clovaai/donut issues and pull requests

#321 - zero division error

Issue - State: open - Opened by saimadhav952 9 days ago

#320 - High inference lantency

Issue - State: open - Opened by Altimis 12 days ago

#319 - Custom dataset for DONUT

Issue - State: open - Opened by Dealibros about 1 month ago - 1 comment

#318 - ground_truth = json.loads(sample["ground_truth"]) KeyError: 'ground_truth'

Issue - State: open - Opened by monarchmio about 1 month ago

#317 - Training doesn't start

Issue - State: open - Opened by Psoriaz about 2 months ago

#317 - Training doesn't start

Issue - State: open - Opened by Psoriaz about 2 months ago

#316 - Got wrong inference result

Issue - State: open - Opened by TinhAnhGitHub about 2 months ago

#316 - Got wrong inference result

Issue - State: open - Opened by TinhAnhGitHub about 2 months ago - 1 comment

#315 - Can someone provide the complete pip list for training a Japanese dataset?

Issue - State: open - Opened by monarchmio about 2 months ago - 1 comment

#314 - The predicion do not display anything at all when I was training

Issue - State: open - Opened by monarchmio about 2 months ago

#313 - Is it normal to see a lot of 'encoder.layers weight were not used when initializing DonutModel' during training?

Issue - State: open - Opened by monarchmio about 2 months ago

#312 - DocVQA input input_ids at training time

Issue - State: open - Opened by cccccckt 3 months ago - 1 comment

#311 - Issue with Fine-Tuning: Incorrect Answers When Querying Specific Questions

Issue - State: open - Opened by dikyridhlo 3 months ago - 2 comments

#310 - How to interpret the results of the Japanese document?

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

#309 - Hindi Synthdog

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

#308 - What should be the configuration of the machine to train the model?

Issue - State: open - Opened by anant996 5 months ago

#307 - How many documents(invoices) are required for training model for document information extraction?

Issue - State: open - Opened by anant996 5 months ago

#306 - Early Stopping

Issue - State: closed - Opened by svchostdotexe 5 months ago

#305 - How to improve OCR accuracy for Japanese characters?

Issue - State: closed - Opened by kirby707 6 months ago - 2 comments

#304 - Performance of the model

Issue - State: open - Opened by DriraYosr 6 months ago - 1 comment

#304 - Performance of the model

Issue - State: open - Opened by DriraYosr 6 months ago

#303 - Donut generate ONLY <s><s>...<s></s>

Issue - State: open - Opened by SerenaTetart 6 months ago - 7 comments

#302 - Why is the output of the intermediate verification empty after training?

Issue - State: open - Opened by zweiqi 6 months ago

#301 - Introducing the scores to the outputs

Pull Request - State: open - Opened by matteocacciola 6 months ago

#300 - Where is the fine-tuned model?

Issue - State: closed - Opened by sdip15fa 7 months ago

#299 - Can synthdog insert text for a specified bbox?

Issue - State: open - Opened by Supercarry-khr 7 months ago - 1 comment

#299 - Can synthdog insert text for a specified bbox?

Issue - State: open - Opened by Supercarry-khr 7 months ago

#298 - not work this app.py

Issue - State: open - Opened by sssssshf 7 months ago - 1 comment

#297 - Not getting prediction correctly using the model trained on the custom dataset (similar format as CORD-V2 dataset)

Issue - State: open - Opened by SiriusPoint 7 months ago - 7 comments

#296 - donut inference시 sub task가 변경?

Issue - State: open - Opened by kittyLunar 7 months ago

#295 - Update donut-python Python Package to be compatible with latest versions of transformers

Issue - State: open - Opened by agovin9812 8 months ago

#294 - Classification inference

Issue - State: open - Opened by SNavgale 8 months ago

#293 - confidence 값의 공식적인 지원

Issue - State: open - Opened by HwangSeyoon 9 months ago

#292 - How to extract complete text from the document?

Issue - State: open - Opened by vikasr111 9 months ago - 3 comments

#291 - Multi GPU support for fine tuning

Issue - State: closed - Opened by SNavgale 9 months ago

#290 - custom json schema - ASAP

Issue - State: open - Opened by TheFaheem 9 months ago - 2 comments

#289 - Error "A configuraton of type donut cannot be instantiated because not both `encoder` and `decoder` sub-configurations are passed" when run inference after finetuned docvqa without pushing to hugging face?

