#mathematical optimization
566 related articles
Deep Dives·3 min
Alibaba's QwQ-32B Open-Sourced: How 32B Parameters Rival DeepSeek R1's 671B
Alibaba's open-source reasoning model QwQ-32B achieves performance rivaling DeepSeek R1 (671B) with only 32B parameters through a two-stage reinforcement learning strategy on verifiable tasks.
Read more →
Tutorials·3 min
GPT-OSS Open-Source Model Local Deployment Tutorial: Real-World Performance and VRAM Usage Analysis
Complete guide to deploying OpenAI's open-source GPT-OSS model locally with Ollama. Real-world testing of the 20B version on RTX 4090 covering Chinese comprehension, logical reasoning, and VRAM usage analysis under MoE architecture.
Read more →
Deep Dives·3 min
Reinforcement Learning Drives AI Reasoning Evolution: From Imitator to True Thinker
How RL, Self-Play, and Verifiers work together to evolve LLM reasoning — driving the leap from SFT imitation to true System 2 deep thinking.
Read more →
Product Reviews·4 min
GLM-4.7 In-Depth Review: Coding Capabilities on Par with Claude Sonnet 4.5
In-depth testing of Zhipu AI's open-source GLM-4.7 coding abilities across SVG animation, 3D game dev, iOS native apps, and browser automation, compared against Claude Sonnet 4.5 and DeepSeek V3.2.
Read more →
Product Reviews·4 min
Kimi K2 Thinking Hands-On Review: Can It Replace Sonnet 4.5 in Claude Code?
Hands-on testing Kimi K2 Thinking in Claude Code across text creation, programming, agent building, and full-stack apps. Comparing it to Claude Sonnet 4.5 and DeepSeek as a cost-effective AI coding alternative.
Read more →
Product Reviews·3 min
Claude Haiku 4.5 Review: Sonnet 4-Level Coding at 1% of the Cost
In-depth review of Claude Haiku 4.5: 73.3% on SWE-bench rivaling Sonnet 4, input at just $1/million tokens. Covers code generation, agentic coding, SVG tests, and Sonnet+Haiku collaboration strategies.
Read more →
Tutorials·3 min
Vibe Coding in Practice: Building an AR SLAM System with AI Programming to Save $28K in Licensing Fees
AR entrepreneur uses Vibe Coding to build a custom SLAM system, replacing $5K-$20K/year commercial licenses. Full hands-on experience with AI-powered SLAM development and why most Vibe Coding projects fail.
Read more →
Product Reviews·3 min
Dify-Chat: An Open-Source AI Application Management Platform Built on Dify API
Dify-Chat is an open-source platform built on the Dify API supporting Chatflow, Workflow, deep thinking visualization, chart rendering, and more to help developers quickly deploy AI application interfaces.
Read more →
Expert Opinions·3 min
Sequoia Capital's Conversation with AI Expert: How the Von Neumann Architecture Shapes the Future of AI
Sequoia Capital's deep conversation with AI expert Konstantine explores how von Neumann's architecture and ideas profoundly shape modern AI development, from computing paradigms to the pursuit of AGI.
Read more →
Tech Frontiers·3 min
Isomorphic Labs Raises $2.1 Billion: How the AlphaFold Team Is Using AI to Reshape Drug Discovery
Isomorphic Labs, the AI drug company founded by DeepMind co-founder Hassabis, raises $2.1B. Built on AlphaFold breakthroughs, it aims to conquer all diseases.
Read more →
Tutorials·4 min
Coze AI Agent Tutorial: A Complete Guide to Building AI Agents from Scratch
Complete guide to building AI agents on Coze from scratch, covering LLM configuration, prompt writing, plugin integration, knowledge base setup, and memory systems.
Read more →
Deep Dives·3 min
The Evolution of LLM Reasoning: A Complete Guide from Chain-of-Thought to DeepSeek-R1
A systematic overview of LLM reasoning evolution, covering Chain-of-Thought, Tree-of-Thought, OpenAI o1, and DeepSeek-R1, with insights from the Awesome-LLM-Reasoning project.
Read more →
Tutorials·3 min
4-bit QLoRA Fine-tuning LLaMA 3 in Practice: A Guide to Training 8 Billion Parameter LLMs on Consumer GPUs
Learn how to fine-tune LLaMA 3 8B on consumer GPUs using QLoRA and Unsloth. Covers 4-bit quantization, LoRA adaptation, Alpaca data format, training pipeline setup, and model deployment—runs on RTX 3090.
Read more →
Product Reviews·2 min
Unsloth: 80% VRAM Reduction — The Best Open-Source Tool for Local LLM Fine-Tuning
Unsloth is an open-source LLM training tool with 63K+ GitHub stars, supporting Gemma 4, Qwen 3, DeepSeek. Reduces VRAM by 50–80%, enabling RTX 4090 to fine-tune 7B models with a no-code Web UI.
Read more →
Research·3 min
SVDQuant: 4-bit Quantization Enables Efficient Diffusion Model Inference on Consumer GPUs
SVDQuant, an ICLR 2025 Spotlight paper, achieves 4-bit diffusion model quantization via low-rank decomposition that absorbs outliers, reducing memory by 75%. Open-source engine Nunchaku (3800+ stars) enables FLUX inference on consumer GPUs like RTX 4060.
Read more →
Tech Frontiers·2 min
DeepSeek V3.2 Released: Proprietary Sparse Attention (DSA) + 50% API Price Cut Explained
DeepSeek releases V3.2-Exp with proprietary DeepSeek Sparse Attention (DSA) for faster long-context training and inference, plus API prices cut over 50%.
Read more →
Tech Frontiers·2 min
DeepSeek V4 Pro Promotional Pricing Extended to May 2026: A Complete Breakdown for Developers
DeepSeek extends V4-Pro API promotional pricing to May 31, 2026. Learn how this impacts developers and enterprises, and what it reveals about LLM pricing strategy.
Read more →
Product Reviews·3 min
MiroFlow Open-Source AI Workflow Framework Review: The Strengths and Concerns Behind Top Rankings Across Multiple Benchmarks
In-depth review of MiroFlow open-source AI workflow framework: technical architecture behind 5+ benchmark Top-1 rankings, multi-model support, Web UI, and comparison with LangChain and Dify.
Read more →
Product Reviews·2 min
Deep Research Web UI: An Open-Source AI Deep Research Assistant with DeepSeek R1 Support
Deep Research Web UI is an open-source AI research assistant supporting DeepSeek R1, enabling automated deep research through iterative search, web scraping, and LLM reasoning. 2100+ Stars on GitHub.
Read more →
Product Reviews·2 min
DeepSeek-Reasonix: A Terminal AI Coding Agent Optimized with Prefix Cache Stability
DeepSeek-Reasonix is an open-source terminal AI coding agent natively designed for DeepSeek models, achieving lower latency and API costs through prefix cache stability optimization.
Read more →