This project provides a Chinese-language chatbot that can be trained on custom datasets for applications like intelligent customer service, Q&A, and casual conversation. It aims to evolve with AI trends, with plans to incorporate GPT-like models and multimodal capabilities.

How It Works

The project initially implemented a Seq2Seq architecture for conversational AI. Users can train the model using their own corpus, with a provided example using the "Xiaohuangji" dataset. The development roadmap indicates a future shift towards GPT-based models, incorporating advanced features like Reinforcement Learning from Human Feedback (RLHF) and multimodal understanding (e.g., image-text dialogue).

Quick Start & Requirements

• Install: pip install -r requirements.txt (specific requirements vary by framework version).
• Prerequisites: Ubuntu 18.04, Python 3.6, TensorFlow 2.x (e.g., 2.6.0), PyTorch 1.11.0, Flask 0.11.1, Horovod 0.24 (for distributed training).
• Data: Download the "Xiaohuangji" corpus from the repository and place it in the train_data directory.
• Execution: Follow the sequence: data_utls.py -> execute.py -> app.py.
• Distributed Training: Use horovodrun -np n -H host1_ip:port,... python3 excute.py.
• Links: Xiaohuangji Corpus

Highlighted Details

• Supports training custom Chinese chatbots with user-provided corpora.
• Roadmap includes integration of GPT models, RLHF, and multimodal capabilities (image-text dialogue).
• Offers both Seq2Seq and planned GPT branches, with support for multiple AI frameworks.
• Includes a web-based visualization module for dialogue.

Maintenance & Community

• Contact via QQ: 934389697 for exchange and communication.

Licensing & Compatibility

• The README does not explicitly state a license. Compatibility for commercial use or closed-source linking is not specified.

Limitations & Caveats

The project's primary focus is on Chinese language models. The current Seq2Seq version is described as "effect reference (training progress 50%)", suggesting it may not be fully optimized or production-ready. The roadmap indicates significant future development, implying the current state might be experimental.