Play with LLMs
Share how to train and evaluate large language models, and build interesting LLMs applications based on RAG, Agent, and Chain.
Ready to use | I code this so you don't have to!
- Mistral-8x7b-Instruct stably outputs Json Format, paired with Llamacpp grammar
- Mistral-8x7b-Instruct CoT Agent, Think step by steps
- Mistral-8x7b-Instruct ReAct Agent with tool call
- Llama3-8b-Instruct, transformers, vLLM and Llamacpp
- Llama3-8b-Instruct, CoT with vLLM
- Llama3-8b-Instruct, pure Chinese implementation of ReAct with tool call
- Chinese-Llama3-8b, DPO fine-tuning makes Llama3 more willing to speak Chinese
- llama-cpp-convert-GGUF, convert model quantification into GGUF format and upload huggingface
- AdvancedReAct
?In-depth LLMs | Pretraining, Fine-tuning, RLHF and ?>
- qlora-finetune-Baichuan-7B
Case display
| Mixtral 8x7b ReAct | Llama3-8b ReAct |
|---|
 |  |
