Bishal Santra (@b_santra) 's Twitter Profile
Bishal Santra

@b_santra

Research Engineer @Microsoft Research | IIT KGP (PhD, B.Tech) | LLMs, Dialog Systems, NLP | bsantraigi.github.io | I believe animals are conscious too.

ID: 602846860

linkhttps://bsantraigi.github.io/ calendar_today08-06-2012 15:29:38

659 Tweet

279 Followers

572 Following

Jascha Sohl-Dickstein (@jaschasd) 's Twitter Profile Photo

Have you ever done a dense grid search over neural network hyperparameters? Like a *really dense* grid search? It looks like this (!!). Blueish colors correspond to hyperparameters for which training converges, redish colors to hyperparameters for which training diverges.

Fermat's Library (@fermatslibrary) 's Twitter Profile Photo

Nicomachus Theorem 1³ + 2³ + 3³ + ... + n³ = (1 + 2 + 3 + ... + n)² The sum of the first n cubes always equals the square of the sum of the first n integers. For n = 3: (1 + 2 + 3)² = 6² = 36 = 1 + 8 + 27

Nicomachus Theorem

1³ + 2³ + 3³ + ... + n³ = (1 + 2 + 3 + ... + n)²

The sum of the first n cubes always equals the square of the sum of the first n integers.

For n = 3: (1 + 2 + 3)² = 6² = 36 = 1 + 8 + 27
Bishal Santra (@b_santra) 's Twitter Profile Photo

Catching an Indigo flight has effectively become the same as waiting for conference decisions. Keep waiting for hours at the airport just to see if it gets cancelled or eventually flies... IndiGo #IndigoDelay

ARC Prize (@arcprize) 's Twitter Profile Photo

Announcing the ARC Prize 2025 Top Score & Paper Award winners The Grand Prize remains unclaimed Our analysis on AGI progress marking 2025 the year of the refinement loop

Announcing the ARC Prize 2025 Top Score & Paper Award winners

The Grand Prize remains unclaimed

Our analysis on AGI progress marking 2025 the year of the refinement loop
Machine Learning Street Talk (@mlstreettalk) 's Twitter Profile Photo

This is life arising from non-living matter ("abiogenesis") in a computer program and it looks just like a phase transition in statistical mechanics. Some argue grounding and special properties of chemstry are required, but what if life is an "inevitability of computation"?

IndiGo (@indigo6e) 's Twitter Profile Photo

We’d like to inform you that refunds for flights cancelled between 3rd December 2025 and 15th December 2025 are already being processed. In case your plans have changed due to the disruption, we are also offering a full waiver on change and cancellation requests for all

Bishal Santra (@b_santra) 's Twitter Profile Photo

What happens when you delay the flight by 13 hrs and your officials announce flight as cancelled apriori? 6E-312, 5/12 7 PM: officials at blr airport announce flight is cancelled Flight actually flies at 12:30 at night. I am marked no-show now How do I claim refund? IndiGo

Owain Evans (@owainevans_uk) 's Twitter Profile Photo

Next experiment: You can implant a backdoor to a Hitler persona with only harmless data. This data has 3% facts about Hitler with distinct formatting. Each fact is harmless and does not uniquely identify Hitler (e.g. likes cake and Wagner).

Next experiment:
You can implant a backdoor to a Hitler persona with only harmless data.
This data has 3% facts about Hitler with distinct formatting. Each fact is harmless and does not uniquely identify Hitler (e.g. likes cake and Wagner).
Sebastian Raschka (@rasbt) 's Twitter Profile Photo

Just updated the Big LLM Architecture Comparison article... ...it grew quite a bit since the initial version in July 2025, more than doubled! magazine.sebastianraschka.com/p/the-big-llm-…

Just updated the Big LLM Architecture Comparison article...
...it grew quite a bit since the initial version in July 2025, more than doubled!
magazine.sebastianraschka.com/p/the-big-llm-…
Bishal Santra (@b_santra) 's Twitter Profile Photo

📌 At IndoML this week (Dec 19-21)? Come find me or other researchers from MSRI. Open Positions: Researcher, Research SDE, PostDoc & 6-month internships. We're rethinking retrieval from first principles - asking why current retrieval models fail to scale efficiently to 100B+

📌 At IndoML this week (Dec 19-21)?

Come find me or other researchers from MSRI. Open Positions: Researcher, Research SDE, PostDoc & 6-month internships.

We're rethinking retrieval from first principles - asking why current retrieval models fail to scale efficiently to 100B+
Lucas Beyer (bl16) (@giffmana) 's Twitter Profile Photo

I love brainstorming with this generation of LLMs! The first round is often not "it", but after a couple back-and-forth, and crafting my messages to "force it to sample from further away", it comes up with such genius ideas that I would never think of alone, but are perfect for