- The rapid development of large models has provided many possibilities for downstream tasks using LLM.
Finetune LLM
- We fine-tuned the Llama3 8B language model based on open-source datasets.
- Further optimization on different private datasets to achieve more accurate responses for various tasks.

Smart Healthcare
- Based on Open Medical Image Datasets, we fine-tuned the Multimodal Large Model Idefics2.
- Performing image-based question answering to achieve preliminary diagnoses based on medical images.

Chongqing dialect recognition
- We collected approximately 40 hours of Chongqing dialect.
- We trained a lightweight model.