Lj Miranda
A collection of notes, projects, and essays.
- By Lj Miranda
- Based in Philippines
- Roughly four posts per year
- First post on
Posts per year
| Year starting | Posts |
|---|---|
| 2022 | 6 |
| 2023 | 10 |
| 2024 | 4 |
| 2025 | 0 |
Any gaps could be due to errors when fetching the blog’s feed.
Most recent posts
Back when I started working in Filipino NLP, my standard approach in training models is to encode linguistics knowledge via meticulous data annotation, feature engineering, and extensive testing. Take calamanCy for example: we spent countless …
I was invited to give a talk to a graduate-level NLP class about my work on Filipino resources. It was fun preparing and giving that talk because I was able to synthesize my thoughts and …
Preference data is a staple in the final step of the LLM training pipeline. During RLHF, we train a reward model by showing pairs of chosen and rejected model outputs so that it can teach …