ET

steelskull_etheria-55b-v0.1

EtheriaGGUF by TheBlokeUpdated 2024-01-26
Download on Hugging Face
Parameters55.6B
Context195K
Layers98
Architecturellama
Smallest quant20.6 GB
Quick answer

steelskull_etheria-55b-v0.1 needs about 37.1 GB of memory at Q4_K_M (a 33.3 GB file) with an 8K context window, so it fits on a 48 GB GPU. This repository has 11 quants, from 20.6 GB (Q2_K) to 45.6 GB (Q6_K). Longer context needs more: at 32K tokens the KV cache alone adds about 13.2 GB.

Steel's 55B Etheria model in GGUF format

steelskull_etheria-55b-v0.1 is a 55.6B parameter model based on the Etheria family. It was merged using the DARE TIES method and supports a theoretical context of 200k tokens, though 32k is recommended for stability. The model is available in multiple GGUF quantisations.

Good at

  • Large context handling
  • High parameter count
  • Merge of yi-34b models

Limits

  • No vision support
  • License not specified
  • Requires large VRAM for higher quants

Best for

  • Long‑form content generation
  • Research prototyping
  • Storytelling with extended context

Who it’s for. Developers and researchers needing a large‑parameter LLM for local deployment. The model is suitable for users with access to GPU resources that can handle the memory demands of higher‑quality quantisations.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
Q2_K20.6 GB 24.4 GB
Too big Get file →
Q3_K_S24.1 GB 27.9 GB
Too big Get file →
Q3_K_M26.9 GB 30.6 GB
Too big Get file →
Q3_K_L29.3 GB 33.1 GB
Too big Get file →
Q4_031.4 GB 35.2 GB
Too big Get file →
Q4_K_S31.6 GB 35.4 GB
Too big Get file →
Q4_K_M33.3 GB 37.1 GB
Too big Get file →
Q5_038.3 GB 42.1 GB
Too big Get file →
Q5_K_S38.3 GB 42.1 GB
Too big Get file →
Q5_K_M39.3 GB 43.1 GB
Too big Get file →
Q6_K45.6 GB 49.4 GB
Too big Get file →

For 8 GB VRAM, use Q2_K (20.61 GB file, ~24.4 GB memory at 8k context). For 16 GB VRAM, Q3_K_S (24.07 GB file, ~27.9 GB memory). For 24 GB VRAM, Q4_K_M (33.34 GB file, ~37.1 GB memory).

What GPU do you need for steelskull_etheria-55b-v0.1?

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: the best quant that fits is Q3_K_M (26.9 GB file, about 30.6 GB in use).
  • 48 GB VRAM: the best quant that fits is Q5_K_M (39.3 GB file, about 43.1 GB in use).

Which steelskull_etheria-55b-v0.1 quant should you choose?

Q4_K_M is the usual starting point for steelskull_etheria-55b-v0.1: it balances file size and quality well. Here it is a 33.3 GB file.

If memory is tight, the smallest file is Q2_K at 20.6 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
Parameters55.6B
Layers98
Trained context200,000 tokens
KV cache (fp16)392 KB per token

Source

Quantized byTheBloke
FamilyEtheria
Downloads (HF)833
Last updated2024-01-26

Frequently asked questions about steelskull_etheria-55b-v0.1

How much VRAM does steelskull_etheria-55b-v0.1 need?

About 37.1 GB at Q4_K_M with an 8K context window. That is the 33.3 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 steelskull_etheria-55b-v0.1 on an 8 GB GPU?

Not at an 8K context. The smallest file needs about 24.4 GB, so you would need a larger GPU or CPU offloading.

What is the file size of steelskull_etheria-55b-v0.1 GGUF?

It depends on the quant: Q2_K is 20.6 GB, Q3_K_S is 24.1 GB, Q3_K_M is 26.9 GB, Q3_K_L is 29.3 GB, Q4_0 is 31.4 GB, Q4_K_S is 31.6 GB and 5 more.

What context length does steelskull_etheria-55b-v0.1 support?

The GGUF header lists a trained context length of 200,000 tokens (195K). The KV cache grows with the context you use: about 3.3 GB at 8K and 13.2 GB at 32K.

Where can I download steelskull_etheria-55b-v0.1 GGUF?

The files are on Hugging Face in the TheBloke/Etheria-55b-v0.1-GGUF repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.

Can I run steelskull_etheria-55b-v0.1 on 8 GB VRAM?

Use the Q2_K quant which requires about 24.4 GB memory at 8k context, so 8 GB VRAM is insufficient.

What context length is safe to use?

Start at 32k context; the model theoretically supports 200k but stability is unverified.

Does the model support vision?

No, the model has no vision capabilities.

Other GGUFs of Etheria-55b-v0.1

Similar models

Browse all →

More from TheBloke

All by TheBloke →

More Etheria models

All Etheria →

Other large models

All large models →