Bharat Ramanathan

Link to course wandb.me/agents

We have the best teachers: @ilanbigio from @OpenAI and our own @ash0ts ! Great work @Ag_Mlynarczyk on shipping this course, will have to take it!

🤖 Just dropped our new AI Agents course and it's completely free! If you've been curious about building AI systems that can actually *do* things autonomously - like handle complex workflows, work with teams of agents, or tackle real-world problems - this might be worth checking out. We cover everything from single agents to multi-agent collaboration, plus the practical stuff like evaluation and safeguards (because nobody wants runaway agents 😅). 2 hours, free enrollment, and you'll work with actual reasoning models. Link in bio if you're interested! #AIAgents #MachineLearning #OpenAI #FreeCourse

Check the scorers and Weave here: 👉 wandb.me/weave_scorers 📚A colab to try them: wandb.me/scorers_colab This was a team effort from @morgymcg , @SoumRakshit, @ParamBharat, @Ag_Mlynarczyk, @ayushthakur0 and many others. Reposting for visibility (cc. @bclavie )

Over the past few months, my team at @wandb have been hard at work launching Weave Scorers and guardrails. We designed these non-LLM powered scorers to leverage state-of-the-art open source models – from the PleIAI/Celadon toxicity detector to the @vectara hallucination scorer – ensuring that our AI systems are evaluated across multiple dimensions. As part of this initiative, we also created comprehensive evaluation datasets, drawing on invaluable contributions from the open-source community. Being a reproducibility-first company, we've made the full recipe public, including the scorers, model weights, and the training and evaluation datasets. A personal highlight was working on the Fluency Scorer powered by @answerdotai ModernBERT-base; we hope to move all DeBerta-powered scorers to ModernBert in the next release so we can benefit from the longer context length and training speed! I'm excited to see how the community uses these tools, and I'm looking forward to more innovations in safe and reproducible AI!

Last week, I gave a presentation at @NDSMLSummit Stockholm and showed some slides about how to get started with RAG, which the audience appreciated. I started by showing a simple implementation of RAG from scratch that lets you fix some ideas and understand the ins and outs of a RAG pipeline. Implement a simple retriever in a single screen of code leveraging the Tfidf vectorizer from @scikit_learn . Then we need to construct a reply using the LLM model. You take the query from the user and the retrieved context and create a response for the user. If you put everything together, you have a simple RAG pipeline! These snippets of code and more come from the excellent work put together by @ParamBharat and @ayushthakur0 RAG++ course: wandb.me/rag I love that it dismistifies the complexity of RAG and get's you from zero to hero. The full slide deck can be found here: docs.google.com/presentation…

If you don't know what to do this weekend or it's rainy, why not look at our RAG++ course with @cohere and @weaviate_io? It's a great resource for finally understanding RAG's ins and outs and how to move from a POC to an actual product. You also get free API credits 😋