---
language:
  - zh
license: cc-by-nc-sa-4.0
library_name: peft
pipeline_tag: text-classification
base_model: Qwen/Qwen3-8B-Base
base_model_relation: finetune
datasets:
  - thu-coai/COLD
tags:
  - toxicity-detection
  - hate-speech-detection
  - cyberbullying-detection
  - content-moderation
  - chinese
  - pinyin
  - adversarial-robustness
  - homophone-attack
  - multi-task
  - lora
  - peft
  - qwen3
metrics:
  - f1
  - accuracy
inference: false
extra_gated_prompt: >-
  CC BY-NC-SA 4.0 — non-commercial research and educational use only.
extra_gated_fields:
  Intended use: text
  Affiliation: text
  I agree to the non-commercial license: checkbox
model-index:
  - name: CyberPuppy-v6-pinyin-LoRA
    results:
      - task:
          type: text-classification
          name: Toxicity Detection (Pinyin-only)
        dataset:
          name: COLD
          type: thu-coai/COLD
          split: test
        metrics:
          - type: f1
            value: 0.7983
            name: F1 (weighted, pinyin-only dev)
---

# CyberPuppy v6 — Pinyin LoRA (LoRA-B) | Strengthened Homophone Defense

> 拼音分支 v6 · Qwen3-8B + LoRA **r=64** + 同音字攻擊強化防禦
>
> Companion to [v6-bilingual](https://huggingface.co/thc1006/cyberpuppy-v6-bilingual).
> Doubled rank from r=32 → **r=64** for stronger phonetic pattern recognition.

## What changed from v5

- LoRA rank: 32 → **64** (doubled)
- LoRA alpha: 64 → **128**
- Epochs: 3 → 5 (best at epoch 2)
- Max length: 128 → **192**
- Consistency loss: λ=0 → **λ=0.5**

## Performance impact (in v6 dual-LoRA ensemble)

| Metric | v5.1 | v6 |
|---|---|---|
| Pinyin-only dev F1 | 0.7979 | 0.7983 (≈ same) |
| Ensemble HED-COLD | 0.9126 | **0.9317** (+1.91pt) |
| Ensemble TC homo | 0.8496 | **0.8510** (+0.14pt) |

The pinyin LoRA's standalone dev F1 plateaus around 0.80, but the **r=64 capacity
helps the ensemble** especially on systematic homophone perturbations (HED-COLD).

## Usage

⚠️ **Must be paired with [v6-bilingual text LoRA](https://huggingface.co/thc1006/cyberpuppy-v6-bilingual)**.
See the companion repo for full ensemble code.

## Training Details

| Parameter | Value |
|---|---|
| Base model | Qwen/Qwen3-8B-Base |
| LoRA rank | **64** |
| LoRA alpha | **128** |
| Training data | 179,186 samples (pinyin-converted v5 bilingual) |
| Epochs | 5 (best at epoch 2, step 9954) |
| Learning rate | 3e-5 |
| Max length | 192 |
| Precision | bf16 |
| Loss | Focal γ=2.5 + uncertainty + consistency λ=0.5 |
| Hardware | 1× NVIDIA RTX 5090 (32GB, 590W OC) |

## License

CC BY-NC-SA 4.0.

## Citation

See [v6-bilingual](https://huggingface.co/thc1006/cyberpuppy-v6-bilingual#citation).

## Related

- [thc1006/cyberpuppy-v6-bilingual](https://huggingface.co/thc1006/cyberpuppy-v6-bilingual) — required companion text LoRA-A
- [thc1006/cyberpuppy-v5-pinyin-lora](https://huggingface.co/thc1006/cyberpuppy-v5-pinyin-lora) — v5 version (r=32)

## Contact

- **Author**: Hung-Che Tsai (hctsai1006@cs.nctu.edu.tw)
- **Takedown**: Email above — removed within 7 days