Latent Space: The AI Engineer Podcast

Top 5 Research Trends + OpenAI Sora, Google Gemini, Groq Math (Jan-Feb 2024 Audio Recap) + Latent Space Anniversary with Lindy.ai, RWKV, Pixee, Julius.ai, Listener Q&A!

45 snips
Mar 9, 2024
Florent Crivello, founder of Lindy.ai, and Eugene Cheah, CEO of RecursalAI, dive into the latest research trends in AI. They explore the impact of OpenAI's Sora and Google's Gemini Pro 1.5, including long inference features and multimodal capabilities. The conversation highlights Lindy.ai's innovative productivity tools and RWKV’s advancements as a transformer alternative. They also tackle the complexities of AI integration and the challenges of data management, all while celebrating their first anniversary in the AI podcasting scene.
Ask episode
AI Snips
Chapters
Transcript
Episode notes
ADVICE

Scale Inference Time

  • Scale inference time, as other LLM scaling dimensions like parameters and data have limits.
  • Explore chain of thought, flow engineering, and tools like SGLANG for efficient long inference.
INSIGHT

Npm for Plans

  • There's a need for an "npm for plans" to share and reuse successful LLM planning strategies.
  • Personalization and varying needs might hinder the creation of such a library.
INSIGHT

Synthetic Data Potential

  • Synthetic data generated by LLMs can improve themselves, even without teacher LLMs.
  • This offers a solution to data scarcity issues but risks model collapse.
Get the Snipd Podcast app to discover more snips from this episode
Get the app