Oxen Courses

How RWKV-7 "Goose" and Its Linear Inference Work

Dive into RWKV-7 "Goose" with author Eugene Cheah, exploring its linear inference capabilities, architecture innovations, benchmarks, and the World Tokenizer, with practical tips for running and fine-tuning this RNN-based model.

• YouTube
• 1 hour 11 minutes
• Self-Paced
• Free Video

How Vision Language Models Reason - Understanding LLaVa Chain of Thought

Delve into the technical aspects of LLaVa-CoT's vision language models, exploring data generation, training methodologies, and inference-time scaling for enhanced reasoning capabilities.

• YouTube
• 37 minutes
• Self-Paced
• Free Video

Fine-Tuning a Tab Tab Code Completion Model

Discover how to fine-tune AI models for intelligent code completion, building Cursor-like "tab tab" functionality with practical examples and model comparisons.

• YouTube
• 45 minutes
• Self-Paced
• Free Video

Inside OpenCoder - Data Processing and Training Pipeline for Code LLMs

Delve into the technical details of OpenCoder's development, from data preprocessing and deduplication to training methodology and evaluation metrics for building effective code LLMs.

• YouTube
• 55 minutes
• Self-Paced
• Free Video

How DeepSeek R1's Reinforcement Learning Works Through GRPO

Delve into the mechanics of DeepSeek R1's reinforcement learning, exploring GRPO's memory optimization, group advantages, and practical applications in AI model training.

• YouTube
• 35 minutes
• Self-Paced
• Free Video

RAGAS - A Framework for Evaluating Retrieval Augmented Generation Systems

Explore RAGAS framework's key evaluation criteria for Retrieval Augmented Generation (RAG) systems, covering faithfulness, answer relevance, and context relevance metrics for improved AI performance.

• YouTube
• 43 minutes
• Self-Paced
• Free Video

How to Fine-Tune FLUX-dev and Comparing it to a Fine-Tuned PixArt Model

Master FLUX-dev fine-tuning with LoRA techniques, data requirements, hardware setup, and code walkthrough. Compare results against PixArt models for optimal AI image generation.

• YouTube
• 59 minutes
• Self-Paced
• Free Video

Fine-Tuning a Diffusion Transformer to Generate OpenAI Level Images

Master fine-tuning diffusion transformers to create high-quality AI images using synthetic data generation and advanced training techniques for professional results.

• YouTube
• 54 minutes
• Self-Paced
• Free Video

How We Fine-tuned Qwen3.0-6B to Beat GPT-4o on Text2SQL

Discover how to fine-tune Qwen3.0-6B for Text2SQL tasks, outperforming GPT-4o through dataset preparation, model evaluation, and optimization techniques in this detailed walkthrough.

• YouTube
• 51 minutes
• Self-Paced
• Free Video

Understanding DeepSeek R1 and GRPO - A Technical Deep Dive

Dive into the technical architecture and development process of DeepSeek's R1 model, exploring GRPO implementation, reasoning capabilities, and model optimization techniques through detailed examples.

• YouTube
• 45 minutes
• Self-Paced
• Free Video

How Phi-4 Cracked Small Multimodality

Explore the architecture and training process of Phi-4, a small multimodal model, including its mixture of LoRAs approach, vision and audio capabilities, and practical applications.

• YouTube
• 36 minutes
• Self-Paced
• Free Video

The Prompt Report: Understanding Prompting Techniques and Taxonomy - Part 1

Explore systematic prompting techniques through an in-depth analysis of template structures, zero-shot methods, emotion prompting, and thought generation approaches for enhanced AI interactions.

• YouTube
• 57 minutes
• Self-Paced
• Free Video

Inside the Rectified Flow Transformer Model - Understanding Advanced Image Generation

Dive into the technical architecture and methodology behind Flux, exploring rectified flow transformers, latent diffusion models, and the innovative approaches that led to superior image generation results.

• YouTube
• 33 minutes
• Self-Paced
• Free Video

Understanding How Llama 3.1 Works - A Technical Deep Dive

Dive into Meta's groundbreaking Llama 3 architecture, exploring pre-training techniques, model capabilities, synthetic data quality, and implementation strategies for advanced AI development.

• YouTube
• 52 minutes
• Self-Paced
• Free Video

Samba: Simple Hybrid State Space Models for Language Modeling

Dive into the mechanics of Samba, a hybrid state space model built on Mamba that enables efficient unlimited context language modeling for advanced AI applications.

• YouTube
• 22 minutes
• Self-Paced
• Free Video