AI Weekly 002

AI Weekly #002

🆕 What's new

Enjoy last week's new tutorial:

  • How to generate App logo using ComfyUI? (opens in a new tab): Our tutorials have taught a lot about how to use ComfyUI, but some students have reported not knowing how to get started when they want to apply ComfyUI to their work. Also, there are a lot of little configuration items in ComfyUI, and the tutorials haven't covered them one by one, so it's not clear how to configure them. Therefore, we've decided to write a series of hands-on tutorials to teach everyone how to apply ComfyUI to real work cases, as well as sharing some tricks to help everyone use ComfyUI better.
  • Stable Diffusion Advanced (opens in a new tab): Added the Stable Diffusion XL workflow, which can generate more refined images.

Article updates:

🤩 Weekly‘s AI highlights

If you encounter interesting AI-related articles or perhaps products during your studies, you are welcome to share them on our Discord Share Channel (opens in a new tab), or submit a newsletter Issue (opens in a new tab) on Github. If your submission is accepted, we will feature it in our newsletter to be seen by more people.

📄 Noteworthy papers and technic

An open-source hardware product developed by Stanford University. The technology is quite complex. In simple terms, users can control the robot's arm to perform a certain task (such as cooking) through the controller. After completing this task more than 50 times, the machine can independently mimic and complete the task on its own.

This model uses the FaceID of facial recognition models to replace CLIP image embedding, thus more accurately capturing and reproducing specific character facial features.

Microsoft released a new small language model Phi-2. Phi-2 only has 2.7B parameters, but the performance of Phi-2 surpasses the Mistral and Llama-2 models, which have 7B and 13B parameters respectively. The screenshot below shows Phi-2's score. The core reason for achieving this effect is that Microsoft compiled a high-quality dataset to train this model.

🛠️ Products you should try

Convai has released its latest game AI feature - Scene perception and NPC interactive generation, in the CES2024 NVIDIA special live broadcast. Using this feature, the NPC's dialogue will be different each time in the same scene, greatly improving the gaming experience. What's even cooler is that NPCs can now have voice conversations with players.

A very popular AI image enlargement tool recently, it works impressively well. What surprised me most is that there are only two people in the company that developed this product.

Finally, in 2024, there was an AI hardware product that officially went on sale. The industrial design is very interesting. And the sales on the first day reached 10,000 units.

AI products are updated every day, but are you curious about which AI products are shut down? This website lists many discontinued products. If you have an idea for developing an AI product, why not take a look at this list and see if any predecessors have developed it.

Subscribe for free to receive new posts and support my work. Or join our Discord.