Interconnects

Exploring the performance trade-offs of open LLM models like Mistrel 8 and DBRX. Discussion on the compute efficiency, growth trends, and optimizing resources for large language models. Debate on non-commercial licenses and comparisons between open and closed LLMs.

The podcast discusses the varied perspectives on open-source AI, including accelerationists, scientists promoting transparency, inclusion and fighting power concentration, and freedom advocates. It explores the complexities of defining 'openness' in the tech industry and emphasizes the importance of diverse viewpoints in shaping open-source projects.

Exploring Databricks' new model DBRX, surpassing Mixtral and Llama 2 in performance. Discussion on AI generated audio, Python, and 11Labs. Details on open LLM and AI strategy, playing with DBRX Instruct. Digging into the narrative and strategic ML implementation. Exploring model capabilities, author's AI newsletter, and system dynamics.

Exploring the shift towards trust and performance-focused evaluations, the rising costs of evaluation tools, and the introduction of RewardBench for evaluating reward models. Discussing the challenges in evaluating different AI models, the need for standardized frameworks, and incremental upgrades in evaluation systems.

Exploring challenges in the AI industry with the rise of GPT4 class models and the development of unique language models. Insights on building moats despite commoditization, opportunities in the Open, and the usefulness of LLMs. Discussion on upcoming developments and the competitive dynamics between companies in the LLM space.

Exploring the complexities of defining an open-source LLM, new naming schemes, bio-risks, transparency, safety, licenses, and copyright in AI models. Debunking myths, discussing regulatory implications, and the impact of Elon Musk's lawsuit against OpenAI on the tech ecosystem.

Louis Castricato, a researcher at EleutherAI and founder of Synth Labs, dives deep into the fascinating world of RLHF. He explores the complexities of preference learning and the shift from PPO to DPO in reinforcement learning. The conversation highlights the challenges of biases in AI, especially regarding representation in training data. Castricato also shares insights on Gemini's impact on data safety, the evolution of model evaluation techniques, and the importance of collaborative efforts in advancing AI research.

Tips on assessing and curating AI content, focusing on model credibility, depth vs. breadth, reproducibility, and community dynamics. Emphasizes the importance of research papers, networking, and building relationships in the AI field.

Google rejoins the open model party and gets some backlash for a frequent problem for generative AI.This is AI generated audio with Python and 11Labs. Music generated by Meta's MusicGen.Source code: https://github.com/natolambert/interconnects-toolsOriginal post: https://www.interconnects.ai/p/gemma-google-ships-it00:00 Google ships it: Gemma open LLMs and Gemini backlash03:12 Getting to know Gemma07:11 Alignment details08:55 Aside: What is REINFORCE? Some history of RL11:08 Implementation details and RLHF12:18 Terms of use: RAIL Licenses history repeated14:05 Is Google back on top? Gemini's woesFigure 1: https://huggingface.co/datasets/natolambert/interconnects-figures/resolve/main/gemma/img_008.webpFigure 2: https://huggingface.co/datasets/natolambert/interconnects-figures/resolve/main/gemma/img_014.pngFigure 3: https://huggingface.co/datasets/natolambert/interconnects-figures/resolve/main/gemma/img_035.pngFigure 4: https://huggingface.co/datasets/natolambert/interconnects-figures/resolve/main/gemma/img_051.pngFigure 5: https://huggingface.co/datasets/natolambert/interconnects-figures/resolve/main/gemma/img_055.png This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.interconnects.ai/subscribe

10 Sora and Gemini 1.5 follow-ups: code-base in context, deepfakes, pixel-peeping, inference costs, and moreThis is AI generated audio with Python and 11Labs. Music generated by Meta's MusicGen.Source code: https://github.com/natolambert/interconnects-toolsOriginal post: https://www.interconnects.ai/p/sora-gemini-follow-up00:00 10 Sora and Gemini 1.5 follow-ups: code-base in context, deepfakes, pixel-peeping, inference costs, and more00:46 1. Deepfake detection of Sora01:59 2. Playing with long-context, problem settings, and prompting03:39 3. Gemini paper snooping: contamination and citation games05:42 4. Training data and token estimates of YouTube07:42 5. Unlocking model-based RL and downstream research08:52 6. Midjourney style matching, V-JEPA, replicating Sora in the open10:09 7. Architectures and academic links10:57 8. Pixel peeping from the arts11:58 9. Inference costs13:24 10. Pressure on Llama and Mistral14:03 11. Sound effects, physics, and the complete pictureFigure 1: https://huggingface.co/datasets/natolambert/interconnects-figures/resolve/main/sora-2/img_003.pngFigure 2: https://huggingface.co/datasets/natolambert/interconnects-figures/resolve/main/sora-2/img_007.mp4Figure 3: https://huggingface.co/datasets/natolambert/interconnects-figures/resolve/main/sora-2/img_009.mp4Figure 4: https://huggingface.co/datasets/natolambert/interconnects-figures/resolve/main/sora-2/img_011.mp4Figure 5: https://huggingface.co/datasets/natolambert/interconnects-figures/resolve/main/sora-2/img_037.mp4Figure 6: https://huggingface.co/datasets/natolambert/interconnects-figures/resolve/main/sora-2/img_044.pngFigure 7: https://huggingface.co/datasets/natolambert/interconnects-figures/resolve/main/sora-2/img_047.pngFigure 8: https://huggingface.co/datasets/natolambert/interconnects-figures/resolve/main/sora-2/img_049.mp4 This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.interconnects.ai/subscribe