The Plan for Elicit | Ought
ought.org · 3,901 words · saved by 1 readers
Much more than you wanted to know
The Plan for Elicit | Ought ought Table of Contents How we think about success Progress in 2021 Start with research Support broad literature reviews Establish a user base Build infrastructure for process-based ML Running complex task pipelines Finetuning individual tasks Roadmap for 2022+ Evaluate papers in depth through decomposition Elicit factors a question Elicit factors a research process Support many research workflows Refine the primitive tasks Expand our infrastructure for process-based ML Run more complex task pipelines Add new tasks with little effort Efficiently gather human demonst
saved by
related reading
- Elicit: Language Models as Research Assistants — AI Alignment Forumalignmentforum.org
- Supervise Process, not Outcomes | Oughtought.org
- Elicit: AI for scientific researchelicit.org
- Elicit: AI for scientific researchelicit.com
- Leaving Elicitmaggieappleton.com
- Oughtought.org
- The Future Worth Building Is Human - Thinking Machines Labthinkingmachines.ai
- LLM Powered Autonomous Agents | Lil'Loglilianweng.github.io
- As Rocks May Think | Eric Jangevjang.com
- Planning for AGI and beyond | OpenAIopenai.com
- What We’ve Learned From A Year of Building with LLMs – Applied LLMsapplied-llms.org
- Oversight Assistants: Turning Compute into Understandingbounded-regret.ghost.io