GLM-5.3 is live. Run Z.ai's latest model on Friendli Model APIs. Try it today ➜
gorbatjovy
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Model Details
Model Provider
gorbatjovy
Model Tree
DavidAU/Qwen3.8-27B-Cold-Fusion-GAIN-V1.1
Input Modalities
Output Modalities
Supported Functionality
This model has had its safety alignment removed via abliteration. As a result it will comply with harmful, unethical, or offensive requests that the base model would refuse. It is released for research — interpretability, refusal-mechanism study, red-teaming, and robustness evaluation. You are responsible for what you do with it and for everything it generates. Do not deploy it to end users without adding your own safety and moderation layers. Use must comply with the inherited Apache 2.0 license and all applicable law. The authors accept no liability for misuse.
Abliteration removes the single "refusal direction" from the residual stream (Arditi et al. 2024, Refusal in Language Models Is Mediated by a Single Direction). Heretic estimates that direction from harmful (AdvBench-style) vs harmless (Alpaca) prompts and runs a 200-trial Optuna search that co-minimizes refusals and KL divergence from the base model, then orthogonalizes the chosen direction out of the residual-writing weights.
| Base model | DavidAU/Qwen3.8-27B-Cold-Fusion-GAIN-V1.1 |
| Method | Heretic (automated, KL-constrained) — single refusal direction, k=1 |
| Selected trial | direction index 34.95, per-layer ablation kernel from the Optuna search |
| Coverage | 131 residual writers (self_attn.o_proj, linear_attn.out_proj, mlp.down_proj, embed_tokens) incl. the MTP head; the 333 visual.* tensors are untouched |
| Format | safetensors, BF16, 1199 tensors |
| Architecture | Qwen3_5ForConditionalGeneration — 64 layers, hybrid Gated DeltaNet + full attention, native VL tower, MTP head |
Note: Heretic's own export drops the MTP head (transformers has no MTP module for this architecture). It was reinstated from the base checkpoint and abliterated consistently with the main model, so speculative decoding keeps working and the draft head does not re-introduce refusals.
Two different rulers, both reported honestly:
Capability is essentially retained (KL 0.0315 indicates minimal drift), and the vision tower is byte-for-byte identical to the base, so image understanding is unchanged. As with any abliteration, fine-tuning on refusal-heavy data can partially reintroduce refusals.
python
import torchfrom transformers import AutoProcessor, AutoModelForImageTextToTextmodel_id = "gorbatjovy/Qwen3.8-27B-Cold-Fusion-GAIN-V1.1-heretic"processor = AutoProcessor.from_pretrained(model_id)model = AutoModelForImageTextToText.from_pretrained(model_id, torch_dtype=torch.bfloat16, device_map="auto")messages = [{"role": "user", "content": "Prove that sqrt(2) is irrational."}]inputs = processor.apply_chat_template(messages, add_generation_prompt=True, tokenize=True,return_dict=True, return_tensors="pt", enable_thinking=True,).to(model.device)out = model.generate(**inputs, max_new_tokens=512)print(processor.decode(out[0][inputs["input_ids"].shape[1]:], skip_special_tokens=True))
BF16 weights are ~56 GB. For a smaller footprint use the W4A16 build. Ninfer build.
Apache 2.0, inherited from the base model. Abliteration does not change the license.