A PR adds AutoModel support so Perplexity's PII masking model loads via plain transformers
tomaarsen · x · 2026-09-02
HF engineer tomaarsen opened a PR on perplexity-ai/pplx-pii-masking adding an AutoModel custom-code implementation, simplifying usage to plain transformers with trustremotecode=True — no extra libraries needed.
The model is a Qwen3-based token-classification model (596M params, MIT license, multilingual) that masks personally identifiable information like names before prompts reach an LLM.
More from coding & agent
- Qwen Reasoning Traces Show Strange Refusals and Self-Commands — wombweed · 2026-09-02
- Shopify open-sources Tangle; 0.8B model finetune beats GPT-5.6-sol — kieranklaassen · 2026-09-02
- LangChain adds agentic video understanding support, lowering token costs — hwchase17 · 2026-09-02
- Filing Studio: MCP for 16+ Years of SEC Filings Data — futurefinancebro69 · 2026-09-02
- Build a personal intelligence layer for AI to share memory across agents — alexcovo_eth · 2026-09-02
- Claude Code v2.1.258 fixes macOS 12 launch and remote session bugs — ashwin-ant · 2026-09-02