LL

Meta Llama 3.1 13B Instruct Abliterated

LlamaGGUF by mradermacherUpdated 2024-08-02
Download on Hugging Face
Parameters13.3B
Context128K
Layers56
Architecturellama
Smallest quant3.1 GB
Quick answer

Meta Llama 3.1 13B Instruct Abliterated needs about 10.4 GB of memory at Q4_K_M (a 8.1 GB file) with an 8K context window, so it fits on a 12 GB GPU. This repository has 21 quants, from 3.1 GB (IQ1_S) to 10.9 GB (Q6_K). Longer context needs more: at 32K tokens the KV cache alone adds about 7.5 GB.

Meta Llama 3.1 13B Instruct Abliterated – a 13.3B parameter instruction model

Meta Llama 3.1 13B Instruct Abliterated is a 13.3B parameter model trained on 131072 context tokens. It is available in multiple GGUF quantizations from IQ1_S to Q6_K. The model supports instruction following but has no vision capabilities.

Good at

  • Instruction following
  • Large context handling
  • Low‑memory quant options

Limits

  • No vision support
  • License not specified
  • Requires 8k context for memory estimates

Best for

  • Chatbot development
  • Text summarization
  • Code generation

Who it’s for. Developers and researchers needing a 13B instruction model with flexible quantization options. Suitable for local deployment on systems with 8–13 GB VRAM.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
IQ1_S3.1 GB 5.5 GB
Fits Get file →
IQ1_M3.4 GB 5.8 GB
Fits Get file →
IQ2_XXS3.8 GB 6.2 GB
Fits Get file →
IQ2_XS4.2 GB 6.5 GB
Fits Get file →
IQ2_S4.4 GB 6.8 GB
Fits Get file →
IQ2_M4.7 GB 7.1 GB
Fits Get file →
Q2_K5.1 GB 7.5 GB
Tight Get file →
IQ3_XXS5.3 GB 7.7 GB
Tight Get file →
IQ3_XS5.7 GB 8.0 GB
Too big Get file →
Q3_K_S5.9 GB 8.3 GB
Too big Get file →
IQ3_S5.9 GB 8.3 GB
Too big Get file →
IQ3_M6.1 GB 8.5 GB
Too big Get file →
Q3_K_M6.5 GB 8.9 GB
Too big Get file →
Q3_K_L7.1 GB 9.4 GB
Too big Get file →
IQ4_XS7.2 GB 9.6 GB
Too big Get file →
Q4_07.6 GB 10.0 GB
Too big Get file →
Q4_K_S7.7 GB 10.0 GB
Too big Get file →
Q4_K_M8.1 GB 10.4 GB
Too big Get file →
Q5_K_S9.2 GB 11.6 GB
Too big Get file →
Q5_K_M9.4 GB 11.8 GB
Too big Get file →
Q6_K10.9 GB 13.3 GB
Too big Get file →

For 8 GB VRAM, use IQ1_S (3.13 GB, 5.5 GB memory) or IQ1_M (3.38 GB, 5.8 GB memory). With 10 GB VRAM, Q4_0 (7.63 GB, 10 GB memory) or Q4_K_M (8.06 GB, 10.4 GB memory) are suitable. For 13 GB VRAM, Q6_K (10.89 GB, 13.3 GB memory) fits the budget.

What GPU do you need for Meta Llama 3.1 13B Instruct Abliterated?

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 IQ1_M (3.4 GB file, about 5.8 GB in use).
  • 8 GB VRAM: the best quant that fits is IQ3_XS (5.7 GB file, about 8.0 GB in use).
  • 12 GB VRAM: the best quant that fits is Q5_K_M (9.4 GB file, about 11.8 GB in use).
  • 16 GB VRAM: the best quant that fits is Q6_K (10.9 GB file, about 13.3 GB in use).

Which Meta Llama 3.1 13B Instruct Abliterated quant should you choose?

Q4_K_M is the usual starting point for Meta Llama 3.1 13B Instruct Abliterated: it balances file size and quality well. Here it is a 8.1 GB file.

If memory is tight, the smallest file is IQ1_S at 3.1 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
Parameters13.3B
Layers56
Trained context131,072 tokens
KV cache (fp16)224 KB per token

Source

Quantized bymradermacher
FamilyLlama
Downloads (HF)1,446
Last updated2024-08-02

Frequently asked questions about Meta Llama 3.1 13B Instruct Abliterated

How much VRAM does Meta Llama 3.1 13B Instruct Abliterated need?

About 10.4 GB at Q4_K_M with an 8K context window. That is the 8.1 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 Meta Llama 3.1 13B Instruct Abliterated on an 8 GB GPU?

Yes, with IQ3_XS: the 5.7 GB file needs about 8.0 GB at 8K context. Larger quants in this repository do not fit in 8 GB.

What is the file size of Meta Llama 3.1 13B Instruct Abliterated GGUF?

It depends on the quant: IQ1_S is 3.1 GB, IQ1_M is 3.4 GB, IQ2_XXS is 3.8 GB, IQ2_XS is 4.2 GB, IQ2_S is 4.4 GB, IQ2_M is 4.7 GB and 15 more.

What context length does Meta Llama 3.1 13B Instruct Abliterated support?

The GGUF header lists a trained context length of 131,072 tokens (128K). The KV cache grows with the context you use: about 1.9 GB at 8K and 7.5 GB at 32K.

Where can I download Meta Llama 3.1 13B Instruct Abliterated GGUF?

The files are on Hugging Face in the mradermacher/Meta-Llama-3.1-13B-Instruct-abliterated-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, use IQ1_S or IQ1_M quantizations; they require about 5.5–5.8 GB memory at 8k context.

Does this model support vision tasks?

No, the model has no vision capabilities.

What is the maximum context length?

The model was trained on 131072 context tokens.

Similar models

Browse all →

More from mradermacher

All by mradermacher →

More Llama models

All Llama →

Other medium models

All medium models →