The Llama 3 Herd of Models

Name: The Llama 3 Herd of Models
Uploaded: 2026-10-10T14:32:04.779956+00:00
Duration: 676 s
Description: Llama 3 is a foundation model designed for multilinguality, coding, and reasoning. It was developed using massive pre-training on diverse data and refined through Supervised Finetuning (SFT) and Direct Preference Optimization (DPO). The resulting model shows competitive performance across various benchmarks, especially in code and video understanding.

Llama 3 is a foundation model designed for multilinguality, coding, and reasoning. It was developed using massive pre-training on diverse data and refined through Supervised Finetuning (SFT) and Direct Preference Optimization (DPO). The resulting model shows competitive performance across various benchmarks, especially in code and video understanding.

Open the full episode