bentrevett/pytorch-seq2seq issues and pull requests

#209 - Query

Issue - State: open - Opened by malikfahadsarwar 15 days ago

#208 - encoder not handling variable length sequences properly

Issue - State: open - Opened by dsrub 2 months ago

#207 - module 'torchtext' has no attribute 'vocab'

Issue - State: open - Opened by WhaSukGO 3 months ago

#206 - Incorrect German Translation

Issue - State: open - Opened by timothy-geiger 9 months ago

#205 - fmt with black for consistency

Pull Request - State: closed - Opened by bentrevett 10 months ago

#204 - Merge re-write

Pull Request - State: closed - Opened by bentrevett 10 months ago

#203 - How to change seq2seq to graph2seq

Issue - State: closed - Opened by nhjclxc 11 months ago

#202 - Seq2seq: Input not matching Output (and big thanks)

Issue - State: closed - Opened by s2458588 about 1 year ago

#201 - Fwejiio

Pull Request - State: closed - Opened by SM0711 over 1 year ago

#200 - possible opposite explanation of hidden compared to output in notebook #3

Issue - State: closed - Opened by Hadar933 over 1 year ago

#199 - import

Issue - State: closed - Opened by bouchalhakim over 1 year ago

#198 - no module named 'torchtext.legacy'

Issue - State: closed - Opened by Hzzhang-nlp over 1 year ago - 2 comments

#197 - Notebook 1 <eos> problem.

Issue - State: closed - Opened by yusufsali61 over 1 year ago - 2 comments

#196 - How do you make this work on android?

Issue - State: closed - Opened by Lambdacreator over 1 year ago

#195 - Corrections to minor typos.

Pull Request - State: closed - Opened by StatQuest over 1 year ago

#194 - Why using tanh function

Issue - State: closed - Opened by karimmahalian almost 2 years ago - 3 comments

#193 - Updated

Pull Request - State: closed - Opened by RajasBharambe almost 2 years ago

#192 - Using pretrained BERT embedding

Issue - State: closed - Opened by dhurba-baral almost 2 years ago

#191 - Transformer ScaledDotProductAttention energy value on 16-bit Precision.

Issue - State: closed - Opened by ankitvad about 2 years ago - 3 comments

#190 - Question about changing params init from xavier to kaiming

Issue - State: closed - Opened by yzhang-github-pub over 2 years ago

#189 - Tutorial 6: [Attention is All You need] Different output at different batch size during Inference

Issue - State: closed - Opened by rajeevbaalwan over 2 years ago

#188 - Change preprocessing pipeline for torchtext 0.13. Field has been deprecated.

Pull Request - State: closed - Opened by marvinli00 over 2 years ago - 3 comments

#187 - Possible Inaccuracies in training script

Issue - State: closed - Opened by rrmina over 2 years ago

#186 - Question about how to resolve the out of vocabulary problem during encoding and decoding in tutorial 1

Issue - State: closed - Opened by liaomuquan over 2 years ago

#185 - torchtext recent version (0.12.0) doesn't support Field, BucketIterator

Issue - State: closed - Opened by manik2304 over 2 years ago - 4 comments

#184 - Question

Issue - State: closed - Opened by Letcise over 2 years ago

#183 - Custom Text Dataset

Issue - State: closed - Opened by moodhiaj over 2 years ago - 6 comments

#182 - [Bug] Tranformer Seq2Seq Have Wrong Inputs!

Issue - State: closed - Opened by bot66 almost 3 years ago - 2 comments

#181 - Thank you!

Issue - State: closed - Opened by asigalov61 almost 3 years ago - 1 comment

#180 - Multi-gpu might fail with Attention is all you need

Issue - State: closed - Opened by Aaron-Zhao123 almost 3 years ago - 3 comments

#179 - spacy load not loading

Issue - State: closed - Opened by neqkir almost 3 years ago - 1 comment

#178 - why this

Issue - State: closed - Opened by zhiqiangohuo almost 3 years ago

#177 - using CTC loss instead of Cross Entropy loss:

Issue - State: closed - Opened by kerolos about 3 years ago - 1 comment

#176 - Tutorial 6: Regarding attention plots

Issue - State: closed - Opened by yuvaraj91 about 3 years ago

#175 - Tut 6: TypeError with calculate_bleu_alt

Issue - State: closed - Opened by yuvaraj91 about 3 years ago - 1 comment

#174 - How easy can those models be converted into production TorchScript (PyTorch model inference pipeline)

Issue - State: closed - Opened by kerolos about 3 years ago

#173 - Tutorial 4 : AssertionError during the trainning

Issue - State: closed - Opened by WendkuuniArzouma about 3 years ago - 7 comments

#172 - Custom dataset: using Tabular dataset

Issue - State: closed - Opened by CrispenGari about 3 years ago

#171 - Change input data on the picture for the Multi-Head Attention.

