Trelis Research Courses

How to Quantize a Large Language Model with GGUF or AWQ

1 rating

Learn to quantize large language models using GGUF and AWQ. Explore techniques, compare methods, and gain practical skills for efficient model compression and deployment.

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

Train an LLM from Scratch with Karpathy's Nanochat

Master LLM training from scratch using Karpathy's Nanochat with just $100, covering tokenization, pretraining on 10B tokens, and building a functional chat interface.

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

ARC Prize 2025 Retrospective

Explore the ARC Prize 2025 competition through winner analysis, Trelis TRM fine-tuning methods, and firsthand competition insights in this comprehensive retrospective.

• YouTube
• 1 hour 38 minutes
• Self-Paced
• Free Video

Solving ARC Tasks by Writing Python Programs

Discover how to tackle ARC Prize 2025 challenges through Python programming techniques and research insights from Trelis Research experts.

• YouTube
• 1 hour 8 minutes
• Self-Paced
• Free Video

How to Avoid Going Out of Memory (OOM) in GPU Training

Master GPU memory management techniques to prevent out-of-memory errors during machine learning model training and fine-tuning processes.

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

Run a Kokoro Text-to-Speech Model Locally or as a High-Throughput Server

Master Kokoro TTS deployment locally and as a high-throughput server for synthetic voice generation with various accents using this permissively licensed model.

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

Transcribe British and American English Variants with Whisper Models

Master fine-tuned Whisper models for accurate British and American English transcription variants with performance evaluation and open-source conversion tools.

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

Align Audio and Text for Speech Recognition Model Training

Master audio-text alignment techniques for speech recognition training using CTC aligners, Viterbi processes, and Wave2Vec models to create clean datasets and avoid catastrophic forgetting.

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

Understanding TRM and HRM for Solving ARC Tasks and Machine Learning Applications

Explore TRM and HRM methodologies for solving ARC tasks and discover their broader applications in LLMs and machine learning systems.

• YouTube
• 1 hour 1 minute
• Self-Paced
• Free Video

Whisper Data Preparation and Fine-Tuning with Unsloth

Master data preparation and fine-tuning techniques for Whisper speech recognition models using Unsloth, covering timestamped transcription, dataset creation, and performance evaluation.

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

Transcription Models Zero to Hero - Data Prep, Train and Serve

Master the complete pipeline for building custom transcription models from data preparation through training to deployment with hands-on implementation.

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

Training Recursive Models - A Frontier in Adaptive Compute

Explore recursive neural network architectures that enable adaptive computation, efficient inference on smaller devices, and dynamic resource allocation for AI models.

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

Training Robotics Models - Buy vs Build - Part 1

Explore robotics AI model fundamentals, comparing SO-101 robot options, leader-follower setups, and key differences between ACT, GR00T, and pi0 models for hands-on learning.

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

Multi-GPU Training with Unsloth

Master multi-GPU training techniques using Unsloth for faster model fine-tuning, covering data parallelism, script modifications, and distributed training setup.

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

Fine-Tuning with Custom Compute Metrics

Master advanced fine-tuning techniques by implementing custom compute metrics, exploring teacher-forced vs auto-regressive decoding, and optimizing memory management for ML models.

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