issues
search
pratyushasharma
/
laser
The Truth Is In There: Improving Reasoning in Language Models with Layer-Selective Rank Reduction
https://pratyushasharma.github.io/laser/
MIT License
358
stars
25
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
After your code is saved, the size of the weights is the same as the pre-trained ones, and no memory is saved. What is the reason for this?
#28
pursure-D
closed
1 month ago
6
Reproducing LLAMA-2 metrics
#27
sidhantls
closed
3 months ago
2
How to reproduce Figure 5 analysis in this paper?
#26
sanyalsunny111
closed
2 months ago
4
Testing
#25
Gurshaan25
closed
4 months ago
1
fix get_counterfact.py command
#23
sidhantls
closed
5 months ago
1
Problem Encountered During Reproduction
#22
ZY123-GOOD
opened
5 months ago
2
how to get base model accuracy
#21
pursure-D
closed
5 months ago
2
Generic model?
#20
forresti
closed
7 months ago
3
method of composing reductions across layers
#19
KTALS
closed
7 months ago
2
Llama2-7B + TruthfulQA reproduce issue
#18
JiwenJ
closed
7 months ago
8
Application to three-dimensional tensors
#17
xl-lei
opened
7 months ago
1
Potential improvements for evaluation
#15
BenjaminBossan
opened
7 months ago
1
what does the 'rate' parameters actually mean in code?
#14
KTALS
closed
7 months ago
2
Rank reduction using random matrix theory
#12
fernando-neto-ai
closed
8 months ago
1
Feature Request for Upcoming Refactoring
#9
dkmisra
opened
8 months ago
0
Rank-reduced models?
#8
turboderp
opened
8 months ago
4
Do you think it could work for MoE models like Mixtral?
#7
BadisG
closed
8 months ago
2
Question
#6
fakerybakery
opened
8 months ago
2
Where to Get the Dataset
#5
fakerybakery
closed
8 months ago
5
Mistral Support
#4
fakerybakery
opened
8 months ago
16
License
#3
fakerybakery
closed
8 months ago
6
What is the ETA on the code
#2
MrigankRaman
closed
8 months ago
2
Excellent work, looking forward to following up with further research!
#1
caiyuchen-ustc
closed
8 months ago
3