30 Latest AI Open Source Projects of the Week(2024.12.23-2024.12.29)
Discover 30 cutting-edge AI open-source projects (12/23–12/29/2024), including models for NLP, time series, robotics, and video analysis. Explore now!
I’m sharing some interesting AI open-source models and frameworks for this week (2024.12.23-2024.12.29).
There are a total of 30 AI open-source projects.
Project: Bamba
Bamba-9B is a decoder-only language model based on the Mamba-2 architecture, designed for various text generation tasks. The model is trained from scratch using a two-stage training approach.
The first stage involves training on 2 trillion tokens from the Dolma v1.7 dataset, while the second stage adds another 200 billion tokens with a curated mix of high-quality data to further enhance performance and output quality.
https://github.com/foundation-model-stack/bamba
Project: Tiny LLM
This project aims to build a lightweight Chinese large language model, enabling users to quickly learn about large model-related knowledge.
The project adopts an open-source general architecture, supports various optimization techniques and inference frameworks, and provides full access to data and code, covering the complete process from tokenization to deployment.
https://github.com/wdndev/tiny-llm-zh
Project: OmniEval
OmniEval is a comprehensive automated RAG evaluation benchmark tailored for the financial domain.
The project offers tools for constructing retrieval corpora, generating evaluation data samples, and conducting model inference and evaluation. OmniEval aims to improve evaluation efficiency through automation and provides researchers with a reliable framework for financial studies.
https://github.com/RUC-NLPIR/OmniEval
Project: LLM4AD
LLM4AD is an open-source Python-based platform leveraging large language models (LLMs) for automated algorithm design.
Initially developed for optimization tasks, the framework is flexible enough to be applied in other areas, including machine learning, scientific discovery, game theory, and engineering design. LLM4AD provides documentation and sample materials, enabling users and developers to easily test, build, and deploy their applications and conduct related research.
https://github.com/Optima-CityU/llm4ad
Project: Kheish
Kheish is an open-source, multi-agent orchestration platform leveraging large language models (LLMs) to handle complex tasks. Through multiple specialized "agents" and flexible workflows, Kheish coordinates various steps like proposal generation, review, validation, and formatting to produce reliable, high-quality results.
It seamlessly integrates external modules such as file system access, shell commands, or vector storage to enrich context and handle large codebases or documents.