LA

Lamma3merge3-15B-MoE

LammaGGUF by RichardErkhovUpdated 2024-07-21
Download on Hugging Face
Parameters13.7B
Context8K
Layers32
Architecturellama
Smallest quant3.9 GB
Quick answer

Lamma3merge3-15B-MoE needs about 9.9 GB of memory at Q4_K_M (a 8.3 GB file) with an 8K context window, so it fits on a 12 GB GPU. This repository has 19 quants, from 3.9 GB (IQ4_NL) to 9.7 GB (Q5_K_M).

Lamma3merge3-15B-MoE is a 13.7B-parameter Lamma model in GGUF format, quantized by RichardErkhov. Its trained context length is 8K tokens. The repository has 19 quant files, from 3.9 GB (IQ4_NL) to 9.7 GB (Q5_K_M).

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
IQ4_NL3.9 GB 5.5 GB
Fits Get file →
Q4_K4.8 GB 6.3 GB
Fits Get file →
Q2_K5.2 GB 6.8 GB
Fits Get file →
IQ3_XS5.8 GB 7.4 GB
Tight Get file →
Q3_K_S6.1 GB 7.7 GB
Tight Get file →
IQ3_S6.1 GB 7.7 GB
Tight Get file →
IQ3_M6.2 GB 7.8 GB
Tight Get file →
Q3_K6.7 GB 8.3 GB
Too big Get file →
Q3_K_M6.7 GB 8.3 GB
Too big Get file →
Q3_K_L7.2 GB 8.8 GB
Too big Get file →
IQ4_XS7.5 GB 9.1 GB
Too big Get file →
Q4_07.8 GB 9.4 GB
Too big Get file →
Q4_K_S7.9 GB 9.5 GB
Too big Get file →
Q4_K_M8.3 GB 9.9 GB
Too big Get file →
Q4_18.7 GB 10.2 GB
Too big Get file →
Q5_09.5 GB 11.0 GB
Too big Get file →
Q5_K_S9.5 GB 11.0 GB
Too big Get file →
Q5_K9.7 GB 11.3 GB
Too big Get file →
Q5_K_M9.7 GB 11.3 GB
Too big Get file →

What GPU do you need for Lamma3merge3-15B-MoE?

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 IQ4_NL (3.9 GB file, about 5.5 GB in use).
  • 8 GB VRAM: the best quant that fits is IQ3_M (6.2 GB file, about 7.8 GB in use).
  • 12 GB VRAM: the best quant that fits is Q5_K (9.7 GB file, about 11.3 GB in use).

Which Lamma3merge3-15B-MoE quant should you choose?

Q4_K_M is the usual starting point for Lamma3merge3-15B-MoE: it balances file size and quality well. Here it is a 8.3 GB file.

If memory is tight, the smallest file is IQ4_NL at 3.9 GB, with the quality trade-offs that come with fewer bits. The usual sweet spot between file size and quality, and the quant most people start with.

New to quant names? Read GGUF quantization explained.

Specifications

Architecture from the GGUF header

Architecturellama
Parameters13.7B (approximate, mixture of experts)
Layers32
Trained context8,192 tokens
KV cache (fp16)128 KB per token

Source

Quantized byRichardErkhov
FamilyLamma
Downloads (HF)96
Last updated2024-07-21

Frequently asked questions about Lamma3merge3-15B-MoE

How much VRAM does Lamma3merge3-15B-MoE need?

About 9.9 GB at Q4_K_M with an 8K context window. That is the 8.3 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 Lamma3merge3-15B-MoE on an 8 GB GPU?

Yes, with IQ3_M: the 6.2 GB file needs about 7.8 GB at 8K context. Larger quants in this repository do not fit in 8 GB.

What is the file size of Lamma3merge3-15B-MoE GGUF?

It depends on the quant: IQ4_NL is 3.9 GB, Q4_K is 4.8 GB, Q2_K is 5.2 GB, IQ3_XS is 5.8 GB, Q3_K_S is 6.1 GB, IQ3_S is 6.1 GB and 13 more.

What context length does Lamma3merge3-15B-MoE support?

The GGUF header lists a trained context length of 8,192 tokens (8K). The KV cache grows with the context you use: about 1.1 GB at 8K and 1.1 GB at 8K.

Where can I download Lamma3merge3-15B-MoE GGUF?

The files are on Hugging Face in the RichardErkhov/allknowingroger_-_Lamma3merge3-15B-MoE-gguf repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.

Similar models

Browse all →

More from RichardErkhov

All by RichardErkhov →

More Lamma models

All Lamma →

Other medium models

All medium models →