Age | Commit message (Collapse) | Author |
|
|
|
The comment resulted from experimenting with running using distributed parallel nodes
|
|
|
|
spelling mistake
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Decoder
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Fix/ultimate
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Vaile reads code
|
|
|
|
|
|
|
|
|
|
add limited supervision training (10hr)
|
|
|
|
Plot
|
|
|
|
|
|
|
|
|
|
Refactor modularize
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Tokenizer
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Not distributed but still cool
|
|
|
|
|
|
For now decided against it, so I will overwrite the changes soon.
|
|
Slurm distributor attempt one
|
|
|
|
|
|
|
|
|
|
|
|
| ||
|| |_
|