34Bx2 MoE 60B need?

About 39.2 GB at Q4_K_M with an 8K context window. That is the 36.7 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 Yi 34Bx2 MoE 60B on an 8 GB GPU?

Yes, with IQ3_S: the 2.7 GB file needs about 5.2 GB at 8K context. Larger quants in this repository do not fit in 8 GB.

What is the file size of Yi 34Bx2 MoE 60B GGUF?

It depends on the quant: IQ3_S is 2.7 GB, Q4_K_S is 13.6 GB, Q2_K is 22.4 GB, IQ3_XS is 25.0 GB, Q3_K_S is 26.3 GB, IQ4_NL is 26.6 GB and 16 more.

What context length does Yi 34Bx2 MoE 60B support?

The GGUF header lists a trained context length of 200,000 tokens (195K). The KV cache grows with the context you use: about 2.0 GB at 8K and 8.1 GB at 32K.

Where can I download Yi 34Bx2 MoE 60B GGUF?

The files are on Hugging Face in the RichardErkhov/cloudyu_-_Yi-34Bx2-MoE-60B-DPO-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 Yi 34Bx2 MoE 60B 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.

Similar models

Browse all →

More from RichardErkhov

All by RichardErkhov →

More Yi models

All Yi →

Other large models

All large models →