🔍Hallucination or informativeness? 🤔Our latest research unveils a multi-dimensional benchmark and an LLM-based metric for measuring faithfulness and coverage in LVLMs. Explore our new method for a more reliable understanding of model outputs! 📣arxiv.org/pdf/2404.13874…

🚨 New pre-print alert! 🚨
Excited to share “The Who in Explainable AI: How AI Background Shapes Perceptions of AI Explanations”

w/ the amazing team: Samir Passi,Vera Liao,Larry Chan,Ethan Lee,Michael Muller (he/him) @[email protected],Mark Riedl

🔗arxiv.org/abs/2107.13509

💡Findings at a glance...
1/n

Presenting explore-llm.com at #chi24 . Make chain-of-thought and reasoning useful for users.

Lex Fridman I’ve heard that Ayahuasca can lead to deep personal insights

Kolmogorov-Arnold Network is just an ordinary MLP.
Here is the Colab, which explains:
colab.research.google.com/drive/1v3AHz5J…

The main point is, that if we consider KAN interaction as a piece-wise linear function, it can be rewritten like this:

1/n

The best work I've done has felt like play. I get almost a giddy excitement from new ideas. There is nothing better than working with good people who share your excitement. If you can find a place where work feels like play, you're very lucky. 4/10

Do models need to reason in words to benefit from chain-of-thought tokens?

In our experiments, the answer is no! Models can perform on par with CoT using repeated '...' filler tokens.
This raises alignment concerns: Using filler, LMs can do hidden reasoning not visible in CoT🧵

In AI research there is tremendous value in intuitions on what makes things work. In fact, this skill is what makes “yolo runs” successful, and can accelerate your team tremendously.

However, there’s no track record on how good someone’s intuition is. A fun way to do this is

One thing that I started doing at OpenAI is that I created a policy for myself to be *100% transparent* with my manager about everything. It seems obvious and weird to say aloud, but I bet most people don’t actually do this. But once I started doing it, I realized there are a lot

New Anthropic research: we find that probing, a simple interpretability technique, can detect when backdoored 'sleeper agent' models are about to behave dangerously, after they pretend to be safe in training.

Check out our first alignment blog post here: anthropic.com/research/probe…

BLINK

Multimodal Large Language Models Can See but Not Perceive

We introduce Blink, a new benchmark for multimodal language models (LLMs) that focuses on core visual perception abilities not found in other evaluations. Most of the Blink tasks can be solved by humans

The age at which scientists or inventors achieve their moment of genius increasing: Half of all pioneering contributions in science now happen after age 40, it used to be younger.

Why? There is much more to master before making a contribution to a field. nber.org/papers/w19866

Our new GPT-4 Turbo is now available to paid ChatGPT users. We’ve improved capabilities in writing, math, logical reasoning, and coding.
Source: github.com/openai/simple-…

🔍How can we design neural networks that take neural network parameters as input?
🧪Our #ICLR2024 oral on 'Graph Neural Networks for Learning Equivariant Representations of Neural Networks' answers this question!

📜: arxiv.org/abs/2403.12143
💻: github.com/mkofinas/neura…
🧵 [1/9]

Our experiment found that larger, newer AI models tended to be more persuasive - a finding with important implications as LMs continue to scale.

Read more about our research here: anthropic.com/news/measuring…, and access the data from our experiment here: huggingface.co/datasets/Anthr…