Name: Leander Girrbach
Type: User
Company: University of Tübingen
Bio: Hi there! I am a PhD student at the Technical University Munich (TUM). I'm interested in deep learning.
Location: Munich, Germany
Blog: lgirrbach.github.io
Leander Girrbach's Projects
Contains modification of hard transducers to predict cognate reflexes (SIGTYP 2022 shared task). The encoder is changed in order to process multiple-sequence-alignment grids of multiple cognates instead of single sequences.
Implementation of a straight-through gradient wrapper to allow for discrete latent representations. Provides binary discretizer which maps hidden representations to {0, 1} and a learnable multi-value discretizer, which maps hidden activations to their closest value in a set of given size.
Python implementation of the many-to-many aligner proposed by Jiampojamarn et al. (2007): Applying Many-to-Many Alignments and Hidden Markov Models to Letter-to-Phoneme Conversion
End-to-end trainable autoregressive and non-autoregressive transducers using hard attention
An Extendible (General) Continual Learning Framework based on Pytorch - official codebase of Dark Experience for General Continual Learning
Proof-of-concept implementation of QRNN Layer in Pytorch. Recurrent pooling is implemented purely in PyTorch, no Python loop and also no extra CUDA code.
LSTM model for sequence labelling. Includes option to expand input length by constant. Currently implements cross-entropy, crf, and ctc losses.
Code for our SIGMORPHON 2023 shared task description paper: Tü-CL at SIGMORPHON 2023: Straight-Through Gradient Estimation for Hard Attention
Code for our SIGMORPHON 2023 shared task description paper: Tü-CL at SIGMORPHON 2023: Straight-Through Gradient Estimation for Hard Attention
TüReuth Legal Code for SemEval 2023 Task 6: LegalAI.