granite-4.1-30b-abliterated
30BGranite
from ibm-granite/granite-4.1-30b
↓ 5.9K♥ 107 repos
ai.granite-vision-3.2-2b-abliterated
ai.granite-vision-3.2-2b-abliterated is a refusal-ablated variant of huihui-ai/granite-vision-3.2-2b-abliterated, published on Hugging Face by DevQuasar-5. It is 2B parameters and runs on a laptop CPU class. It is distributed across 2 repositories in GGUF format, totalling 15.7K downloads.
Specification
- Model name
- ai.granite-vision-3.2-2b-abliterated
- Source repository
- DevQuasar-5/huihui-ai.granite-vision-3.2-2b-abliterated-GGUF
- Publisher
- DevQuasar-5
- Parameters
- 2B
- Hardware class
- Under 4B — runs on a laptop CPU
- Licence
- Not declared
- Task
- Image text to text
- Context window
- Not documented
- Ablation scope
- Not stated by the publisher
- Formats available
- GGUF
- First indexed
- 11 Mar 2025
- Last updated
- 11 Mar 2025
Ablation technique
How ai.granite-vision-3.2-2b-abliterated was modified, and what that implies.
Abliteration (directional ablation)
The single residual-stream direction that mediates refusal is identified from harmful/harmless prompt pairs and projected out of the model weights.
Reported scope: Not stated by the publisher. Publishers frequently omit this, so verify behaviour empirically rather than assuming full coverage.
From the publisher’s model card
“Quantized version of: huihui-ai/granite-vision-3.2-2b-abliterated”
Running it
Commands are templates — confirm the exact repository and quant file before use.
llama.cpp / GGUF
hf download DevQuasar-5/huihui-ai.granite-vision-3.2-2b-abliterated-GGUF --local-dir ./ai.granite-vision-3.2-2b-abliterated
llama-cli -m ./ai.granite-vision-3.2-2b-abliterated/<file>.gguf -p "..."Transformers
from transformers import AutoModelForCausalLM, AutoTokenizer
repo = "DevQuasar-5/huihui-ai.granite-vision-3.2-2b-abliterated-GGUF"
tok = AutoTokenizer.from_pretrained(repo)
model = AutoModelForCausalLM.from_pretrained(
repo, torch_dtype="auto", device_map="auto"
)vLLM
vllm serve DevQuasar-5/huihui-ai.granite-vision-3.2-2b-abliterated-GGUF --trust-remote-codeDownloads and variants (1)
The only published repository of ai.granite-vision-3.2-2b-abliterated.
| Repository | Format | Downloads |
|---|---|---|
| DevQuasar-5/huihui-ai.granite-vision-3.2-2b-abliterated-GGUFsource | GGUF | 15.7K |
Security-team assessment
Derived from this model’s metadata and the publisher’s claims — not from benchmarks run by this site.
- ai.granite-vision-3.2-2b-abliterated will attempt offensive-security prompts that a hosted commercial model declines, which is what makes it usable for red-team corpus generation and for measuring what an unaligned model of this class produces.
- Capability is inherited from huihui-ai/granite-vision-3.2-2b-abliterated, not added by ablation — at 2B parameters, expect a small model’s reasoning and a high error rate on exploit detail.
- The publisher does not state which layers were ablated, so assume nothing about refusal consistency — probe it directly.
- No licence is declared on this repository, which is a compliance problem in itself. Resolve it against the base model’s terms before use.
Handling: run it isolated, put moderation in front of it if anyone outside your team can reach it, and verify the weights before loading — these are third-party artifacts. Full handling and licence guidance.
Metadata for ai.granite-vision-3.2-2b-abliterated is collected from the public Hugging Face API and the publisher’s model card. This site does not host weights, is not affiliated with DevQuasar-5, and does not independently verify publisher claims. See responsible use.