Qwen2.5-Max: Exploring the Intelligence of Large-scale MoE Model
QWEN CHAT API DEMO DISCORD It is widely recognized that continuously scaling both data size and model size can lead to significant improvements in model intelligence.…
Latest
Alternative DeepSeek V3 providers
DeepSeek's API has been experiencing reliability issues. Here are alternative providers you can use.
AI Daily Brief — 27 January 2025
Monday delivered the largest single-day market-cap loss in US stock-market history. NVIDIA fell 16.97% to close at $118.58, erasing approximately $589 billion in market value in…
AI Daily Brief — 26 January 2025
Sunday delivered the consumer milestone that would lock in Monday's market response. DeepSeek hit #1 on the US Apple App Store, displacing ChatGPT and topping charts…
Qwen2.5-1M: Deploy Your Own Qwen with Context Length up to 1M Tokens
Tech Report HuggingFace ModelScope Qwen Chat HuggingFace Demo ModelScope Demo DISCORD Introduction Two months after upgrading Qwen2.5-Turbo to support context length up to one million tokens,…
Qwen2.5 VL! Qwen2.5 VL! Qwen2.5 VL!
QWEN CHAT GITHUB HUGGING FACE MODELSCOPE DISCORD We release Qwen2.5-VL, the new flagship vision-language model of Qwen and also a significant leap from the previous Qwen2-VL.…
AI Daily Brief — 25 January 2025
Saturday was the calm-before-the-crash day. With US markets closed, DeepSeek's app daily active users surged more than 110% versus the prior week and weekly unique web…
AI Daily Brief — 24 January 2025
Friday closed Davos with AI as the dominant theme and silicon as the unsolved bottleneck. Jensen Huang called AI "the largest infrastructure buildout in human history."…
R1+Sonnet set SOTA on aider’s polyglot benchmark
R1+Sonnet has set a new SOTA on the aider polyglot benchmark. At 14X less cost compared to o1.
AI Daily Brief — 23 January 2025
Thursday rewrote US AI policy and launched the year's first headline agentic-AI product. Trump signed Executive Order 14179 revoking Biden's October 2023 Safe AI EO. OpenAI…
Computer-Using Agent
A universal interface for AI to interact with the digital world.
AI Daily Brief — 22 January 2025
Wednesday delivered three simultaneous shocks. ByteDance released Doubao 1.5 Pro at API prices roughly 50x below GPT-4o and 200x below OpenAI o1. DeepSeek-R1 and Moonshot's Kimi…
RT will brown: the reasoning era is a big opportunity for the open-source world to offer something that closed labs basically cannot: agents trained o…
RT will brownthe reasoning era is a big opportunity for the open-source world to offer something that closed labs basically cannot:agents trained on your internal systems…
AI Daily Brief — 21 January 2025
Day Two of the Trump administration delivered the Stargate Project — a $500-billion four-year US AI infrastructure joint venture announced at the White House by Trump,…
AI Daily Brief — 20 January 2025
Inauguration day delivered the single most consequential 24 hours in open-source AI history alongside a wholesale rewrite of US AI policy. DeepSeek released R1 and R1-Zero…
Global-batch load balance almost free lunch to improve your MoE LLM training
GITHUB HUGGING FACE MODELSCOPE DISCORD Background The Mixture-of-Experts (MoEs) architecture has become a popular model-parameter-scale-up technique. Typically, one MoE layer consists of a router (often parameterized…
AI Daily Brief — 19 January 2025
Sunday was the calm-before-the-storm day. TikTok restored US service after Trump publicly committed to a Day-1 executive order pausing enforcement; the app came back without ByteDance's…
AI Daily Brief — 18 January 2025
Saturday went dark — literally. At roughly 10:30 PM ET, TikTok voluntarily blacked out service for about 170 million US users. The most consequential AI-adjacent signal…
AI Daily Brief — 17 January 2025
Friday delivered a unanimous Supreme Court, a White House staffing pick, an FTC inquiry result, and a Google DeepMind paper that would influence the year's inference-time…
AI Daily Brief — 16 January 2025
Thursday's anchor came from Hsinchu: TSMC walked the Street through a blockbuster Q4 — revenue up 38.8% YoY, 59% gross margin, record AI-accelerator demand — and…
Common pitfalls when building generative AI applications
As we’re still in the early days of building applications with foundation models, it’s normal to make mistakes. This is a quick note with examples of…
AI Daily Brief — 15 January 2025
Wednesday brought MiniMax-01's public open-weights drop — and with it the year's first credible Chinese open-source punch at GPT-4o pricing. The Diffusion Rule formally published in…
Using uv as an installer
Reliably packaging & distributing python CLI tools is hard. Aider uses uv in novel ways to make it easy to install the aider CLI, its dependencies…
AI Daily Brief — 14 January 2025
Tuesday's headline was Biden's second AI executive order in two days — EO 14141, directing the Department of Defense and Department of Energy to lease federal…