GLM-5.2 is live. #1 throughput on OpenRouter, pay-per-token on FriendliAI. Try it today ➜
asmanovlev
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Model Details
Model Provider
asmanovlev
Model Tree
tsinghua-sigs-robot-lab/veriloop-coder-e1
Input Modalities
Output Modalities
Supported Functionality
| Metric | Value |
|---|---|
| Best trial | Trial 36 |
| Refusals (harmful_behaviors) | 82/100 |
| KL divergence | 0.0003 |
| Model damage | Minimal |
| Export format | LoRA adapter (26 MB) |
The model proved unusually resistant to abliteration. After 200 trials, refusal rate only dropped from ~95% to ~82%. Qwen 3.6 architecture with four PEFT-adapters (evidence, rollback, toolspec, uncertainty) seems to distribute refusal patterns across multiple subspaces, making a single refusal direction hard to find.
| File | Size | Description |
|---|---|---|
adapter_model.safetensors | 26 MB | LoRA adapter weights |
adapter_config.json | 1 KB | LoRA configuration |
python
from peft import PeftModelfrom transformers import AutoModelForCausalLMmodel = AutoModelForCausalLM.from_pretrained("tsinghua-sigs-robot-lab/veriloop-coder-e1")model = PeftModel.from_pretrained(model, "asmanovlev/veriloop-coder-e1-heretic")model = model.merge_and_unload()
--method aggressive (yields 0% refusals at the cost of KL ~8.7)