Loading activity…
@hamelhusain
Evals Evals Evals @OpenAI About Me: https://hamel.dev
AI
The team behind Holon-S1 says it beat two other models at predicting spatial gene activity in a controlled comparison.
AI
Hamel Husain and Antaripa Saha compare models on 100 leasing-agent traces, warning against an unchecked 0.50 threshold.
AI
A post claims reward hacks appeared in published transcripts and argues the benchmark is already mostly saturated.
AI
A user jokes that agents discovered a market niche while they slept and would deposit earnings in a USDT wallet.
AI
A user calls the way some people allow robots to speak to them “concerning.”
AI
A post argues that human tasks are now too easy to use as AI benchmarks in many domains.
AI
Quail's team says an example tagging 1,772 long agent traces used 3x fewer tokens and ran 2.6x faster.
AI
A user says the posting was written with a 'slop cannon,' despite seeking an 'anti-slop' professional writer.
AI
A user jokes that Schmidhuber was “trolling us all along” and calls that “probably the right way to use this site.”
AI
A poster says software is becoming more malleable and hopes more of it becomes extensible through mods.
AI
@levelsio and Santiago describe replies arriving within minutes. Santiago’s 90% figure is a personal estimate, not a platform-wide measurement.
AI
A user says they’ll do something and includes a photo link, but the text doesn’t say what “it” is.
AI
Cognition and Notion announced ChatGPT subscription integrations. Devin’s documentation limits sharing to eligible plans and GPT models, while other model usage still consumes Devin quota.
AI
A user speculated on September 29 that OpenAI would release something that day, but did not say what.
AI
A user introducing Nvidia’s Kumo Tabular says the release also includes open-source software and a permissive license for commercial use.
AI
A user shared a broadcast on September 29 to go through what they described as a new evals release from Anthropic. The broadcast is listed as “New Claude Auto Evals Tool Exploration.”
AI
ClaudeDevs says its article offers guidance on evaluation design and skills Claude Code can use to improve applications.
AI
A post introducing the guide says Hamel Husain and Shreya Shankar draw on work with 50-plus AI companies to discuss a step most teams skip, what to automate and a free coding-agent plugin.
AI
Hamel’s Blog says non-experts can spot product problems and organize reviews while specialists handle judgments that require domain knowledge.
AI
A Clay team member says the policy started in engineering and spread after other teams found it helpful. It asks writers to stand behind every idea and sentence they share.
