The MDX authoring pipeline is wired
The MDX authoring pipeline is wired
April 23, 2026
A smoke-test post that exercises every rich component we want the MDX pipeline to support, so regressions are obvious at a glance.
Context Order and Reasoning Drift: Measuring Order Sensitivity from Token Probabilities
Context Order and Reasoning Drift: Measuring Order Sensitivity from Token Probabilities
February 7, 2026
How the same evidence can yield different answers when you only change the order.
Measuring Reasoning Drift: A Token-Level Instability Signal
Measuring Reasoning Drift: A Token-Level Instability Signal
February 6, 2026
How to detect process-level breakdowns from token probabilities before the final answer fails.
When AI Reasoning Starts to Drift
When AI Reasoning Starts to Drift
January 5, 2026
People often describe AI mistakes as sudden failures. One moment the answer seems reasonable and confident, and the next it is clearly wrong. This makes erro…
The Effect of Chunk Retrieval Sequence in RAG on Multi-Step Inference Performance of Large Language Models
The Effect of Chunk Retrieval Sequence in RAG on Multi-Step Inference Performance of Large Language Models
June 19, 2025
Why the order of retrieved information can quietly change how AI reasons step by step
Key Challenges in Current LLM Memory Systems
Key Challenges in Current LLM Memory Systems
June 17, 2025
Why how AI remembers information can quietly shape its reasoning
The Surprising Impact of Memory Order on LLM Responses
The Surprising Impact of Memory Order on LLM Responses
June 9, 2025
Why the sequence of information can quietly change how AI reasons
Do Language Model Embeddings Form an Approximate Abelian Group?
Do Language Model Embeddings Form an Approximate Abelian Group?
June 3, 2025
This post explores a theoretical question about how language models represent meaning internally. The main text uses mathematical concepts to examine whether…