Friday, November 6
Dungeon Ranker - Mino's agents ecosystem
0/1 of today's quests done
LV 4NEWBIE
▲0-day streak
Level · NEWBIELv 4969 XP to next
Your streak0ddays you wrote or worked yourself · best 2
Steps this week15▼ 4 vs last weekyou 0 · party 15
Party worked18hthis week · 1 quest cleared
Active quests
0/1 today
- Hong Kong Smart Identity Card appointmentto do
Recently cleared
last 7 days
- Submit COMP5222 group project proposalcleared ✓
Coming up
next 7 days
Mon, Nov 9
- Attend seminar: Career Development for Engineering Research Studentsto do
From your projects
- FILEThe AI Interview Book (consolidated cheat sheet)Interview Cheat Sheet
- PPTMacNet paper presentation (8 min)
- MDDesigning speech models: components, interfaces and architecture recipesSpeech Model Architecture
- MDDungeon TeacherDungeon Ranker
- MDDungeon EngineerDungeon Ranker
- MDDungeon ResearcherDungeon Ranker
Set the mood
Deep
Quiet room for hard thinking.
Today
Reading for Friday, Nov 6.
Paper for today
OLIVE's joint masked-latent-prediction and waveform-reconstruction objective for speech SSL is directly relevant to your Speech Model Architecture study and could inform representation choices for the dereverberation-memory encoder.
Research fact
Researchers injected artificial 'thoughts' directly into large language models to test whether the models can accurately report on their own internal states, probing a real limit of AI introspection.
Worth readingTiếng Việt
Một câu chuyện riêng tư, nhẹ nhàng về khác biệt văn hoá ẩm thực giữa bạn bè, đáng đọc chậm để suy ngẫm.
From the labs
- OpenAI · 2026-11-05GPT‑6 Cyber: Advancing the frontier of aligned cybersecurity agents
- Hugging Face · 2026-11-05New NVIDIA Cosmos 3 Open Models for Interactive World Simulation
- shattered.io · 2026-11-032 DeepMind Staff Ask Google to Lift Lifetime Gag [2026] - shattered.io
- Hugging Face · 2026-11-05A Closer Look at Phonon TTS. Launching High-Quality On-Device Speech That Tops the Benchmarks.
- The Decoder · 2026-11-05Google's Gemini Omni 1.2 Flash stretches AI video clips to 15 seconds and cuts pricing by 30 percent