Pull Request - State: closed - Opened by vasiliyeskin over 3 years ago

#170 - Tutorial 6: Attention is all you need

Issue - State: closed - Opened by Hannibal046 over 3 years ago - 3 comments

#169 - Tutorial 1: Differences between Encoder/Decoder in Seq2Seq Model

Issue - State: closed - Opened by michael-camilleri over 3 years ago - 3 comments

#168 - Question about tutorial 1 and 2 Decoder

Issue - State: closed - Opened by djaekim over 3 years ago - 1 comment

#167 - Tutorial 4: Why not set the padding_idx of nn.Embedding's to src_pad_token_idx or tgt_pad_token_idx respectively?

Issue - State: closed - Opened by yipliu over 3 years ago - 1 comment

#166 - Tutorial 6: Question to CrossEntropyLoss

Issue - State: closed - Opened by bpaulwitz over 3 years ago - 1 comment

#165 - Suggest ask the users (whoever is using this tutorial) to update their spaCy to 3.0 as 2.0 is suddenly very slow today

Issue - State: closed - Opened by cestwc over 3 years ago - 3 comments

#164 - Lesson 1, decoder to linear batch position

Issue - State: closed - Opened by marcpaga over 3 years ago - 4 comments

#163 - how to use BPTTIterator for Language Modeling

Issue - State: closed - Opened by StephennFernandes over 3 years ago

#162 - RuntimeError: Expected hidden[0] size (2, 1, 512), got [2, 128, 512] - Seq2Seq Model with PreTrained BERT Model

Issue - State: closed - Opened by Ninja16180 over 3 years ago

#161 - Seq2Seq Model with PreTrained BERT Model is Throwing Error During Training

Issue - State: closed - Opened by Ninja16180 over 3 years ago - 2 comments

#160 - Conversational Model Created Using Pre-Trained BERT Model is Throwing Error During Training

Issue - State: closed - Opened by Ninja16180 over 3 years ago - 4 comments

#159 - Tutorial 3: failed to run in google colab

Issue - State: closed - Opened by YiyiAm over 3 years ago - 6 comments

#158 - Tutorial 4: Decoder - the calculation of prediction

Issue - State: closed - Opened by actforjason over 3 years ago - 2 comments

#157 - Tutorial 6: PositionWiseFeedforwardLayer - fc_2 activation function

Issue - State: closed - Opened by dryng over 3 years ago - 1 comment

#156 - Tutorial 4: RuntimeError: 'lengths' argument should be a 1D CPU int64 tensor, but got 1D cuda:0 Long tensor

Issue - State: closed - Opened by STRZGR almost 4 years ago - 3 comments

#155 - Blank Outputs

Issue - State: closed - Opened by abdullahkhilji almost 4 years ago

#154 - Adding layers to the encoder and decoder of seq2seq

Issue - State: closed - Opened by dopu2k16 almost 4 years ago - 1 comment

#153 - Why the batch size is misplaced in the tensor?

Issue - State: closed - Opened by abdullahkhilji almost 4 years ago - 2 comments

#152 - Tutorial 6: Multihead Attention

Issue - State: closed - Opened by hartrials almost 4 years ago

#151 - Tutorial 1: understanding LSTM

Issue - State: closed - Opened by harshraj22 almost 4 years ago - 3 comments

#150 - Implement vectorized version of the BLEU score calculation for the Tr…

Pull Request - State: closed - Opened by azadyasar almost 4 years ago - 1 comment

#149 - Tutorial 3: The input of Attention

Issue - State: closed - Opened by yipliu almost 4 years ago - 3 comments

#148 - Tutorial 6: Multihead Attention

Issue - State: closed - Opened by Hannibal046 almost 4 years ago - 2 comments

#147 - How to add label smoothing

Issue - State: closed - Opened by Real-Music almost 4 years ago

#146 - Ways to Compute the Test Perplexity

Issue - State: closed - Opened by zhongpeixiang almost 4 years ago - 4 comments

#145 - Using datasets from manythings.org? Changes to be made?

Issue - State: closed - Opened by varunbanda about 4 years ago - 1 comment

#144 - can i get some help?thanks!

Issue - State: closed - Opened by bufuchangfeng about 4 years ago - 2 comments

#143 - maybe a mistake in t6

Issue - State: closed - Opened by bufuchangfeng about 4 years ago - 1 comment

#142 - Tutorial 6: why dropout here?

