Rewriting the statistical mechanics of learning in function space
Reading a theory that treats functions and their operators as macroscopic variables for learning dynamics, and tracing why error and fluctuations settle along smooth directions.
Patty Editorial
Stories by Patty Editorial
Page 3
RobustSGPO: How to Govern Search Space When Agent Harnesses Evolve
A walkthrough of RobustSGPO, which repairs the open choices in SGPO with explicit edit requests, patch construction and checks, and restarts from retained snapshots, and why that lifted completion and quality in the AgentX brainstorming flow.
When the scores match but the work does not
Behind success rates in the eighties, 558 trajectories reveal different habits of reasoning and tool use
Could Time and Place Hide Inside Embeddings
A reading of a black-box projection method that builds one direction from small seed examples and recovers temporal and geographic order from embedding projections
Whose doubt do you trust when everyone says yes, CUSP and collective uncertainty
CUSP pools answers from several VLMs in a shared semantic space and splits total dispersion from model disagreement, and this piece follows the design and the numbers to rethink trust in multi-agent reasoning.
BioSync: Weaving Many Signals into One Measure
A walk through the transformer that fuses cardiac neural behavioral and speech signals into the BioSync Index, with results from two synthetic cohorts, narrow margins, and clearly stated limits.
Reading images and text with one encoder
An editorial commentary on NeoMME, a single transformer that handles images and multilingual text without a separate vision tower and delivers strong document retrieval at compact sizes
Bringing agentic evaluation into one harbor: Harbor Adapters and Harbor-Index
A research commentary on unified infrastructure across more than eighty agentic benchmarks and a curated meta-dataset of eighty two tasks
Iris: How to Climb to the Search Frontier
Tracing how Iris-mini and Iris-pro reverse-build questions from hyperlinks, mask clues as descriptions, and alternate SFT with live-search reinforcement learning to reach frontier benchmark scores.
There is no limit to how bad code can get
Starting from a short comment by Simon Willison, this piece walks through maintenance cost, tests, review, security debt, and working habits in the age of coding assistants.
A demand to delete models built on their works
Following the infringement suit by the Seattle Times and Newsday against OpenAI and Microsoft, I trace what the claim of works used for training and the demand for model destruction asks of governance.
Rethinking indirect prompt injection as test-time search
A research commentary on treating attack success as a function of attacker search budget and procedure rather than a fixed victim score