CE

L3.1 Celestial Stone 2x8B

CelestialGGUF by mradermacherUpdated 2024-09-20
Download on Hugging Face
Parameters13.7B
Context128K
Layers32
Architecturellama
Smallest quant3.2 GB
Quick answer

L3.1 Celestial Stone 2x8B 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 24 quants, from 3.2 GB (IQ1_S) to 11.2 GB (Q6_K). Longer context needs more: at 32K tokens the KV cache alone adds about 4.3 GB.

13.7B parameter Llama model with 131072 token context

L3.1 Celestial Stone 2x8B is a 13.7B parameter Llama architecture model trained with a 131072 token context window. It is a mixture of experts and offers a range of quantizations for different memory budgets.

Good at

  • Large context handling
  • High parameter count
  • Mixture of experts

Limits

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

Best for

  • Long-form content generation
  • Complex reasoning tasks
  • Research prototyping

Who it’s for. Developers and researchers needing a large language model with extended context. Suitable for environments where GPU memory can accommodate the chosen quantization.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
IQ1_S3.2 GB 4.7 GB
Fits Get file →
IQ1_M3.4 GB 5.0 GB
Fits Get file →
IQ2_XXS3.9 GB 5.4 GB
Fits Get file →
IQ2_XS4.2 GB 5.8 GB
Fits Get file →
IQ2_S4.4 GB 6.0 GB
Fits Get file →
IQ2_M4.8 GB 6.4 GB
Fits Get file →
Q2_K5.2 GB 6.8 GB
Fits Get file →
IQ3_XXS5.4 GB 7.0 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_K_M6.7 GB 8.3 GB
Too big Get file →
Q3_K_L7.2 GB 8.8 GB
Too big Get file →
IQ4_XS7.4 GB 9.0 GB
Too big Get file →
Q4_07.8 GB 9.4 GB
Too big Get file →
Q4_07.8 GB 9.4 GB
Too big Get file →
Q4_07.8 GB 9.4 GB
Too big Get file →
Q4_07.9 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 →
Q5_K_S9.5 GB 11.0 GB
Too big Get file →
Q5_K_M9.7 GB 11.3 GB
Too big Get file →
Q6_K11.2 GB 12.8 GB
Too big Get file →

For 8‑GB VRAM use IQ1_S (3.15 GB, 4.7 GB memory) or IQ1_M (3.42 GB, 5 GB). For 12‑GB VRAM use Q4_K_M (8.33 GB, 9.9 GB) or Q5_K_S (9.48 GB, 11 GB). For 16‑GB VRAM use Q6_K (11.22 GB, 12.8 GB). Estimates are approximate.

What GPU do you need for L3.1 Celestial Stone 2x8B?

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 IQ2_S (4.4 GB file, about 6.0 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_M (9.7 GB file, about 11.3 GB in use).
  • 16 GB VRAM: the best quant that fits is Q6_K (11.2 GB file, about 12.8 GB in use).

Which L3.1 Celestial Stone 2x8B quant should you choose?

Q4_K_M is the usual starting point for L3.1 Celestial Stone 2x8B: it balances file size and quality well. Here it is a 8.3 GB file.

If memory is tight, the smallest file is IQ1_S at 3.2 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
Parameters13.7B (approximate, mixture of experts)
Layers32
Trained context131,072 tokens
KV cache (fp16)128 KB per token

Source

Quantized bymradermacher
FamilyCelestial
Downloads (HF)962
Last updated2024-09-20

Frequently asked questions about L3.1 Celestial Stone 2x8B

How much VRAM does L3.1 Celestial Stone 2x8B 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 L3.1 Celestial Stone 2x8B 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 L3.1 Celestial Stone 2x8B GGUF?

It depends on the quant: IQ1_S is 3.2 GB, IQ1_M is 3.4 GB, IQ2_XXS is 3.9 GB, IQ2_XS is 4.2 GB, IQ2_S is 4.4 GB, IQ2_M is 4.8 GB and 18 more.

What context length does L3.1 Celestial Stone 2x8B support?

The GGUF header lists a trained context length of 131,072 tokens (128K). The KV cache grows with the context you use: about 1.1 GB at 8K and 4.3 GB at 32K.

Where can I download L3.1 Celestial Stone 2x8B GGUF?

The files are on Hugging Face in the mradermacher/L3.1-Celestial-Stone-2x8B-DPO-i1-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 this model on 8 GB VRAM?

Yes, with IQ1_S or IQ1_M quantizations, which use about 4.7–5 GB of VRAM at 8k context.

Does this model support vision?

No, it is a text-only Llama model with no vision capabilities.

What is the context length of this model?

The model supports up to 131072 tokens of context.

Similar models

Browse all →

More from mradermacher

All by mradermacher →

More Celestial models

All Celestial →

Other medium models

All medium models →