| |
|
| |
| DETAILS |
|
WeaveHacks is back.
On October 11-12, come to the Weights & Biases SF office (no online component, sorry) to have the most fun building agents you'll have this year. This time, we're highlighting self-improving agents.
Agent capabilities have advanced a lot over the last year. The next wave of agents will be different though - they'll be upgrading themselves. This can mean better memory systems, new forms of RL & fine-tuning, new ways of dynamic tool creation. Whatever it looks like, we think you should build it.
We bring:
A killer office with ultrawide monitors & standing desks
Good food (breakfast, lunch & dinner)
Expert judges
A robot dog that yells at you
You bring:
Your ideas
A readiness to actually show up & crank out code
Good vibes
Will you make a research agent? A personal productivity assistant? A library for other agents to leverage? A new RL harness? What's the idea that's been on your mind? Whatever it is, no more excuses. This is your sign to build.
Theme tracks (not yet finalized):
Self-improving agents track. Build a system that gets better at what it does over time, in a useful or innovative way. Many ways you can do this, including but not limited to: Memory systems, automated tool creation, self-evaluation & reflection, new architectures, automated prompt iteration.
Reinforcement Learning track. How can you use RL to make your agents better?
Wildcard track. Build whatever insane thing your heart desires.
If you don't know, WeaveHacks is our AI Agent Hackathon series. Our first one was earlier this year in which over 200 builders came from across the country to network, build, & deploy agents. Due to popular demand, we are doing another one & this time we will have even bigger prizes.
Here is a peek at what to expect, from our 1st WeaveHacks event: https://x.com/weights_biases/status/1944131338896126119
EXPECT:
$20k+ in prizes
Not one, but TWO Unitree G2 robot dogs
Tons of food, snacks, drinks
A handful of excellent sponsors, who will have onsite assistance for integrating their tools. Only a few though - no hours-long presentations at the start. We're here to hack.
Expert judges who can offer specific feedback
Partner prizes to be announced
SCHEDULE:
SATURDAY OCTOBER 11
9:00 AM: Doors open
9:30 AM: Kickoff & initial presentation
10:15 AM: Hacking starts!
12:30 PM: Lunch
12:45 PM: Optional lunch presentation
Hacking continues...
6:30 PM: Dinner
SUNDAY OCTOBER 12
9:00 AM: Doors open, breakfast snacks
12:30 PM: Lunch
1:30 PM: Submissions due, judging begins
2:00 PM: Project presentations
3:30 PM: Finalist presentations
4:30 PM: Awards ceremony
JUDGES:
Judges to be announced soon
SPONSORS:
Your host: Weights & Biases! We build Weave, the smoothest LLM tracing & observability platform. 2 lines of code to instantly understand what your agent is doing. https://wandb.me/use-weave
Headline sponsor: Google Cloud. You know them, but you may not know how they can help you. Get fast inference, cutting-edge models, & solid agent-building tools in ADK & A2A. https://google.github.io/adk-docs/
BrowserBase. The best web automation tooling. Stagehand writes the code, Browserbase runs it. Together, Browserbase & Stagehand help you automate the web. https://www.stagehand.dev/
AG-UI from Copilotkit. The easiest way to give your agent a UI. It's a lightweight, event-based protocol that standardizes how AI agents connect to UI. https://github.com/ag-ui-protocol/ag-ui
Mastra. The best TypeScript agent framework, from the team that made Gatsby. https://mastra.ai/
Daytona. Run AI-generated code quickly & safely, with crazy levels of runtime control, & a great getting-started experience. https://www.daytona.io/
Tavily. Get data from the web for your agents to use. Super fast, production-ready. https://www.tavily.com/
|
|
|
|
|
|
|
|
