#LoRA training
695 related articles
·7 min
The Complete AI Character LoRA Training Workflow: Hands-On Multi-Character Consistency Generation with Local Deployment
A detailed guide to a complete local AI character generation workflow: from the five golden rules of LoRA training and automated ComfyUI dataset construction to hands-on comparisons of Crea2, Ideogram4, and Wan for multi-character same-frame interaction—all running free on personal hardware.
Read more →
·3 min
LoRA Training Showdown: We Tested 6 Base Models — Does Ideogram Beat Flux?
A developer ran a 4-day benchmark testing LoRA training across Ideogram, Flux 1 Dev, Flux 2 Dev & more — revealing overfitting traps and surprising rankings.
Read more →
·2 min
LoRA Training Showdown: Which Base Model Is Best for Character Fine-Tuning? A 6-Model Real-World Test
Real-world LoRA training comparison across Ideogram, Flux 1 Dev, Z Image, Flux 2 Klein, and Krea — revealing which base model best handles face fidelity and generalization for AI portrait developers.
Read more →
·2 min
Directly Optimizing Face Similarity: A Practical Guide to Accelerating Character LoRA Training
Ditch inefficient noise prediction and train character LoRAs by directly optimizing face similarity. Using the DRaFT differentiable reward method, training completes in 10-12 minutes on an RTX 4090, far outperforming traditional SFT. Includes open-source code.
Read more →
·3 min
Moonshot AI Open-Sources FlashKDA: A Deep Dive into High-Performance CUDA Kernels for KDA Attention
Moonshot AI open-sources FlashKDA, providing high-performance CUDA kernels for Kimi Delta Attention. Explore its technical principles, performance gains, and value for long-context training and inference.
Read more →
·2 min
Reverse Engineering Apple's ANE: Open-Source Project Enables On-Device Training on the Neural Engine
Deep dive into the maderix/ANE GitHub project that reverse engineers Apple's private APIs to enable neural network training on the Apple Neural Engine, exploring its technical approach, efficiency gains, compliance risks, and implications for on-device AI.
Read more →
·3 min
The ISNAD Framework: Building a Trust Verification Layer for Multi-Agent AI Systems Using a Millennium-Old Scholarly Tradition
The ISNAD framework adapts Islamic chain-of-transmission verification to build a trust layer for multi-agent AI systems, focusing on claim verification over agent authentication to combat hallucinations and silent failures.
Read more →
·3 min
The Era of AI Capability Overhang: Why You Need to Reset Your Ambition Every 3 Months
Understanding Capability Overhang in the AI era: when model capabilities far exceed application imagination, how teams should reset feasibility boundaries quarterly to avoid ceding advantages to competitors.
Read more →
·2 min
Transformer²: Achieving Co-Design of Robot Morphology and Control with a Unified Architecture
Deep dive into how Transformer² uses a unified Transformer architecture to integrate robot morphology design and motion control into one model, enabling task-driven end-to-end co-design for embodied AI.
Read more →
·4 min
A Guide to Full-Stack Machine Learning Project Repository Structure: From Chaos to Convention
A detailed guide to organizing full-stack ML project repositories, covering directory structure design, data-code separation, and externalized configuration to help ML developers move from experimental code to production-grade engineering standards.
Read more →
·4 min
Full-Stack Machine Learning Project Repository Structure Guide: From Chaos to Convention
A detailed guide to organizing full-stack ML project repositories, covering directory structure design, data-code separation, and configuration externalization to help ML developers move from experimental code to production-grade engineering.
Read more →
·2 min
OpenReviewer: An Open-Source LLM Built Specifically for Scientific Paper Review
OpenReviewer is an open-source LLM for generating critical scientific paper reviews. This article analyzes its technical approach, use cases, and limitations.
Read more →
·3 min
New Gemini Jailbreak Method: Deep Analysis of the Observer and Accomplice Technique
Deep analysis of a Gemini jailbreak technique—the Observer and Accomplice method—examining how it exploits contextual manipulation and reasoning chain inconsistencies to bypass AI safety alignment.
Read more →
·2 min
OpenReviewer: An Open-Source LLM Built Specifically for Scientific Paper Review
OpenReviewer is an open-source LLM for generating critical scientific paper reviews. This article analyzes its technical approach, use cases, and limitations.
Read more →
·2 min
Training AI to Play Devil May Cry 3 with Reinforcement Learning: Challenges and Practices in Action Game AI
An indie developer trains AI to autonomously play Devil May Cry 3 using reinforcement learning. Explore the core challenges of action game AI including sparse rewards, high-dimensional action spaces, and real-time decision-making.
Read more →
·2 min
Training AI to Play Devil May Cry 3 with Reinforcement Learning: Challenges and Practices for Action Game AI
An indie developer trains AI to autonomously play Devil May Cry 3 using reinforcement learning. This article analyzes the core challenges including sparse rewards, high-dimensional action spaces, and real-time decision-making.
Read more →
·3 min
Kimi Delta Attention Explained: The Evolution from Standard Attention to Linear Attention
A deep dive into Kimi Delta Attention (KDA): tracing the evolution from quadratic Softmax attention through linear attention, Delta rules, and gated decay mechanisms, with insights on associative memory and hardware optimization.
Read more →
·3 min
Kimi Delta Attention Explained: The Evolution from Standard Attention to Linear Attention
Deep dive into Kimi Delta Attention (KDA): from standard Softmax attention's quadratic bottleneck through linear attention, Delta Rule, and gated decay mechanisms — the complete evolution explained.
Read more →
·3 min
PIRL Framework Explained: A Smooth Transition from Open-Loop Exploration to Closed-Loop Reinforcement Learning
Deep dive into the PIRL reinforcement learning framework: how to smoothly transition from open-loop exploration to closed-loop RL, mitigating the exploration-exploitation dilemma and improving sample efficiency.
Read more →
·2 min
Gemini 3.6 Flash Hands-On: Upgrades in Both Quality and Token Efficiency
In-depth analysis of Google Gemini 3.6 Flash's core upgrades including output quality improvements and token consumption optimization, with developer migration advice.
Read more →