hiyouga / FastEdit

🩹Editing large language models within 10 seconds⚡

hiyouga/FastEdit Issues

It seems like there is a ignored value in delta calculation?
Closed 2 months ago3
Why modifying down_proj in llama?
Updated 3 months ago1
[Llama-2-7b-chat] RuntimeError: expected scalar type Float but found Half
Updated 7 months ago7
使用在线量化的baichuan 13b chat 报错 LookupError: model.layers.5.mlp.down_proj.weight
Updated 7 months ago3
Llama-2-7b-chat - RuntimeError: Inference tensors cannot be saved for backward
Closed 7 months ago2
训练方式和LLaMA-Efficient-Tuning-main区别
Updated 8 months ago
请问编辑后的模型储存在哪里了
Updated 8 months ago3
运行时报错
Updated 9 months ago2
RuntimeError: computing v Vector
Updated 9 months ago
qwen support
Updated 9 months ago1
Would you consider supporting ChatGLM2-6B?
Updated 10 months ago5
显存占用
Closed 10 months ago1
this is not a good idea, may lead to severe overfitting.
Updated 10 months ago1
LLaMA-2-7b-chat Editing failed
Updated 10 months ago
数据集格式
Closed 10 months ago
编辑完的baichuan-13b该如何保存
Updated 10 months ago3
Is there any way to apply this interesting algorithm to the chatGLM-6B or chatGLM2-6B models?
Updated 10 months ago2
单张80G卡编辑7B模型报显存不足想请教一下如何单机多卡去run
Updated 10 months ago2
这种编辑的方式有副作用吗？比如模型遗忘问题
Updated 10 months ago2
A little mistake in HyperParams
Closed a year ago1
Error occurs when editing Baichuan-13B
Updated a year ago2
错误：TypeError: can't convert cuda:0 device type tensor to numpy.
Closed a year ago1
编辑baichuan13b的时候报错NotImplementedError
Closed a year ago
请教如何配置config？
Closed a year ago1
初步看起来，这个的原理是，捕捉到两个data在模型内部的参数diff？
Updated a year ago