Sarashina2 7b Abliterated needs about 9.3 GB of memory at Q4_K_M (a 4.5 GB file) with an 8K context window, so it fits on a 12 GB GPU. This repository has 25 quants, from 0.0 GB to 6.0 GB (Q6_K).
Sarashina2 7b Abliterated – a 7.32B llama model with refusal direction removed
Sarashina2 7b Abliterated is a 7.32B parameter llama model that has had its refusal direction orthogonalized. The model is released under the MIT license and supports a 4096 token context window. It is intended for research into safety and alignment.
Good at
- Research on refusal mechanisms
- Red‑teaming and adversarial testing
- Instruction‑response tasks
Limits
- Safety features removed – may generate harmful content
- No vision support
- Requires 8k context for memory estimates
Best for
- AI safety research
- Alignment studies
- Red‑team testing
Who it’s for. Researchers and developers working on AI safety, alignment, and red‑team testing. The model is suitable for those who need to study or evaluate refusal mechanisms in language models.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| — | 0.0 GB | 2.7 GB | Fits | Get file → | |
| IQ1_S | 1.8 GB | 4.5 GB | Fits | Get file → | |
| IQ1_M | 1.9 GB | 4.6 GB | Fits | Get file → | |
| IQ2_XXS | 2.1 GB | 4.8 GB | Fits | Get file → | |
| IQ2_XS | 2.3 GB | 5.0 GB | Fits | Get file → | |
| IQ2_S | 2.5 GB | 5.2 GB | Fits | Get file → | |
| Q2_K_S | 2.7 GB | 5.3 GB | Fits | Get file → | |
| IQ2_M | 2.7 GB | 5.3 GB | Fits | Get file → | |
| Q2_K | 2.9 GB | 5.5 GB | Fits | Get file → | |
| IQ3_XXS | 2.9 GB | 5.6 GB | Fits | Get file → | |
| IQ3_XS | 3.2 GB | 5.8 GB | Fits | Get file → | |
| IQ3_S | 3.3 GB | 6.0 GB | Fits | Get file → | |
| Q3_K_S | 3.3 GB | 6.0 GB | Fits | Get file → | |
| IQ3_M | 3.5 GB | 6.1 GB | Fits | Get file → | |
| Q3_K_M | 3.7 GB | 6.3 GB | Fits | Get file → | |
| Q3_K_L | 4.0 GB | 6.6 GB | Fits | Get file → | |
| IQ4_XS | 4.0 GB | 6.7 GB | Fits | Get file → | |
| IQ4_NL | 4.2 GB | 6.9 GB | Fits | Get file → | |
| Q4_0 | 4.2 GB | 6.9 GB | Fits | Get file → | |
| Q4_K_S | 4.3 GB | 6.9 GB | Fits | Get file → | |
| Q4_K_M | 4.5 GB | 7.1 GB | Fits | Get file → | |
| Q4_1 | 4.7 GB | 7.3 GB | Tight | Get file → | |
| Q5_K_S | 5.1 GB | 7.7 GB | Tight | Get file → | |
| Q5_K_M | 5.2 GB | 7.9 GB | Tight | Get file → | |
| Q6_K | 6.0 GB | 8.7 GB | Too big | Get file → |
For 8 GB VRAM use Q4_K_M (4.48 GB, ~9.3 GB memory at 8k context). For 12 GB VRAM use Q5_K_M (5.22 GB, ~10 GB memory). For 16 GB VRAM use Q6_K (6 GB, ~10.8 GB memory). Estimates are approximate.
What GPU do you need for Sarashina2 7b 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 the main file (0.0 GB file, about 4.8 GB in use).
- 8 GB VRAM: the best quant that fits is IQ3_XS (3.2 GB file, about 8.0 GB in use).
- 12 GB VRAM: the best quant that fits is Q6_K (6.0 GB file, about 10.8 GB in use).
Which Sarashina2 7b Abliterated quant should you choose?
Q4_K_M is the usual starting point for Sarashina2 7b Abliterated: it balances file size and quality well. Here it is a 4.5 GB file.
If memory is tight, the smallest file is the first one in the table at 0.0 GB, with the quality trade-offs that come with fewer bits.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about Sarashina2 7b Abliterated
How much VRAM does Sarashina2 7b Abliterated need?
About 9.3 GB at Q4_K_M with an 8K context window. That is the 4.5 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 Sarashina2 7b Abliterated on an 8 GB GPU?
Yes, with IQ3_XS: the 3.2 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 Sarashina2 7b Abliterated GGUF?
It depends on the quant: file is 0.0 GB, IQ1_S is 1.8 GB, IQ1_M is 1.9 GB, IQ2_XXS is 2.1 GB, IQ2_XS is 2.3 GB, IQ2_S is 2.5 GB and 19 more.
What context length does Sarashina2 7b Abliterated support?
The GGUF header lists a trained context length of 4,096 tokens (4K). The KV cache grows with the context you use: about 4.3 GB at 8K and 2.1 GB at 4K.
Where can I download Sarashina2 7b Abliterated GGUF?
The files are on Hugging Face in the mradermacher/sarashina2-7b-abliterated-i1-GGUF repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.
What license does Sarashina2 7b Abliterated use?
Hugging Face lists the license as mit. Licenses can differ between the quantized files and the original model, so check the repository and the base model before commercial use.
Can I run this model on 8 GB VRAM?
Yes, the Q4_K_M quant uses 4.48 GB file size and requires about 9.3 GB memory at 8k context.
Does the model support vision?
No, the model has no vision capabilities.
Is the model safe for general use?
The safety guardrails have been removed; the model can produce harmful content and is intended for research only.