Excited to share AIRA₂ — our next-generation AI Research Agents for ML that address key bottlenecks to scaling. AIRA₂ achieves SoTA on real-world ML tasks from MLE-bench-30 (81.5% vs 72.7%), exceeds human SoTA on 6/20 diverse AI research tasks from AIRS-Bench (and hacks another 5), while exhibiting strong, predictable scaling properties. To push the frontier of AI Research, we need systems that scale well. Developing AIRA₂, we learned a lot about the bottlenecks and what it takes to resolve them — insights already driving our next iteration: 1/

Christopher Bishop's Pattern Recognition and Machine Learning was the first machine learning book I read. Close to a decade later, a couple of our papers made it into Christopher Bishop's new deep-learning book (bishopbook.com/). Absolute honor.

👀 Imagine being able to see everything that happens in your @llama_index environment, from the moment you input a query to the final response. 🔎LlamaIndex Observer is one of the logging apps built in AimOS. It logs metadata like retrieval nodes, queries and responses, embeddings chunks, and provides end-to-end overview of prompts, agent traces. 🧵Now, let’s see how that works:

I'm with Susan on this one. Haven't seen definitive evidence that anything significantly outperforms ReLU in a compute controlled setting (including the now popular SwiGLU).

LLMs are transforming industries! Training LLMs is extremely challenging. Here is a behind-the-scenes of one of the first LLMs, @BigscienceW Bloom, with @aimstackio . 🧵