EmoLLM is a Python-based project focused on mental health large language models, supporting pre/post-training, datasets, evaluation, deployment, and RAG with multiple model backbones. It provides extensive configuration and a variety of tuning configurations via xtuner, QA, RAG, and deployment guides.
Collecting history — the radar snapshots this repo daily. The trend line appears after 3 days of data (1 so far).
What it is
EmoLLM is a psychology-focused large language model project. It aims to understand, support, and assist with mental health conversations by fine-tuning LLMs (e.g., InternLM, Qwen, Baichuan, DeepSeek, Mixtral, LLaMA, GLM series) and providing pre-training, post-training, datasets, evaluation, deployment, and RAG pipelines. It includes a listing of model configurations and corresponding tuning configs for different backbones.
How it works
The repository aggregates multiple fine-tuning configurations and data pipelines to enable incremental pretraining, full-model fine-tuning, LoRA, and QLoRA approaches. It references xtuner-based scripts, ms-swift workflows, and LLaMA-Factory-based guides for training and deployment. It also integrates RAG pipelines and evaluation frameworks.
Getting started
- Clone the repo:
git clone https://github.com/SmartFlowAI/EmoLLM.git - Follow sections in the README for rapid experience, data construction, incremental pretraining and fine-tuning, deployment, and RAG usage. The README points to detailed guides in various directories:
- quick_start/quick_start.md
- quick_start/Baby_EmoLLM.ipynb
- generate_data/tutorial.md
- xtuner_config/pt/README.md
- xtuner_config/README.md
- swift/README.md
- doc/GLM-4-9B-chat%20Lora%20微调（llama-factory）.md
- demo/README.md
- deploy/lmdeploy.md
- rag/README.md
- evaluate/README.md
Recent releases
- v0.6 (2025-05-18): Merging careyou into EmoLLM by @HaiyangPeng in https://github.com/SmartFlowAI/EmoLLM/pull/305; New Contributor: @HaiyangPeng
- v0.5 (2025-03-23): DOC updates and README/免责声明, Upload ps
- v0.4 (2024-10-21): EmoLLM V3.0 release notes and related dev updates
- v0.3 (2024-07-11): interlm2.5 update and Dev merge
- v0.2 (2024-06-28): Dev merge and GLM-4-9B 4-9b 文档 updates
Traction
Stargazers: 1769
Behind the repo
Not provided in the facts block.
Caveats
- License: MIT
- Created: 2024-01-11
- Last push: 2026-06-18
- Open issues: 5
- Language: Python
- Topics include dataset, depoly, evaluation, llm, post-training, the-big-model-of-mental-health