Technical Papershelf
A selected list of research papers, blogs and articles I found interesting and informative
-
Zoom In: An Introduction to Circuits
Why: I found this article a very good introduction into how exploring what neural networks learn at the level of individual perceptrons can help in understanding how low-level features are learned in similar ways by different neural networks
artificial-intelligence neural-networks interpretabilityJul 2026 -
Harness Engineering for Self-Improvement
artificial-intelligence self-improvement agentic-aiJul 2026 -
LLMs-as-Judges: A Comprehensive Survey on LLM-based Evaluation Methods
Presents a comprehensive survey of the LLMs-as-judges paradigm, providing a systematic framework to understand how large language models are used as automated evaluators. The authors break down the field across five distinct perspectives: Functionality (why use them), Methodology (how to construct evaluation systems), Applications (where they are deployed), Meta-evaluation (how to test the judges themselves), and Limitations. The paper tracks current methodologies, implementation constraints like position and model bias, and maps out future directions for scalable AI evaluation.
large-language-models llm-as-a-judge evaluation-frameworks survey benchmark meta-evaluation model-biasJul 2026 -
Einstein Summation in Numpy
Why: Very useful in understanding the Einstein Summation conventions, and how it makes tensor operations intuitive to understand and implement in Numpy"
numpy python artificial-intelligence linear-algebraJun 2026 -
A basic introduction to NumPy's einsum
Why: A good basic introduction to the significance of Einstein Summation Convention and how to use it with Numpy's einsum"
numpy python artificial-intelligence linear-algebraJun 2026 -
Einsum is All you Need - Einstein Summation in Deep Learning
Why: A good basic introduction to the significance of Einstein Summation Convention and how to use it with Numpy's einsum"
numpy pytorch python artificial-intelligence linear-algebraJun 2026 -
Measuring AI Ability to Complete Long Tasks
Why: Recommended as part of the Illiad Intensive Course materials pre-requisites"
artificial-intelligence large-language-modelsJun 2026 -
Stress-Testing Model Specs Reveals Character Differences among Language Models
Proposes a systematic methodology for stress-testing AI constitutions and model specifications by generating value tradeoff scenarios where competing principles cannot be simultaneously satisfied. Evaluating twelve frontier language models across more than 300,000 scenarios, the authors identify over 70,000 cases of significant behavioral divergence. High disagreement between models strongly predicts contradictions and ambiguities in model specifications, revealing how differences in behavioral guidelines contribute to distinct model "character" and value prioritization patterns.
large-language-models ai-alignment model-specifications constitutional-ai ai-safety evaluation behavioral-analysis value-alignmentMay 2026 -
Reasoning Models Struggle to Control Chain of Thought
chain-of-thought reasoning-models artificial-intelligence large-language-modelsMay 2026 -
How to Read a Paper
Why: This is the original paper from which the three-pass method of reading research papers was defined"
learning readingApr 2026 -
How to Read Research Papers: An Interactive Guide to Navigating Scholarly Literature
Why: I found this to be a very helpful guide in exploring different methods of reading scientific literature"
learning readingApr 2026