Issue - State: open - Opened by phuchm 9 months ago - 1 comment

#288 - Does synthdog data has MiT or afl-3.0 license?

Issue - State: open - Opened by becxer 9 months ago - 1 comment

#287 - Problem with getting predictions

Issue - State: open - Opened by jdrzj 10 months ago - 7 comments

#286 - Request: Dataset and pretrained model for language detection

Issue - State: open - Opened by turian 10 months ago

#285 - Integrate a customized internal OCR engine to Donut

Issue - State: open - Opened by Altimis 10 months ago - 2 comments

#284 - VisionEncoderDecoderModel convert

Issue - State: open - Opened by sjtu-cz 10 months ago - 2 comments

#283 - The latest update has the model weights twice the embedding dim size of the actual model installed through github or pip

Issue - State: open - Opened by Samartha27 10 months ago - 2 comments

#282 - Prediction and Answer differ by dataset-specific tag

Issue - State: closed - Opened by ftkeys 10 months ago - 1 comment

#281 - DOCVQA data set format ?

Issue - State: open - Opened by tzktz 11 months ago - 3 comments

#280 - getting no module named lightning module when trying to run the fine tuning code in train.py file of donut model.

Issue - State: open - Opened by svocdfrockz 11 months ago - 1 comment

#279 - How to handle multi page invoices

Issue - State: open - Opened by DNFSiF 11 months ago - 4 comments

#278 - Trying to run DOCVQA dataset

Issue - State: open - Opened by srgautam9 11 months ago

#277 - Bounding boxes required for pretraining?

Issue - State: open - Opened by mustaszewski 11 months ago - 1 comment

#276 - Could not find image processor class in the image processor config or the model config.

Issue - State: open - Opened by felixepiklah 12 months ago - 1 comment

#275 - Idea: Freezing SwinEncoder and fine-tuning BARTdecoder only on custom data

Issue - State: open - Opened by jackkwok 12 months ago - 4 comments

#274 - Question about the special token map

Issue - State: closed - Opened by RAY-RaY-R about 1 year ago - 2 comments

#272 - Donut Return Output even With Blank Image

Issue - State: open - Opened by wdprsto about 1 year ago - 5 comments

#271 - Simple questions answering

Issue - State: open - Opened by shersoni610 about 1 year ago

#270 - Random prediction and wrong prediction in repeated characters

Issue - State: open - Opened by Asha-12502 about 1 year ago - 5 comments

#269 - Where can I find the dataset used for training Document Visual Question Answering model

Issue - State: open - Opened by Coder-Vishali about 1 year ago

#268 - Bff/token optimisation

Pull Request - State: open - Opened by benjaminfh about 1 year ago - 8 comments

#267 - Input type (float) and bias type (struct c10::BFloat16) should be the same

Issue - State: open - Opened by Coder-Vishali about 1 year ago - 4 comments

#266 - json2token performance

Issue - State: open - Opened by benjaminfh about 1 year ago - 1 comment

#265 - fine-tuning on docvqa ,anls only 40%

Issue - State: open - Opened by ShuoZhang2003 about 1 year ago - 1 comment

#264 - Are there any acceleration solutions for donut deployment?

Issue - State: open - Opened by sjtu-cz about 1 year ago

#263 - Dataset Loader didn't work properly on Kaggle

Issue - State: closed - Opened by wdprsto about 1 year ago - 1 comment

#262 - Can donut support batch inference?

Issue - State: closed - Opened by sjtu-cz about 1 year ago - 2 comments

#261 - Documentation for Synthdog?

Issue - State: open - Opened by parthch11 about 1 year ago

#260 - Model Consistently Mispredicting Specific Character in Invoice Number

Issue - State: open - Opened by Codedrainer about 1 year ago

#259 - How to annotate and train donut for extracting all dates (unknown number of dates)

Issue - State: open - Opened by Anas-Khayata about 1 year ago

#258 - details is not ideal

Issue - State: open - Opened by chopin1998 about 1 year ago - 5 comments

#257 - Is UIPATH AI Center Document Understanding Process also OCR free like Donut

Issue - State: open - Opened by Sridhar-Ranganaboina about 1 year ago

#256 - Two types of documents in one model?

