#LLM pre-training
379 related articles
Deep Dives·3 min
Complete Guide to LLM Training: Pre-training, SFT Fine-tuning, and Preference Alignment Explained
Complete guide to the three core LLM training stages: pre-training, supervised fine-tuning (SFT), and preference alignment (DPO/PPO), covering LoRA, distillation, quantization, and pruning.
Read more →
·3 min
Gemini's Hilarious Reply Goes Viral on Reddit: A Deep Dive into the AI Personification Trend
Google Gemini went viral on Reddit for a humorous reply, dubbed an "undercover wasp." This article explores the technical origins of AI humor, RLHF-driven personality shaping, and the future of AI personification.
Read more →
·3 min
Symbio: Technical Analysis and Challenges of the AI Self Fine-Tuning Loop
In-depth analysis of Symbio's AI self fine-tuning loop mechanism, exploring the technical logic of self fine-tuning loops, personalization value, and challenges like catastrophic forgetting and model drift.
Read more →
·2 min
Why Does AI Write Like a Redditor? How Training Data Shapes AI Language Style
Exploring why AI LLMs write with a distinct Reddit style. From Reddit's high proportion in GPT training data to typical AI sentence patterns, revealing how training corpora shape model personality.
Read more →
·3 min
LLMs Rely on Tools for Math: Intelligence or a Shortcut?
When LLMs need calculators for math, is it intelligence or proof they can't compute? Exploring tool calling vs. human cognition and two frameworks for evaluating AI intelligence.
Read more →
·3 min
The AI Replacement Wave: Can Switching to a Trade Really Save You?
Can switching to plumbing or electrical work really protect you from AI long-term? This article analyzes white-collar vs. blue-collar replacement timelines, the durability of the physical moat, and personal strategies more important than picking the right career track.
Read more →
·3 min
LLM Post-Training on an 8GB GPU: A Practical Guide to SFT, DPO, and GRPO
Learn how to complete LLM post-training on a consumer GPU with just 8GB VRAM, covering SFT, DPO, and GRPO methods using LoRA quantization and other techniques.
Read more →
·3 min
AI Cross-Validation: A Practical Guide to Using Gemini to Review ChatGPT Outputs
Explore cross-validation methods using Gemini to review ChatGPT outputs. Analyze the value and limitations of AI peer review with a rational multi-model collaboration framework.
Read more →
·3 min
The Fundamental Conflict Between Backpropagation and Continual Learning: Can Catastrophic Forgetting Be Solved?
Exploring why standard backpropagation causes catastrophic forgetting, its fundamental conflict with continual learning, and whether solutions like EWC and experience replay can bridge the gap.
Read more →
·3 min
AI Can't Build Usable Products — Developers' Jobs Haven't Disappeared
AI can generate code snippets and demos, but usable products still require human engineers' judgment and responsibility. This article analyzes AI coding tools' limits and developers' evolving roles.
Read more →
·3 min
System Prompts Should Be Version-Controlled Like Code
System prompts drive LLM apps but often lack version control and regression testing. Learn how to manage them with versioning, structured separation, testing, and code review.
Read more →
·2 min
General Agents Are Essentially Coding Agents: The Paradigm Shift from Tool Calling to Code Execution
Explore why general AI agents are essentially coding agents. From Turing completeness to composability and verifiability, discover the paradigm shift from Function Calling to Code as Action.
Read more →
·3 min
Kimi-K3 Scores 60.4% on ARC-AGI-2: A Breakthrough in Abstract Reasoning
Kimi-K3 scores 60.4% on ARC-AGI-2, far surpassing most LLMs. This article analyzes what ARC-AGI-2 tests, what this score means for abstract reasoning, and its implications for the AI industry.
Read more →
·3 min
Self-Hosting a Code Review AI Agent: From Architecture Design to Engineering Implementation
A deep dive into building and self-hosting a code review AI Agent from scratch, covering architecture design, context management, model selection, and noise control.
Read more →
·5 min
Have AI Doomers Actually Used AI? A Developer Exposes the Gap Between Hype and Reality
AI Doomers warn AI will destroy humanity, but have they actually built AI apps? A developer's sharp critique reveals the vast gap between AI demos and real engineering practice.
Read more →
·5 min
The Complete Learning Path to Understanding the Kimi K3 Technical Report
A systematic learning path for understanding the Kimi K3 technical report, covering MoE, MLA, distributed training, and modern post-training techniques.
Read more →
·3 min
LaunchPanda Review: 255 Startup Directory Rankings to Boost SEO & GEO Growth
In-depth review of how LaunchPanda ranks 255+ startup directories with data to provide indie developers a step-by-step SEO backlink and GEO promotion roadmap, reducing cold start decision costs.
Read more →
·2 min
LaunchPanda Review: 255 Startup Directory Rankings to Boost SEO & GEO Growth
In-depth review of how LaunchPanda ranks 255+ startup directories to help indie developers build SEO backlinks and optimize for GEO, reducing cold-start decision costs.
Read more →
·4 min
Automatic Tibetan-Chinese Bilingual Subtitle Generation: A Complete Technical Breakdown
A complete technical guide to automatic Tibetan-Chinese bilingual subtitle generation, covering Tibetan ASR (Whisper/wav2vec), machine translation (NLLB), timeline alignment, and subtitle export for low-resource language creators.
Read more →
·3 min
Can Talking Like a Caveman Save 65% on Tokens? An In-Depth Analysis
Can caveman-style minimal prompts save 65% on Tokens? We analyze task quality, hidden cost transfers, and model robustness to reveal the right Token optimization strategies.
Read more →