TH

Theia 21B v2d

TheiaGGUF by BeaverAIUpdated 2024-08-30
Download on Hugging Face
Parameters20.4B
Context1000K
Layers70
Architecturellama
Smallest quant7.8 GB
Quick answer

Theia 21B v2d needs about 15.2 GB of memory at Q4_K_M (a 12.4 GB file) with an 8K context window, so it fits on a 16 GB GPU. This repository has 6 quants, from 7.8 GB (Q2_K) to 21.7 GB (Q8_0). Longer context needs more: at 32K tokens the KV cache alone adds about 9.4 GB.

Theia 21B v2d is a 20.4B-parameter Theia model in GGUF format, quantized by BeaverAI. Its trained context length is 1000K tokens. The repository has 6 quant files, from 7.8 GB (Q2_K) to 21.7 GB (Q8_0).

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
Q2_K7.8 GB 10.6 GB
Too big Get file →
Q3_K_M10.0 GB 12.9 GB
Too big Get file →
Q4_K_M12.4 GB 15.2 GB
Too big Get file →
Q5_K_M14.5 GB 17.3 GB
Too big Get file →
Q6_K16.8 GB 19.6 GB
Too big Get file →
Q8_021.7 GB 24.6 GB
Too big Get file →

What GPU do you need for Theia 21B v2d?

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 Q2_K (7.8 GB file, about 10.6 GB in use).
  • 16 GB VRAM: the best quant that fits is Q4_K_M (12.4 GB file, about 15.2 GB in use).
  • 24 GB VRAM: the best quant that fits is Q6_K (16.8 GB file, about 19.6 GB in use).
  • 32 GB VRAM: the best quant that fits is Q8_0 (21.7 GB file, about 24.6 GB in use).

Which Theia 21B v2d quant should you choose?

Q4_K_M is the usual starting point for Theia 21B v2d: it balances file size and quality well. Here it is a 12.4 GB file.

The Q8_0 file is 21.7 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 7.8 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
Parameters20.4B
Layers70
Trained context1,024,000 tokens
KV cache (fp16)280 KB per token

Source

Quantized byBeaverAI
FamilyTheia
Downloads (HF)12
Last updated2024-08-30

Frequently asked questions about Theia 21B v2d

How much VRAM does Theia 21B v2d need?

About 15.2 GB at Q4_K_M with an 8K context window. That is the 12.4 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 Theia 21B v2d on an 8 GB GPU?

Not at an 8K context. The smallest file needs about 10.6 GB, so you would need a larger GPU or CPU offloading.

What is the file size of Theia 21B v2d GGUF?

It depends on the quant: Q2_K is 7.8 GB, Q3_K_M is 10.0 GB, Q4_K_M is 12.4 GB, Q5_K_M is 14.5 GB, Q6_K is 16.8 GB, Q8_0 is 21.7 GB.

What context length does Theia 21B v2d support?

The GGUF header lists a trained context length of 1,024,000 tokens (1000K). The KV cache grows with the context you use: about 2.3 GB at 8K and 9.4 GB at 32K.

Where can I download Theia 21B v2d GGUF?

The files are on Hugging Face in the BeaverAI/Theia-21B-v2d-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 BeaverAI

All by BeaverAI →

More Theia models

All Theia →

Other medium models

All medium models →