GLM-5.3 is live. Run Z.ai's latest model on Friendli Model APIs. Try it today ➜
masterman331
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Run this model inference with full control and performance in your environment.
Model Details
Model Provider
masterman331
Model Tree
WeiboAI/VibeThinker-3B
Input Modalities
Output Modalities
Supported Functionality
The model has had its refusal direction ablated via a difference-of-means LoRA on the residual stream, then merged back into the base weights. The goal is to remove knee-jerk refusals while preserving the base model's reasoning capability.
This is a derivative of WeiboAI/VibeThinker-3B. Go take a look at the original repo for full details on the base — architecture, training procedure, intended use cases, and licensing terms all carry over from there.
| Metric | This model | Original (by definition) |
|---|---|---|
| KL divergence | 0.0870 | 0 |
| Refusals | 2/100 | ~64/100 |
KL divergence of 0.087 is low — well under the 0.5 threshold above which abliteration typically starts damaging model capability. The 2/100 refusal count on the held-out eval set indicates the abliteration generalized rather than just memorizing training prompts.
Based on the benchmarks, this model appears to be quite strong. However, from my own testing, I have had difficulty maintaining meaningful conversations with it. The model seems to struggle with certain uncensored instructions and generally does not appear to be optimized for conversational use.
I have not personally tested its one-shot generation capabilities, but according to the official repository, that is where much of its advertised performance comes from.
Overall, I recommend using strong, specific prompts when seeking particular information or task completion, rather than relying on it for extended conversation.
heretic-llm)mean(bad_residuals) - mean(good_residuals) per layer, projected out of attention and MLP weights via a rank-3 LoRA adapter, then mergedmlabonne/harmless_alpaca (train[:400])mlabonne/harmful_behaviors (test[:100])You are a helpful assistant.To reproduce the same pipeline on any other model:
bash
pip install -U heretic-llmheretic --model WeiboAI/VibeThinker-3BThen follow the interactive prompts. See the Heretic repo for the full parameter space.
Same as the base model, minus refusal behavior. Apply your own application-level review or safety layer for any deployment where end-user-facing safety matters — this model ships with none. It has been specifically designed to be fully uncensored across all topics. Please use it responsibly and with care.
Inherits from WeiboAI/VibeThinker-3B. Check the original repo for the exact terms.
pip install -U heretic-llmheretic --model WeiboAI/VibeThinker-3B