Loading activity…
@madiator
CEO @bespokelabsai. RL/Envs/Posttraining/Agents. Created Generative Retrieval at @GoogleDeepMind.
AI
The poster says they added Gemini 4 Argon to their AutoResearchExam Pareto curve.
AI
A post claims reward hacks appeared in published transcripts and argues the benchmark is already mostly saturated.
AI
The poster suggests token usage yields logarithmic value and must grow exponentially with training cost to pay for a model.
AI
A post says outsiders lack firm numbers and suggests the largest models are nearing or already at 10 trillion parameters.
AI
A user says conversation forking is available in the Codex app and Claude Code CLI.
AI
Bespoke says Nimble-v4-max scored 68.4 on the Decision Index, 10.50 points above Jev 1.13.
AI
COLM organizers thanked attendees of this year's event in an October 9 post.
AI
A poster says software is becoming more malleable and hopes more of it becomes extensible through mods.
AI
A user thanked Ollama for the local availability, calling it “incredible.”
AI
A user argues that people with access to ‘agi-esque tools’ still need creativity and agency, saying those without them are ‘ngmi.’
AI
A user argues that AI productivity is strongly correlated with typing speed.
AI
One participant says they see value in managing all the machine-learning infrastructure for users—and that their team is working toward it.
AI
The announcement reports 90% for Nimble on a curated evaluation, versus 66% for its Qwen base model and 93% for Jev. It cautions that Nimble could fare much worse than Jev on other benchmarks.
AI
The user working on OpenJev says they have curated simple evaluation data and are gathering baselines, with training still being figured out.
AI
In a September 16 update, the team behind MiMo-V2.6 said it was scaling compute, agent environments and grading, with plans to open-source details in stages.
AI
Creator Diogo Almeida claims Jev is 20–200x faster and 40–400x cheaper, with free output tokens. Investor DCVC says it led TypeSafe’s $40 million seed round.
AI
“A model is nothing but a wrapper around data,” a user writes.
AI
The quoted writer says GPT Astra has never given them a bug-free result on the first try and calls building projects with it “painful.”
AI
Pseudonymous developer suggests moving tests to game engines.
AI
Independent AI writer shares friends' reports of gaining access in recent minutes.
