GoJun AI Daily

English

简体中文

English

简体中文

Appearance

Spend 1 minute each day to get selected cutting-edge AI information.

The content covers, but is not limited to, cutting-edge AI news, AI tools, AI painting, open-source projects, and learning tutorials, etc.

Follow AI Daily to stay updated with AI trends. We hope it helps you. For important information, we will make separate posts with detailed introductions.

Here is the latest AI information for August 9.

Cutting-edge News ​

1. Alibaba releases the Qwen2-Math series models, focusing on solving mathematical problems.

There are three versions: 1.5B, 7B, and 72B, with the 72B surpassing GPT-4o and Claude 3.5 in mathematical capabilities in benchmark tests.

Detailed introduction: https://qwenlm.github.io/blog/qwen2-math/

Model download: https://huggingface.co/collections/Qwen/qwen2-math-66b4c9e072eda65b5ec7534d

However, the model currently only supports English, and the bilingual (Chinese-English) model will be released later.

img

2. Google announces a price cut for Gemini 1.5 Flash!

Input costs have dropped by 78%, and output costs by 71%, while support for over 100 languages has been added to the API.

Detailed introduction: https://developers.googleblog.com/en/gemini-15-flash-updates-google-ai-studio-gemini-api/

Gemini 1.5 Flash price list effective August 8 2024

3. Apple open-sourced a new image generation model and training method.

Using the CC12M dataset containing 12 million images for training, they developed a high-quality image diffusion model.

GitHub: https://github.com/apple/ml-mdm

The GitHub repository includes an overview of the codebase structure, core concepts, tutorials, etc. Those interested can check it out.

mdm text to image outputs

Open-source Models ​

1. A language model optimized for role-playing and conversation, Peach-9B-8k-Roleplay.

Based on the Yi-1.5-9B model, it was fine-tuned using over 100,000 synthetic conversation data, specifically targeting role-playing and dialogue scenarios, and supports both Chinese and English.

Model download: https://huggingface.co/ClosedCharacter/Peach-9B-8k-Roleplay

Peach

Learning Tutorials ​

1. A step-by-step video tutorial on implementing multimodal language models.

The tutorial is about 6 hours long, with detailed explanations and hands-on coding, guiding you through the core mechanisms of multimodal language models.

Detailed introduction: https://t.zsxq.com/KCGfe

img