allenai/bi-att-flow
Bi-directional Attention Flow (BiDAF) network is a multi-stage hierarchical process that represents context at different levels of granularity and uses a bi-directional attention flow mechanism to achieve a query-aware context representation without early summarization.
The implementation targets machine reading comprehension on the SQuAD dataset using TensorFlow r0.11, with GloVe embeddings and NLTK preprocessing. Training involves ~2.5M parameters requiring 12GB+ GPU memory (convergence at ~18k steps), while multi-GPU parallelization is supported for distributed inference across multiple cards. Pre-trained weights are available via CodaLab for reproducible evaluation against official metrics, achieving 77.3% F1 on single models and 80.7% on ensembles.
1,540 stars. No commits in the last 6 months.
Stars
1,540
Forks
672
Language
Python
License
Apache-2.0
Category
Last pushed
May 31, 2023
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/nlp/allenai/bi-att-flow"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Related tools
carpedm20/MemN2N-tensorflow
"End-To-End Memory Networks" in Tensorflow
domluna/memn2n
End-To-End Memory Network using Tensorflow
HKUST-KnowComp/R-Net
Tensorflow Implementation of R-Net
localminimum/QANet
A Tensorflow implementation of QANet for machine reading comprehension
YerevaNN/Dynamic-memory-networks-in-Theano
Implementation of Dynamic memory networks by Kumar et al. http://arxiv.org/abs/1506.07285