issues
search
pmhalvor
/
fgsa
Main repo for my Masters thesis on Fine Grained Sentiment Analysis
4
stars
0
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
FgFlex: debug task layers
#52
pmhalvor
closed
2 years ago
0
FgFlex: debug cnn dim and kernel size
#51
pmhalvor
closed
2 years ago
0
FgFlex: concat attention outputs w/ correct task convolution outputs
#50
pmhalvor
closed
2 years ago
0
Debug: RACL cnn_dim + allow kernel size tuning
#49
pmhalvor
closed
2 years ago
0
RACL gold transmission
#48
pmhalvor
closed
2 years ago
0
FgFlex: task cnn concat w/ embeddings
#47
pmhalvor
closed
2 years ago
0
Split cnn inspired by first layer of IMN
#46
pmhalvor
closed
2 years ago
0
Fix stack for FgFlex
#45
pmhalvor
closed
2 years ago
0
Gold transmission RACL
#44
pmhalvor
closed
2 years ago
0
Statistical inference
#43
pmhalvor
opened
2 years ago
0
`split_cnn` for FgFlex
#42
pmhalvor
closed
2 years ago
0
Build metrics (tail=1,2,3) distribution plotter for runs=[ ]
#41
pmhalvor
opened
2 years ago
0
Working FgFlex implmentation
#40
pmhalvor
closed
2 years ago
0
Document and sync parameters for each model
#39
pmhalvor
opened
2 years ago
0
Gold transmission for RACL
#38
pmhalvor
closed
2 years ago
1
Study RACL and IMN
#37
pmhalvor
closed
2 years ago
1
Fix shared/relations components optimizer assignment
#36
pmhalvor
closed
2 years ago
0
Fix/optimize shared layers
#35
pmhalvor
closed
2 years ago
1
New RACL attentions per stack
#34
pmhalvor
closed
2 years ago
1
Choose runs to show smooth plots
#33
pmhalvor
opened
2 years ago
0
Sanity check of subtasks in BertHead (base)
#32
pmhalvor
closed
2 years ago
0
Study
#31
pmhalvor
closed
2 years ago
0
Dev/metric
#30
pmhalvor
closed
2 years ago
0
Study
#29
pmhalvor
closed
2 years ago
0
Dev/racl
#28
pmhalvor
closed
2 years ago
1
Precision focused loss every 3. epochs, w/ recall based loss on others
#27
pmhalvor
opened
2 years ago
0
Code clean-up
#26
pmhalvor
opened
2 years ago
0
Update README
#25
pmhalvor
opened
2 years ago
0
Debug/0 metric
#24
pmhalvor
closed
2 years ago
0
FgFlex: attention based subtasks architecture
#23
pmhalvor
closed
2 years ago
0
Dev/imn
#22
pmhalvor
closed
2 years ago
1
Dev/data
#21
pmhalvor
closed
2 years ago
1
Make sure imn doesn't fail
#20
pmhalvor
closed
2 years ago
0
Initial hyper parameter searching
#19
pmhalvor
closed
2 years ago
0
Mimic RACL like cross attention between tasks
#18
pmhalvor
closed
2 years ago
0
Debugging ideas for 0 metric
#17
pmhalvor
opened
2 years ago
1
Compare to runs from exam IN5550
#16
pmhalvor
closed
2 years ago
0
Current status of IMN form SAGA
#15
pmhalvor
closed
2 years ago
0
Mimic IMN like information sharing between tasks
#14
pmhalvor
closed
2 years ago
0
Write about WWNLP as a baseline
#13
pmhalvor
closed
2 years ago
0
Test BertHead with unique learning rates, but also dropout
#12
pmhalvor
opened
2 years ago
1
March preprocessing bug
#11
pmhalvor
closed
2 years ago
1
Run studies through slurm
#10
pmhalvor
closed
2 years ago
0
Build FgsaLSTM
#9
pmhalvor
closed
2 years ago
0
Build BertTarget, predicting only targets as linear output w/ Bert head
#8
pmhalvor
closed
2 years ago
0
Start writing about experiment
#7
pmhalvor
closed
2 years ago
0
Get BertSimple to run on Fox GPUs
#6
pmhalvor
closed
2 years ago
0
Build BertHead
#5
pmhalvor
closed
2 years ago
3
Get BertSimple to give meaningful results (above 0)
#4
pmhalvor
closed
2 years ago
0
Implement Travis CI
#3
pmhalvor
closed
2 years ago
1
Next