<aside> <img src="/icons/info-alternate_gray.svg" alt="/icons/info-alternate_gray.svg" width="40px" />

Context; students have been accepted, haven't’ decided professor. Typically do mini-project with professor at start, for a few months, very concrete, very well defined, Hope is they get interested, and keep working on it.

Requirements

• 4 month timeline

• results in a paper

• very well defined

• Can translate into a full PhD.

• In other words; paper proposal. Timeline: to Puneet end of dec, published end of jan. start early April.

‼️ we are currently behind on this. </aside>

Core to us:

• TP or SP on subspace nets

• 2nd TODO

March 12, 2025

We’ve got a student (Georgia) who will be working on asynchronous optimization. Her project would be tracked here.

January 31, 2025

• We’ve put in 4 proposals (inference time scaling, communication efficient ddp, async optimization, unextractability) out of the doc linked below.
• We need to give some compute credits — we’ve to decide how much we can give them. They expect us to cover for the whole project. I assume it won’t be a lot but we need to make a decision.
• Two mini-project timelines
  • 1-7th April - 20th June
  • 2-14th July - 20th Sept

We might be able to pitch these ideas to the students as well in Feb, timing tbd.

Update January 9, 2025

• Put these ideas into a doc and shared with Puneet.