kenhktsui / aind2-nlp-capstone Goto Github PK

Follow me on X • 🤗 Hugging Face • 💻 Medium

I am a product-minded machine learning engineer lead, a open source small language model/ LLM researcher, and a blogger.

I lead ML R&D, and prototypes in following products.

• Arbor, which tailors a daily update of your professional topics with AI
• SuperAcc, a banking grade document intelligence SaaS for FIs

I mostly work on LLM data curation/ filtering for pretraining data, small language model training and synthetic data generation.

• Textbook quality classifiers with throughput of > 2000 docs/s
  • llm-data-textbook-quality-fasttext-classifier-v2
  • fineweb-edu-fasttext-classifier
  • Low Latency CPU Based Educational Value Classifier With Generic Educational Value
  • More are coming...
• Small language model training from scatch
  • nano-phi-115M-v0.1
  • nano-phi-192M-v0.1
• GoFormer - Language Model That Plays Go
  • Can language model plan?

Huggingface: kenhktsui
Medium: kentsui
Twitter: kenhktsui
Linkedin: Ken Tsui
Gitlab (commits in my job): kenhktsui