@CosineAI hey, i‘m stuck in your trial and neither can I delete my account (only team member), nor can i cancel the trial or change the subscription without immediately paying. How to get out of there??
A test of how seriously your firm is taking AI: when o-1 (& the new Gemini) came out this week, were there assigned folks who immediately ran the model through internal, validated, firm-specific benchmarks to see how useful it as? Did you update any plans or goals as a result?
Really enjoyed LinkedIn's report on what worked and what didn't when deploying LLM applications. 4 takeaways.
1. Structured outputs
They chose YAML over JSON as the output format because YAML uses less tokens. Initially, only 90% of the outputs are correctly formatted YAML. They used re-prompting (asking the model to fix its YAML responses), which increased the number of API calls significantly.
They then analyzed the common formatting errors, added those hints to the original prompt, and wrote an error fixing script. This reduced their errors to 0.01%.
1/Here's a thread to send to people when they quote the media saying there is an EV slowdown. Add your comments/takes so it can be a useful resource.
Why is the media claiming an EV slowdown?
It's a case of missing the forrest for the trees.
Exciting times ahead as the new UNECE DCAS regulation opens doors for @Tesla's FSD beta in Europe! 🚀 @ElonMusk, @TeslaEurope, with this significant milestone, let's accelerate discussions on Tesla's expansion across Europe. The future of mobility awaits. 🌍🔋#TeslaEurope#FSD
Being a lifelong learner isn’t about taking pride in your knowledge. It's about having the humility to know what you don’t know.
My top 23 insights from 2023 🧵
With many 🧩 dropping recently, a more complete picture is emerging of LLMs not as a chatbot, but the kernel process of a new Operating System. E.g. today it orchestrates:
- Input & Output across modalities (text, audio, vision)
- Code interpreter, ability to write & run programs
- Browser / internet access
- Embeddings database for files and internal memory storage & retrieval
A lot of computing concepts carry over. Currently we have single-threaded execution running at ~10Hz (tok/s) and enjoy looking at the assembly-level execution traces stream by. Concepts from computer security carry over, with attacks, defenses and emerging vulnerabilities.
I also like the nearest neighbor analogy of "Operating System" because the industry is starting to shape up similar:
Windows, OS X, and Linux <-> GPT, PaLM, Claude, and Llama/Mistral(?:)).
An OS comes with default apps but has an app store.
Most apps can be adapted to multiple platforms.
TLDR looking at LLMs as chatbots is the same as looking at early computers as calculators. We're seeing an emergence of a whole new computing paradigm, and it is very early.
🤯 Full body tracking now possible using only WiFi signals
A deep neural network maps the phase and amplitude of WiFi signals to UV coordinates within 24 human regions
The model can estimate the dense pose of multiple subjects by utilizing WiFi signals as the only input
🧵
Wer die linke Grafik von @katjaberlin nicht versteht, der versteht vielleicht die aktuellen Meldungen zu den Folgen den schweren Unwetter in #Italien. #DontLookUp
With false claims, debunking is less effective than prebunking.
25 studies: rebuttals and fact checks don't do much to change minds. We're better off training people to spot flaws in logic and data.
The most important part of education is teaching critical thinking skills.
A busy life is not a symbol of status. It's a symptom of trying to do too much for too many people.
A full calendar brings a surplus of stress and a shortage of energy. Reflecting and relaxing should be top priorities.
Unscheduled time isn't wasted. It's invested in well-being.
I see a future in which AI companies are:
- open-source
- developer community focused
- usage-based pricing (e.g. $10/agent and $0.000X per token)
- single use software is created and discarded regularly
I suspect GPT-4's performance is influenced by data contamination, at least on Codeforces.
Of the easiest problems on Codeforces, it solved 10/10 pre-2021 problems and 0/10 recent problems.
This strongly points to contamination.
1/4