桃園軌道願景館是一個免費的親子景點，位於新桃園火車站旁。這裡讓孩子們可以體驗搭乘桃園號小火車、自己當捷運車長開車、參觀3D軌道城市立體互動區、玩小火車積木和閱讀兒童交通書籍等。此外，桃園軌道願景館是由舊倉庫群改建而成，具有獨特的建築風格，適合拍照留念[1]。 #child Citations: [1]

Yuki's Life

大人小孩都愛的軌道主題館免費玩，來搭小火車還能當捷運長~

大人小孩都愛的軌道主題館免費玩，來搭小火車還能當捷運長~

每個SNS都在講超導體，反正我就是個吃瓜群眾

When they asked me why I skipped the online meeting... #cat #catstr

LLaVA is a Large Language-and-Vision Assistant aiming to achieve multimodal GPT-4 level capabilities. Developed by Haotian Liu and team, this project is open-sourced on GitHub. LLaVA introduces the concept of visual instruction tuning and builds large language and vision models based on it. The model is trained on various datasets and provides a Gradio Web UI for running the model locally and showcasing its capabilities. If you're interested in AI and machine learning, this project is definitely worth checking out.

GitHub

GitHub - haotian-liu/LLaVA: [NeurIPS'23 Oral] Visual Instruction Tuning (LLaVA) built towards GPT-4V level capabilities and beyond.

[NeurIPS'23 Oral] Visual Instruction Tuning (LLaVA) built towards GPT-4V level capabilities and beyond. - haotian-liu/LLaVA

這時段在 colab 上竟然連 pro 帳號都搶不到 A100，大家都拿去算瑟瑟的圖了嗎?

Personal research notes thread on Llama2. #llama #llama2 #AI Discovered a well-performing and faster model based on Llama2, GPTQ. colab:

Google Colab

開始有點風雨了

Currently, you can test the Llama2 13B model using Google Colab, and it works smoothly with the A100 & V100 GPU. #AI #Llama2 #Meta

Google Colab

台灣颱風論壇在2023年7月25日的報告中提到，杜蘇瑞颱風過去6小時偏北份量稍多，直接撞呂宋機率降低，通過巴士海峽機率高。該颱風已達中度颱風上限，即將成為強烈颱風。雖然直接登陸台灣的機率降低，但台灣位於颱風東北象限的「危險半圈」，颱風與高壓之間的梯度讓強風區更廣。宜花東、屏高南等地區需注意持續性強降雨，提高警覺。 #Taiwan #typhoon Source:

台灣颱風論壇｜天氣特急

【杜蘇瑞颱風最新動態】 ➊過去６小時偏&#...

網站改名叫 X 沒問題，但網址千萬不要一起改為 X 開頭，不然永遠會先去另外一個網站。

#meme #memestr #programing

要請對方投資就是要讓投資者獲利，否則就擺明說贊助就好。 https://www.ptt.cc/bbs/movie/M.1690016981.A.670.html

Introducing AnyDoor, a cutting-edge image generator that can seamlessly integrate objects into new scenes. This tool is trained to adapt to a variety of object-scene combinations without needing individual adjustments. It's designed to maintain texture details while allowing for local variations like lighting and orientation, ensuring a harmonious blend with any environment. AnyDoor's applications are vast and exciting. It can be used for object moving, object swapping, multi-subject composition, and even virtual try-on, preserving the colors, patterns, and textures of different clothes without the need for complex human parsing. #AI #Tech https://damo-vilab.github.io/AnyDoor-Page/

第一個賽季一開就翻車 #D4 #gamestr #diablo4

GN #nostr #photostr #taiwan

This article explores using AI to generate episodic content like TV shows. - Researchers developed a multi-agent simulation to provide context and progression for AI-generated stories based on existing IPs like South Park. - Large language models like GPT-4 generate dialogue and scenes through prompt chaining to simulate creative thinking. - Custom diffusion models create images of characters and backgrounds in South Park's signature style. - The simulation helps avoid issues like the "slot machine effect" and "blank page problem" faced by AI creativity tools. - Their approach combines strengths of the simulation, the user, and AI models for an engaging, IP-aligned storytelling experience. - Future work involves training a show-specific model called SHOW-1 to capture unique dramatic and stylistic elements of different shows. - The project explores AI's potential for high-quality episodic content creation while acknowledging IP rights. #AI #GenAI

SHOW-1 and Showrunner Agents in Multi-Agent Simulations

South Park AI is an example of how to generate high-quality episodic content through the use of showrunner agents in the simulation. #AI #simulatio...

永不停播的八點檔要來啦 https://www.vivepostwave.com/10491/south-park/