⚡Megatron LLM Hub
←返回搜索
Twitter

称把某方法套到 GLM/DeepSeek 等开源权重模型上,效果优于微调小模型

llm
Instead of using fine-tuned less intelligent models. Apply this on any decent open weights model like GLM/DeepSeek, and get substantially better results. Cheaper, effective, fast.