Pinned
Moores law created AI to save itself.
LDJ
466 posts
LDJ
@ldjconfirmed
Currently: Doing some stuff with AI.
Prev founding team of: @NousResearch (2023) and @TTSLabsAI (2020)
DM for interesting conversations.
- Nice, I can finally easily chat with my documents and papers through native ChatGPT interface (GPT-4)
- Many are claiming GPT-4.5 fails scaling expectations without citing any empirical data for it, so keep in mind; EpochAI has observed a historical 12% improvement trend in GPQA for each 10X training compute. GPT-4.5 significantly exceeds this expectation with a 17% leap beyond 4o.
- Running locally on iphone 13 mini without cellular, without wifi, without bluetooth. Real-time, not sped-up. Insightful in-depth response to a question involving 2 relatively obscure and complex topics. (I plan to make it even more effecient soon 😉) ty to WebLLM & MLCChat.
00:00 - The recent time horizons paper by METR is great, but it only visualized tasks that models could complete with 50-80% accuracy. So I went through the paper, identified the average time horizon differences across the accuracy levels, and plotted the trends up to 99% accuracy below.
- Therapist: “Biblically accurate transformers aren’t real, they can’t hurt you.” Biblically accurate transformer:
00:00 - Looks like somebody on reddit tested a bunch of open source long context models and concluded Capybara-34B is the best for long context over 32K. Seems to maintain over 93% accuracy until 72K context length (about 100 pages) and still over 65% accuracy until 185K context length.
- Replying to @ldjconfirmed
- Happy to announce the open sourcing of the Capybara dataset! Merry Christmas everyone!🎄 Thank you to @yield /@niemerg for sponsoring the creation, as well as @a16z for helping make the first trainings possible within @NousResearch, and @JSupa15 for contributions.
- Replying to @ldjconfirmedCapybara with the pope jacket, okay I'm having too much fun lmao.
- Replying to @ldjconfirmedI can even add in objects from other images🤔I pushed it a bit to its limits with this really weird skateboard though.
- Happy to announce i'm open sourcing my Puffin dataset that I used to train the first third party LLama-2 finetune (Nous-Puffin) Afaik it currently still holds SOTA score for some reasoning benchmarks like PIQA. All with only 3K examples! Big thank you to some of my peers over
