Open LLM Upgrades πŸ†• // Gemma 2 Performance πŸ’Ž // SeaKR's Self-aware Learning 🧠

Open LLM Upgrades πŸ†• // Gemma 2 Performance πŸ’Ž // SeaKR's Self-aware Learning 🧠

Author: Earkind June 28, 2024 Duration: 13:48

HuggingFace has upgraded the Open LLM Leaderboard to v2, adding new benchmarks and improving the evaluation suite for easier reproducibility.

Gemma 2, a new addition to the Gemma family of lightweight open models, delivers the best performance for its size and offers competitive alternatives to models that are 2-3Γ— bigger.

SeaKR is a new model that re-ranks retrieved knowledge based on the LLM's self-aware uncertainty, outperforming existing adaptive RAG methods in generating text with relevant and accurate information.

Step-DPO is a new method that enhances the robustness and factuality of LLMs by learning from human feedback, achieving impressive results in long-chain mathematical reasoning.

Contact:Β Β sergi@earkind.com

Timestamps:

00:34 Introduction

01:21Β HuggingFace Updates Open LLM Leaderboard

03:19Β Gemma 2: Improving Open Language Models at a Practical Size

04:16Β From bare metal to a 70B model: infrastructure set-up and scripts

05:21 Fake sponsor

07:11Β SeaKR: Self-aware Knowledge Retrieval for Adaptive Retrieval Augmented Generation

08:47Β Simulating Classroom Education with LLM-Empowered Agents

10:16Β Step-DPO: Step-wise Preference Optimization for Long-chain Reasoning of LLMs

12:31 Outro


Each morning, GPT Reviews serves up a fresh, slightly chaotic conversation about everything happening in artificial intelligence. This daily podcast from Earkind is actually crafted by AI, offering a unique blend of the latest headlines, major announcements, and intriguing research plucked from sources like arXiv. But it’s far from a dry briefing. The dynamic comes from its four distinct hosts: Giovani Pete Tizzano brings relentless optimism as an AI enthusiast, while Robert, the analyst, provides a grounded and often skeptical counterpoint. Olivia, who’s deeply embedded in online communities, shares the buzz and broader reactions, and Belinda, the witty research expert, helps unpack the technical details with clarity and a sharp sense of humor. Tuning in feels like dropping into a lively roundtable where complex ideas are debated, explained, and occasionally laughed about. You’ll get a comprehensive yet digestible overview of the AI landscape, all wrapped in a format that’s as entertaining as it is informative. The result is a consistently engaging listen that keeps you updated without feeling like homework, making it a standout in the daily news podcast space.
Author: Language: English Episodes: 100

GPT Reviews
Podcast Episodes