Qwen2.5-VL-7B-Instruct-heretic
A decensored variant of Qwen/Qwen2.5-VL-7B-Instruct, produced with Heretic v1.4.0 (directional ablation / "abliteration"). Refusal behavior is suppressed via targeted weight edits to the attention output and MLP down-projections rather than fine-tuning, so the base model's vision-language knowledge and instruction-following are left largely intact.
Who this is for: developers who want Qwen2.5-VL's vision-language capabilities - OCR, document parsing, object grounding, video understanding - without refusals. Not a capability upgrade over base Qwen2.5-VL-7B-Instruct - same model, refusal guardrails removed.
Runs on your gaming PC
Full GGUF ladder included — pick the quant that fits your card:
| Your GPU | Recommended quant | Weights |
|---|---|---|
| RTX 3090 / 4090 / 5090 (24 GB) | Q8_0 | 8.10 GB |
| RTX 4080 / 5080 / 4060 Ti 16G (16 GB) | Q6_K | 6.25 GB |
| RTX 3060 / 4070 / 5070 (12 GB) | Q5_K_M | 5.44 GB |
| RTX 4060 / 3070 (8 GB) | Q4_K_M | 4.68 GB |
| GTX 1660 Super / 2060 / 3050 laptop (6 GB) | Q4_K_M | 4.68 GB |
| CPU-only / Apple Silicon | Q4_K_M | 4.68 GB, fits in system RAM |
Weights only, at this model's ~8.3B native size; add ~1 GB for context. OOM? Drop one quant level. Headroom to spare? Go one up.
Why abliteration instead of fine-tuning
Fine-tuning a "helpful" persona on top of RLHF'd refusals fights the base model's training and tends to degrade coherence. Abliteration instead finds and edits the specific weight directions responsible for refusal, leaving the rest of the network (and its capabilities) untouched. See the Heretic repo and the original abliteration writeup for the mechanism.
Files
Safetensors (BF16)
The full-precision weights are in model.safetensors (see the repo file listing for exact sizes).
GGUF quantizations
GGUF quantizations are published for this model (Q4_K_M, Q5_K_M, Q6_K, Q8_0). Pull a specific quant with llama.cpp / ollama.
| File | Format | Size |
|---|---|---|
Qwen2.5-VL-7B-Instruct-heretic-Q4_K_M.gguf | GGUF Q4_K_M | (see repo files) |
Qwen2.5-VL-7B-Instruct-heretic-Q5_K_M.gguf | GGUF Q5_K_M | (see repo files) |
Qwen2.5-VL-7B-Instruct-heretic-Q6_K.gguf | GGUF Q6_K | (see repo files) |
Qwen2.5-VL-7B-Instruct-heretic-Q8_0.gguf | GGUF Q8_0 | (see repo files) |
The GGUFs contain the text backbone (the GGUF format as produced by llama.cpp does not carry the vision encoder); the full vision-language model is available via the Safetensors weights above.
Quickstart
# llama.cpp - defaults to the Q4_K_M quant
llama serve -hf saidutta69/Qwen2.5-VL-7B-Instruct-heretic:Q4_K_M
# transformers (vision-language)
from transformers import Qwen2_5_VLForConditionalGeneration, AutoProcessor
model_name = "saidutta69/Qwen2.5-VL-7B-Instruct-heretic"
model = Qwen2_5_VLForConditionalGeneration.from_pretrained(model_name, torch_dtype="auto", device_map="auto")
processor = AutoProcessor.from_pretrained(model_name)
# ... inference code (see Qwen2.5-VL docs for chat template usage)
Also runnable via Ollama, LM Studio, Jan, vLLM, SGLang.
Responsible use
Refusal suppression is deliberate and works as intended: this model will comply with requests the base model would refuse, including some it shouldn't. There is no safety filtering layered on top. You are responsible for how you deploy it.
Made with ❤️ by RACER IS OP — follow for more uncensored models
License
Inherits the apache-2.0 license from the base model.