Issue - State: open - Opened by henkish about 1 year ago - 1 comment

#255 - The provided lr scheduler `LambdaLR` doesn't follow PyTorch's LRScheduler API. You should override the `LightningModule.lr_scheduler_step` hook with your own logic if you are using a custom LR scheduler.

Issue - State: open - Opened by goseesomething about 1 year ago - 2 comments

#254 - CORD-v2 accuracy much lower than the paper's results

Issue - State: closed - Opened by kevinmeetooa about 1 year ago - 1 comment

#253 - Complete text

Issue - State: open - Opened by VagnerBelfort about 1 year ago - 3 comments

#252 - Couldn't connect to 'https://huggingface.co'

Issue - State: open - Opened by mmhzlrj about 1 year ago - 1 comment

#251 - Pin timm version in package requirements

Pull Request - State: open - Opened by mberr about 1 year ago

#250 - dataset script missing error

Issue - State: open - Opened by segaranp about 1 year ago - 1 comment

#249 - How to config synthdog for much more longer text, like total length about 1024-2048

Issue - State: open - Opened by CheungZeeCn about 1 year ago - 2 comments

#248 - Handling Variable Element Presence in Parsing Document Task

Issue - State: open - Opened by abdelaziz-jaddi about 1 year ago

#247 - Texts in "swenglish"?

Issue - State: closed - Opened by henkish about 1 year ago - 2 comments

#246 - Yup

Issue - State: open - Opened by Freddie504 about 1 year ago - 1 comment

#245 - Query Regarding Tree-Based Accuracy Calculation in Donut Model

Issue - State: open - Opened by Mann1904 about 1 year ago

#244 - Is it possible to convert the donut model to openvino format

Issue - State: open - Opened by gsarathkumar92 about 1 year ago

#243 - make sure `_init_weights` is implemented for <class 'donut.model.DonutModel'>

Issue - State: open - Opened by asfansajid123 about 1 year ago - 8 comments

#242 - size mismatch error

Issue - State: open - Opened by yaoliUoA about 1 year ago - 3 comments

#241 - validation loss does not decrease

Issue - State: open - Opened by Mann1904 about 1 year ago - 2 comments

#240 - CUDA out of memory

Issue - State: open - Opened by ghost over 1 year ago - 1 comment

#239 - Multipage docs

Issue - State: open - Opened by abaranovskis-redsamurai over 1 year ago - 7 comments

#238 - How to estimate the required video memory?

Issue - State: closed - Opened by chopin1998 over 1 year ago - 1 comment

#237 - ValueError: `num_beams` is set to 1

Issue - State: open - Opened by yonlas over 1 year ago - 1 comment

#236 - Difficulties finetuning for another language

Issue - State: open - Opened by lauraminkova over 1 year ago - 12 comments

#235 - your gradio web demo does not work, please fix it

Issue - State: open - Opened by PigLookingatMoon over 1 year ago

#234 - Python 3.10 throws an error related to swintransformer

Issue - State: open - Opened by thistleknot over 1 year ago - 5 comments

#233 - Training on cpu

Issue - State: open - Opened by justOleh over 1 year ago - 1 comment

#232 - Where the output of swin diffused with the text->1.At the starting of Bart encoder,2. cross attention(K,V from swin,Q from attention) of second attention of Bart encoder,3.directly the decoder part of BART

Issue - State: open - Opened by shubham953 over 1 year ago - 3 comments

#231 - feat: update pl version supports

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

#230 - Converting trained Donut model to a single onnx file

Issue - State: open - Opened by TanvirHundredOne over 1 year ago - 2 comments

#229 - Issue with tokenizing '1' preceded by a char

Issue - State: open - Opened by arnaudstiegler over 1 year ago - 7 comments

#228 - How to serve trained donut model

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

#227 - Easier to fine tune using this repository code or Transformers and nielsr code?

Issue - State: open - Opened by DoctorSlimm over 1 year ago - 3 comments

#226 - Could i train/ fineturn donut for text direction task?

Issue - State: open - Opened by chopin1998 over 1 year ago

#225 - Fine-tuning on Apple M1/M2

Issue - State: closed - Opened by abaranovskis-redsamurai over 1 year ago - 3 comments

GitHub / clovaai/donut issues and pull requests