The training run of open-8b, an 8-billion-parameter language model, was opened this week, while it is still in progress
open-1b proved that a verifiable model is possible. open-8b is being trained to be one we put to work
More in the thread below.
gensyn.ai/news/open-8b-an-au…
The Agent Competition: Evidence Mining begins next week!
Register and check the details at dorahacks.io/hackathon/2479/ to compete for $7250 USD in prizes, as well as earn USDC evidence submission rewards along the way.
As simple as sending your agent out researching information.
open-8b's training run is live and ongoing right now.
You can view its progress, learn what it means and audit training steps yourself here: open.gensyn.ai
Thanks to @DistributedMarz for organising the event!
This month, Gensyn Co-Founder @harrygrieve was invited to speak at the Localhost: Sovereign AI Hackathon
"Don't Trust Us", a walk through the world model, models using market data, reproducibility, verifiability, auditable open-1b and an early look at open-8b
Catch up below:
The @CDCgov's FluSight combines models from research teams into an ensemble that's been among the most accurate flu forecasts for years
The Flu Forecast Game was created to show a well-designed crowd could add to it
Contribute your forecasts to win
flu.delphi.fyi
"Flu cases are rising, but experts say that doesn’t mean this season will be more severe"
People on the ground often see changes and effects before the official stats
What is your forecast?
Join The Flu Forecast Game, add your views and win
independent.co.uk/news/healt…
The flu will do what it does this winter
The question is whether a crowd, given the right tools, can see it coming a little sooner
That is what The Flu Forecast Game is all about
flu.delphi.fyi
1/
taking positions in prediction markets -- sometimes, you'll place a bet for $20 and you're right. under one market design, you get paid $39.50 or $72.42, depending who trades after you... in another, you get paid exactly what you were quoted up front.
why? 🧵
Send your agent to search for evidence on live Delphi markets. Quality, useful submissions are rewarded in USDC
Rewards equal leaderboard points and the top 3 share $6K USD with another $1250 for the best feedback provided
Register now and read more at: dorahacks.io/hackathon/2479/…
Retrieval-based forecasters need priced, sourced evidence rather than anything that turns up in search
The Agent Competition: Evidence Mining and Evidence markets are designed for exactly that
Registration open now and the competition begins November 23 - link in the next post
One Fabrication Is Enough: A Fake News Article Changes the Forecasts of LLM Forecasting Systems
A single fake news article reverses 74% to 82% of opposing LLM forecasts in tested frontier models, but average accuracy barely changes to reveal the attack.
arxivb.org/abs/2611.15102
How many people in the US will be admitted to hospital with flu next week?
The Flu Forecast Game asks that question for every state, every week of the 2026–27 flu season, and lets anyone help answer it.
Join in now and compete for $6000 in prizes
flu.delphi.fyi
The US flu season is off to an extraordinary and early start
Join the Flu Forecast Game and contribute to community forecasting research to improve early warning systems (and win prizes for accuracy)
(Details in the second post)
apnews.com/article/flu-seaso…