lancopku/label-words-are-anchors
Repository for Label Words are Anchors: An Information Flow Perspective for Understanding In-Context Learning
PythonMIT
Issues
- 4
attention_adapter.params.grad为None
#28 opened by CoverZhao - 2
- 1
求解释Saliency分数得到的tensor
#26 opened by Patrick-Ni - 19
求复现指导2.1
#14 opened by gaotongxue12138 - 2
How to calculate $I_l(i, j)$
#24 opened by lucky9-cyou - 2
Problem about S_{wq} computation
#25 opened by lucky9-cyou - 9
关于llama2-7b在3.1Anchor Re-weighting 相应代码的修改
#21 opened by Cooperx521 - 4
关于3.2 Anchor-Only Context Compression的问题
#23 opened by Cooperx521 - 9
关于Figure 1作图的疑问以及方法能否适用于Bert、Roberta类模型的疑问
#8 opened by lczx666 - 3
请问下显著性得分和注意力得分的区别是什么
#6 opened by ahxchzt - 3
关于论文Figure 5的疑问
#22 opened by qq31415926 - 3
关于锚点重加权
#20 opened by sumingfirst - 1
能否详细解释一下(7)式中的\hat{\mathbf{x}}与(8)式中的x的区别呀
#19 opened by zhiyunjiang - 12
请问以下如果想计算llama-2-7b在半精度下的显著性得分大概需要多少显存呢?
#11 opened by zhiyunjiang - 4
关于可视化的一些问题
#17 opened by fmm170 - 2
标签词的token超过1
#18 opened by Tincsvsv - 1
- 14
- 2
prefix_idxs如何确定
#15 opened by RENNY-Jenius - 1
请问predictor_classes.py程序中的问题
#13 opened by xsc1234 - 2
程序报错
#12 opened by RENNY-Jenius - 1
anchor reweighting on GPT-J
#10 opened by Tincsvsv - 2
Does this apply to the llama2 model?
#4 opened by Tincsvsv - 1
关于 Anchor Re-weighting 方法的疑问
#9 opened by OStars - 5
请问显著性得分的计算方法为什么是论文中写的那样
#7 opened by zhangzhen-research - 11
OverflowError: int too big to convert
#5 opened by UGUESS-lzx - 2
- 2
不是很懂label words在中间层中具体指的是什么
#1 opened by HuXinjing