-
**Describe the bug**
I am trying to continue the training of the `news-forward` language model on domain specific data, which works in a jupyter notebook but fails when executing the code in a plain …
-
This is my current script
corpus: Corpus = NLPTaskDataFetcher.load_column_corpus(data_folder, columns,
train_file=train_path,
…
-
**Problem**
Specified GPU ('cuda:1') was ignored in `DocumentPoolEmbedding` since I'm using `ELMoEmbeddings` together with flair embeddings and in line 570 of `embeddings.py`, the device to use for e…
-
**Describe the bug**
As mentioned in the title, when I load the BERT embeddings from file, it reports an `UnicodeDecodeError`。
**To Reproduce**
The code is as follows:
```
from flair.embeddings…
-
In order to find the best parameters for different embeddings in flair training, parameter optimization script should be run on different embeddings. Parameters to be decided are the followings:
* hi…
-
**Describe the bug**
When I try to get a sentence embedding using BERT the output is an empty tensor
**To Reproduce**
``` python
import flair
document_embeddings = flair.embeddings.BertEmbedd…
-
I was wondering if it would be possible to add simple one-hot encoding of characters in a sentence. You already provide the following method, but this still requires training on a down-stream task:
`…
-
I am trying to do a simple binary text classification using `DocumentRNNEmbeddings` and `TextClassifier` where the embedding used is the `CharacterEmbeddings()`.
The model looks like so:
```
d…
-
WordEmbeddings work for me as expected, but calling .embed on a StackedEmbedding doesn't return any result:
```python
import flair, torch
from flair.embeddings import CharacterEmbeddings, StackedEm…
-
A clear and concise description of what you want to know:
I want to make a custom NER tagger that can be used inside of Flair to `.predict()` custom tags of words in a text I give to it. Very similar…