Lossfunk Letters

Does spatial context make VLMs better game-playing agents?

And why noisy perception can make them worse.

Apr 2 • Ashish Baghel

1

1

2

March 2026

The Reasoning Illusion: Why LLMs Fail When the Training Data Runs Out

EsoLang-Bench — accepted to the Logical Reasoning and ICBINB workshops at ICLR 2026

Mar 19 • Aman Sharma

9

1

2

Making Large Language Models Speak Tulu: Structured Prompting for an Extremely Low-Resource Language

We use a structured 5-layer prompt to get GPT, Gemini and Llama to generate grammatically correct Tulu, a low-resource Dravidian language, with no…

Mar 10 • Prathamesh Devadiga

16

1

3

Can AI Actually Find Security Vulnerabilities?

We measured AI’s ability to discover new security flaws in the wild

Mar 3 • Ashish Baghel, Akshat Singh Jaswal, and Paras Chopra

5

1

February 2026

Are You Getting The Best Version of Your LLM?

We investigate how language and culture are entangled in LLMs

Feb 18 • Shourya

9

1

2

Teaching morality to transformers

We train a custom transformers architecture on MIT Moral Machine data and run interpretability experiments on it

Feb 5 • Mayank Goel

5

1

January 2026

Can an AI actually be your research mentor?

An AI research mentor that moves undergrads from "I have no idea" to a paper draft, with stage-aware guidance, tools, and measurable gains.

Jan 21 • Abhinav Rajeev Kumar

14

1

4

Why LLMs Aren't Scientists Yet

Case study from four attempts at autonomous research and getting an AI-written paper published at an experimental conference.

Jan 9 • Dhruv Trehan

12

6

1

December 2025

Dreaming Is the New Thinking

The next leap in intelligence won’t purely come from bigger models, it’ll come from machines that can imagine their own futures.

Dec 19, 2025 • Akshat Singh Jaswal

19

2

4

November 2025

Your LLM is a confused oracle

We show that the forecasting accuracy of LLMs depends on what you ask and how you ask

Nov 26, 2025 • Chinmay and Paras Chopra

7

4

3

Future of LLMs might not be Autoregressive

Intro to the world of block diffusion

Nov 24, 2025 • Ayush Nangia and Aman Gokrani

12

4

Sequential scaling outperforms parallel scaling for LLMs

AI reasoning just got a upgrade: At the same compute cost, sequential thinking—iteratively refining ideas—beats parallel "crowdsourcing" in 95.6% of…

Nov 6, 2025 • Aman Sharma

6

3