September 5th, 2023 - Frontiers in AI Efficiency and Capability: From Turbocharged Transformers and Extended Contexts to High-Definition Video Generation and Self-Tuned Learning

ML Cult

• Fast Inference from Transformers via Speculative Decoding
• YaRN: Efficient Context Window Extension of Large Language Models
• VideoGen: A Reference-Guided Latent Diffusion Approach for High Definition Text-to-Video Generation
• RLAIF: Scaling Reinforcement Learning from Human Feedback with AI Feedback

Support the Show.

فصول

1. Intro (00:00:00)

2. Fast Inference from Transformers via Speculative Decoding (00:01:51)

3. YaRN: Efficient Context Window Extension of Large Language Models (00:04:11)

4. VideoGen: A Reference-Guided Latent Diffusion Approach for High Definition Text-to-Video Generation (00:06:06)

5. RLAIF: Scaling Reinforcement Learning from Human Feedback with AI Feedback (00:07:34)