#model-finetuning

One Platform, Every Stage: How Oumi Streamlines the Foundation Model Lifecycle

An End to End Look at Open Source Training, Evaluation, and Deployment Without the Tool Sprawl

Fine-Tuning LLMs: What Actually Happens Under the Hood

Most engineers can make an API call to fine-tune a model. Far fewer understand what’s actually going on when they do — and honestly, that…

The Silent Killer of LLM Fine-Tuning: Why Your Masking is Broken (And How to Fix It)

Many training disasters trace back to dataset formatting problems. A misconfigured masking setup causes the model to train on both the…

The Knowledge Architecture Trap

Most enterprises chose their AI retrieval layer before they understood the question it actually answers. Here is the framework for fixing…

Fine-Tuning LayoutLMv3 for Intelligent Document Processing

How we trained a model to read and understand documents the way humans do — extracting structured data from invoices, forms, and more.

Fine-Tuning, Part 2: Teaching an LLM to Actually Listen

From “why can’t model follow instructions?” to tokenization, padding, the -100 trick, and batching

Understanding LoRA: Parameter Efficient Fine Tuning for Large Language Models

AI Fabric: Moving from Generic Models to Bespoke AI Engineering

AI Engineering has become the defining discipline for organizations moving beyond generic models toward purpose-built intelligence systems…

Setting Up an MLFlow Tracking Server for Machine Learning Experiments (Part II) — Logging Model…

In my previous post, I showed how to set up a simple local MLflow server for running machine learning experiments (you can check it out…

How to Test Wan2.1 LoRA on RunPod + ComfyUI

Running a base diffusion model is easy. The hard part is properly testing custom models and LoRAs in a reliable way at cloud scale. Results…

Hugging Face Transformers vs PyTorch: How They Work Together for NLP and LLMs

The difference between Hugging Face Transformers and PyTorch, when to use each, and how to combine them for powerful LLM and NLP apps.

Relationship of Finetuning Capacity and Model Size: Adaptation without Regret

Fine-tuning is where a pre-trained model transforms from a generalist to a specialist. But, if your fine-tuning capacity doesn’t match your…

From Pretraining to Fine-Tuning: Strategies for Adapting Models to New Tasks — Part2

Introduction: Unlocking the Power of Pretrained Models

Experiment with LLM Model Merging Techniques

In this article, I will share my views on the innovative method of model blending techniques and algorithms, and discuss the results of my…