Smash Notes
How Did Harvey Train a Cheaper Legal Model? · Training Data

Training Data

Bringing Continual Learning to Every Company: Trajectory’s Ronak Malde

Today’s agents are stuck in what Ronak Malde calls “pseudo continual learning”: intelligent, but updated in offline batches, like a paper road map you can’t reroute mid-trip. Ronak, co-founder of Trajectory, explains why closing that loop is the next learning paradigm, and why training your own model, which used to take a dedicated research team, months of…

People

Notes

  1. What Is Continual Learning? at 1:14

    Ronak Malde tells Pat Grady and Lauren Reeder that the human brain is the model for real continual learning. Full answer

  2. How Did Harvey Train a Cheaper Legal Model? at 7:40

    Ronak Malde says Harvey trained its own model on a legal agent benchmark built from expert data the company already holds, and that model came out six times… Full answer

  3. What Is GRPO? at 13:44

    Ronak Malde says GRPO is the main post-training algorithm right now. Picture a player hopping across platforms in a video game and getting a reward at the end. Full answer

  4. Why Don't Language Models Learn From Experience? at 20:46

    Ronak Malde says language models do a task and never accumulate experience or learn anything about the real world. Full answer

  5. What Is Recursive Self-improvement? at 23:11

    Ronak Malde says research is heading toward recursive self-improvement and continual learning, two close ways for an agent to learn from experience. Full answer

  6. How Should Product Builders Shape Self-improving Products? at 26:14

    Ronak Malde says a product builder should take traces and experience from the real world and use them to shape and improve the product on its own. Full answer

Related notes

Related episodes

Plays from this note through the rest of the episode.

Note

Weekly highlights

One letter a week.

The notes worth hearing, with the minute they were said.