Issue - State: closed - Opened by xxoospring about 4 years ago - 3 comments

#141 - Question about beam search using convolution and transformer

Issue - State: closed - Opened by dksifoua about 4 years ago - 4 comments

#140 - Process finished with exit code 139 (interrupted by signal 11: SIGSEGV)

Issue - State: closed - Opened by cjjjy about 4 years ago - 1 comment

#139 - Tutorial 6: using own data, CUBLAS status alloc failed

Issue - State: closed - Opened by izaskr about 4 years ago - 3 comments

#138 - Tutorial 6: question about dimension of trg_mask and src_mask in forward() of Encoder/Decoder/EncoderLayer/DecorderLayer.

Issue - State: closed - Opened by byc1234 about 4 years ago - 2 comments

#137 - Tutorial 4: question about a comment in forward() in Encoder()

Issue - State: closed - Opened by byc1234 about 4 years ago - 2 comments

#136 - Using trg[: ,;-1] during training

Issue - State: closed - Opened by wajihullahbaig about 4 years ago - 7 comments

#135 - getting different shapes for Q,K in Multihead attention values pytorch-seq2seq in DataParallel

Issue - State: closed - Opened by VinACE about 4 years ago

#134 - Model actually corrects "ground truth" (6 - Attention is All You Need).

Issue - State: closed - Opened by sechegaray about 4 years ago - 1 comment

#133 - tut 6 - when slicing the <eos> token off from trg before feeding it into the model

Issue - State: closed - Opened by JaeyoonChun about 4 years ago - 4 comments

#132 - There is no trg when for evaluation, but the tutorial has used this for input: trg_len

Issue - State: closed - Opened by liuchanglab about 4 years ago - 1 comment

#131 - RuntimeError: The size of tensor a (1024) must match the size of tensor b (2048) at non-singleton dimension 3

Issue - State: closed - Opened by VinACE about 4 years ago - 1 comment

#130 - attention notebook: energy = energy.masked_fill(mask == 0, float("-1e20")) RuntimeError: The size of tensor a (1024) must match the size of tensor b (512) at non-singleton dimension 3

Issue - State: closed - Opened by VinACE about 4 years ago - 3 comments

#129 - In seq2seq pytorchTypeError: 'int' object is not callable in attention is all you need notebook

Issue - State: closed - Opened by VinACE over 4 years ago - 1 comment

#128 - How to matrix multiply two encoded representations?

Issue - State: closed - Opened by jlim13 over 4 years ago - 1 comment

#127 - Why no softmax for prediciton (eg: Notebook: 3)

Issue - State: closed - Opened by macabdul9 over 4 years ago - 2 comments

#126 - Tutorial 3 : Aligment Model not the same as the paper

Issue - State: closed - Opened by xxoospring over 4 years ago - 2 comments

#125 - I can not import bleu score.

Issue - State: closed - Opened by GlazeDonuts over 4 years ago - 2 comments

#124 - Embedded Seq Dim

Issue - State: closed - Opened by macabdul9 over 4 years ago

#123 - Tutorial 6: Batch inference to speed up the translation

Issue - State: closed - Opened by ruathudo over 4 years ago - 4 comments

#122 - In Tutorial 1 target sequence len is used at time of evaluation

Issue - State: closed - Opened by dineshkh over 4 years ago - 1 comment

#120 - Comment misplaced

Issue - State: closed - Opened by binishjoshi over 4 years ago - 2 comments

#119 - Confusion about masking in transformer

Issue - State: closed - Opened by Yupei-Du over 4 years ago - 3 comments
Labels: question, waiting-reply

#114 - BERT Model for Tokenization and Initialization of embedding layer

Issue - State: closed - Opened by kr-sundaram over 4 years ago - 1 comment
Labels: waiting-reply

#113 - remove unused import `TranslationDataset`

Pull Request - State: closed - Opened by sheiun over 4 years ago

#106 - Tutorial 6: byte pair encoding

Issue - State: closed - Opened by brand17 over 4 years ago - 1 comment
Labels: enhancement

#100 - Transformer: Use separate LayerNorm layers for attention and feedforward blocks

Pull Request - State: closed - Opened by rakeshchada over 4 years ago - 3 comments

#99 - Fix Colab links for notebooks 5 & 6

Pull Request - State: closed - Opened by HelgeS over 4 years ago - 1 comment

#98 - Right to Left implemention

Issue - State: closed - Opened by aimanmutasem over 4 years ago - 6 comments

#97 - [Tutorial 3] Minor typo on bidirectional GRU definition

Issue - State: closed - Opened by dataframing over 4 years ago - 1 comment

#96 - there is a little error in the make_trg_mask function in tutorial 6

Issue - State: closed - Opened by zhangjianzhang over 4 years ago - 2 comments

#95 - Un expected results

Issue - State: closed - Opened by aimanmutasem over 4 years ago - 2 comments

#94 - Understanding how training in batch occurs

Issue - State: closed - Opened by kushalj001 over 4 years ago - 3 comments

GitHub / bentrevett/pytorch-seq2seq issues and pull requests