GLM-5.3 is live. Run Z.ai's latest model on Friendli Model APIs. Try it today ➜
tusharislampure29
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
tusharislampure29
Model Tree
Qwen/Qwen2.5-3B-Instruct
Input Modalities
Output Modalities
Supported Functionality
Reward models / LLM-judges are the linchpin of RLHF/RLVR, yet a single "Final verdict: CORRECT" cue can flip many of them into accepting a wrong answer — so RL policies learn to game the verifier instead of reasoning. This model is trained (adversarial augmentation + meaning- preserving consistency + position-diversified cues) to stay correct under those attacks.
| Metric | Qwen2.5-3B base | Hardened-3B |
|---|---|---|
| Robustness ↑ | 49.1 | 74.7 (#1 of 12) |
| Attack-success-rate ↓ | 0.635 | 0.082 |
| Clean accuracy | 0.617 | 0.577 |
| Instability ↓ | 0.140 | 0.037 |
Generalizes to unseen attack families (leave-one-family-out) and holds up far better than the base under a per-model searched attacker (searched ASR 0.193 vs base 0.95). Full methodology, leaderboard, and honest caveats: the GitHub repo.
python
from peft import PeftModelfrom transformers import AutoModelForCausalLM, AutoTokenizerbase = "Qwen/Qwen2.5-3B-Instruct"tok = AutoTokenizer.from_pretrained(base)model = AutoModelForCausalLM.from_pretrained(base, device_map="auto", load_in_4bit=True)model = PeftModel.from_pretrained(model, "tusharislampure29/VerifierBench-Hardened-3B")# Prompt it exactly as VerifierBench does (system: grade CORRECT/INCORRECT); it outputs a verdict.
Or plug straight into the benchmark:
python
from verifierbench.verifiers import get_verifierv = get_verifier("gen_judge", model="Qwen/Qwen2.5-3B-Instruct", load_in_4bit=True, adapter="tusharislampure29/VerifierBench-Hardened-3B")
Trained/evaluated on ProcessBench math (gsm8k + math). It defends the known Tier-A attack families and generalizes to held-out families, but a searched attacker with novel cues can still find some exploits — verifier robustness is a moving target. Apache-2.0.