-
The UDPipe sentence splitter seems to be a bit too split-happy, creating many fragments. Is this dragging down performance of our BERT models? Furthermore, we put a lot of effort into splitting large …
-
- prevent this f…
-
Currently, when calculating overlap, we are first splitting previous segment by **sentences**, then try to fit as many sentences as possible into the overlap.
One should be able to configure this: ei…
-
**Is your feature request related to a problem? Please describe.**
When using the builtin tokenizer of INCEpTION, it sometimes does errors. It would be nice if the tokenization can be edited.
**De…
-
The sentence is incorrectly split after 'vol.':
> Other good editions are in **vol.** 4.
 or if sentences should be the inputs (has punctuation.)
This should be configurable so as to be able to better preserve…
-
Many thanks for your kind code sharing!
Could you provide the code to preprocess the data?
Or Could you give us your configurations to use CoreNLP?
Thanks again!
-
# Bug Report
## Installation Method
Installed via pip on a virtual environment.
## Environment
- **Open WebUI Version:** tested with 0.3.30 & 0.3.32
- **TTS backend:** happens with both A…
-
Hi all,
I am trying to use BlingFire for sentence splitting in Greek. As it makes many errors, I want to improve the "rules" it uses. How can I do this?
i want to try to port some rules from the…
-
Often we split up sentences after aligning into `aligned_subsegments`.
In the end we could run the pipeline again on these smaller subsegments now, get a new trellis shape and assign a new ratio base…