LL

llemma 34b

LlemmaGGUF by TheBlokeUpdated 2023-10-17
Download on Hugging Face
Parameters33.7B
Context4K
Layers48
Architecturellama
Smallest quant14.2 GB
Quick answer

llemma 34b needs about 22.3 GB of memory at Q4_K_M (a 20.2 GB file) with an 8K context window, so it fits on a 24 GB GPU. This repository has 12 quants, from 14.2 GB (Q2_K) to 35.9 GB (Q8_0).

llemma 34b is a 33.7B-parameter Llemma model in GGUF format, quantized by TheBloke. Its trained context length is 4K tokens. The repository has 12 quant files, from 14.2 GB (Q2_K) to 35.9 GB (Q8_0).

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
Q2_K14.2 GB 15.5 GB
Too big Get file →
Q3_K_S14.6 GB 15.9 GB
Too big Get file →
Q3_K_M16.3 GB 17.6 GB
Too big Get file →
Q3_K_L17.8 GB 19.1 GB
Too big Get file →
Q4_019.1 GB 20.4 GB
Too big Get file →
Q4_K_S19.1 GB 20.5 GB
Too big Get file →
Q4_K_M20.2 GB 21.5 GB
Too big Get file →
Q5_023.2 GB 24.5 GB
Too big Get file →
Q5_K_S23.2 GB 24.5 GB
Too big Get file →
Q5_K_M23.8 GB 25.1 GB
Too big Get file →
Q6_K27.7 GB 29.0 GB
Too big Get file →
Q8_035.9 GB 37.2 GB
Too big Get file →

What GPU do you need for llemma 34b?

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: the best quant that fits is Q4_K_M (20.2 GB file, about 22.3 GB in use).
  • 32 GB VRAM: the best quant that fits is Q6_K (27.7 GB file, about 29.8 GB in use).
  • 48 GB VRAM: the best quant that fits is Q8_0 (35.9 GB file, about 38.0 GB in use).

Which llemma 34b quant should you choose?

Q4_K_M is the usual starting point for llemma 34b: it balances file size and quality well. Here it is a 20.2 GB file.

The Q8_0 file is 35.9 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 14.2 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
Parameters33.7B
Layers48
Trained context4,096 tokens
KV cache (fp16)192 KB per token

Source

Quantized byTheBloke
FamilyLlemma
Licensellama2
Downloads (HF)360
Last updated2023-10-17

Frequently asked questions about llemma 34b

How much VRAM does llemma 34b need?

About 22.3 GB at Q4_K_M with an 8K context window. That is the 20.2 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 llemma 34b on an 8 GB GPU?

Not at an 8K context. The smallest file needs about 16.3 GB, so you would need a larger GPU or CPU offloading.

What is the file size of llemma 34b GGUF?

It depends on the quant: Q2_K is 14.2 GB, Q3_K_S is 14.6 GB, Q3_K_M is 16.3 GB, Q3_K_L is 17.8 GB, Q4_0 is 19.1 GB, Q4_K_S is 19.1 GB and 6 more.

What context length does llemma 34b 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 llemma 34b GGUF?

The files are on Hugging Face in the TheBloke/llemma_34b-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 llemma 34b use?

Hugging Face lists the license as llama2. Licenses can differ between the quantized files and the original model, so check the repository and the base model before commercial use.

Other GGUFs of llemma_34b

Similar models

Browse all →

More from TheBloke

All by TheBloke →

More Llemma models

All Llemma →

Other medium models

All medium models →