-
Hi, thanks a lot for sharing your codes.
I am trying to reproduce preprocessed data by following the instruction on `preprocessing` folder, but it looks like it does not create the same file as rel…
-
Both [Wikipedia Retriever (Used For Open Squad)
](http://parl.ai.s3-website.us-east-2.amazonaws.com/docs/zoo.html#wikipedia-retriever-used-for-open-squad) and [Wikipedia Retriever (Used For Wizard Of…
-
I was trying to run the example below after installing corenlp "./install_corenlp.sh"
from drqa.tokenizers import CoreNLPTokenizer
tok = CoreNLPTokenizer()
tok.tokenize('hello world').words() # …
-
Hi,
I need to know confidence level of prediction in order to return a default answer if confidence too low.
-
I can't seem to get this working on the most recent version of DrQA. when the server is started, it goes through the steps of loading up the DrQA system but gets stuck on the "[ Initializing tokenizer…
-
Is there are way for DrQA to give me the specific sentence in the document context where it found the answer? Sometimes the context will contain multiple mentions of the DrQA answer and right now I ha…
-
My original motivation was to make Parlai support Chinese, so I digged some information about Spacy. In Parlai code, there is a line spacy.load("en") , so I'm looking for something like spacy.load("ch…
-
I saw in README that reading comprehension model sets predict_raw_text=True, while the pipeline model sets predict_raw_text=False. Are there any differences?
-
Hi !
Nice project, very interresting.
Do you have an interactive mode ?
Like DrQA and its pre-trained model.
Otherwise, after training the model, how can I use this algorithm ?
Thanks a lo…
-
I encountered this error.
THCudaCheck FAIL file=/opt/conda/conda-bld/pytorch_1502004572321/work/torch/lib/THC/generic/THCStorage.cu line=66 error=46 : all CUDA-capable devices are busy or unavaila…