todo update
This commit is contained in:
parent
bef086ab50
commit
de98926d00
7
main.py
7
main.py
|
@ -12,16 +12,9 @@ from gfun.generalizedFunnelling import GeneralizedFunnelling
|
|||
|
||||
"""
|
||||
TODO:
|
||||
- Transformers VGFs:
|
||||
- scheduler with warmup and cosine
|
||||
- freeze params method
|
||||
- General:
|
||||
[!] zero-shot setup
|
||||
- CLS dataset is loading only "books" domain data
|
||||
- documents should be trimmed to the same length (for SVMs we are using way too long tokens)
|
||||
- Attention Aggregator:
|
||||
- experiment with weight init of Attention-aggregator
|
||||
- FFNN posterior-probabilities' dependent
|
||||
- Docs:
|
||||
- add documentations sphinx
|
||||
"""
|
||||
|
|
Loading…
Reference in New Issue