No. 37 The latest achievements of the AI big model open source project
Project name: OctoPack
★ Open source implementation code/training dataset/model weights
★ Includes datasets CommitPack (4TB across 350 programming languages), models OctoCoder and OctoGeeX, and evaluation benchmark HumanEvalPack
★ The proposed models OctoCoder and octogex achieve the best performance in HumanEvalPack, demonstrating the advantages of CommitPack in generalizing to a wider set of languages and natural coding tasks
# Instruction fine-tuning code: A large collection of language models
Project name: Neuralangelo
★ Open source implementation code
★ Combine the representation capabilities of multi-resolution 3D hash meshes with neural surface rendering
★ Even without auxiliary inputs such as depth, Neuralangelo can effectively recover dense 3D surface structures from multi-view images with much fidelity over previous methods, allowing detailed large-scale scene reconstruction from RGB video capture
# High-fidelity neural surface reconstruction model
❸ Project name: LLMClient
★ Open source implementation code
★ Support for all major hosted and open source LLMs is designed to simplify the build process, providing an easy way to interact with any LLM and using advanced features such as automatic function calls
# A client library that focuses on the functionality of LLM for function calls and inference
