.ds Aq ’
lstmtraining --continue_from train_output_dir/continue_from_lang.lstm --old_traineddata bestdata_dir/continue_from_lang.traineddata --traineddata train_output_dir/lang/lang.traineddata --max_iterations NNN --debug_interval 0|-1 --train_listfile train_output_dir/lang.training_files.txt --model_output train_output_dir/newlstmmodel
lstmtraining(1) trains LSTM-based networks using a list of lstmf files and starter traineddata file as the main input. Training from scratch is not recommended to be done by users. Finetuning (example command shown in synopsis above) or replacing a layer options can be used instead. Different options apply to different types of training. Read the [training documentation](m[blue]https://tesseract-ocr.github.io/tessdoc/TrainingTesseract-4.00.htmlm[]) for details.
Aq--debug_interval Aq
How often to display the alignment. (type:int default:0)
Aq--net_mode Aq
Controls network behavior. (type:int default:192)
Aq--perfect_sample_delay Aq
How many imperfect samples between perfect ones. (type:int default:0)
Aq--max_image_MB Aq
Max memory to use for images. (type:int default:6000)
Aq--append_index Aq
Index in continue_from Network at which to attach the new network defined by net_spec (type:int default:-1)
Aq--max_iterations Aq
If set, exit after this many iterations. A negative value is interpreted as epochs, 0 means infinite iterations. (type:int default:0)
Aq--target_error_rate Aq
Final error rate in percent. (type:double default:0.01)
Aq--weight_range Aq
Range of initial random weights. (type:double default:0.1)
Aq--learning_rate Aq
Weight factor for new deltas. (type:double default:0.001)
Aq--momentum Aq
Decay factor for repeating deltas. (type:double default:0.5)
Aq--adam_beta Aq
Decay factor for repeating deltas. (type:double default:0.999)
Aq--stop_training Aq
Just convert the training model to a runtime model. (type:bool default:false)
Aq--convert_to_int Aq
Convert the recognition model to an integer model. (type:bool default:false)
Aq--sequential_training Aq
Use the training files sequentially instead of round-robin. (type:bool default:false)
Aq--debug_network Aq
Get info on distribution of weight values (type:bool default:false)
Aq--randomly_rotate Aq
Train OSD and randomly turn training samples upside-down (type:bool default:false)
Aq--net_spec Aq
Network specification (type:string default:)
Aq--continue_from Aq
Existing model to extend (type:string default:)
Aq--model_output Aq
Basename for output models (type:string default:lstmtrain)
Aq--train_listfile Aq
File listing training files in lstmf training format. (type:string default:)
Aq--eval_listfile Aq
File listing eval files in lstmf training format. (type:string default:)
Aq--traineddata Aq
Starter traineddata with combined Dawgs/Unicharset/Recoder for language model (type:string default:)
Aq--old_traineddata Aq
When changing the character set, this specifies the traineddata with the old character set that is to be replaced (type:string default:)
lstmtraining(1) was first made available for tesseract4.00.00alpha.
Main web site: m[blue]https://github.com/tesseract-ocrm[] Information on training tesseract LSTM: m[blue]https://tesseract-ocr.github.io/tessdoc/TrainingTesseract-4.00.htmlm[]
Copyright (C) 2012 Google, Inc. Licensed under the Apache License, Version 2.0
The Tesseract OCR engine was written by Ray Smith and his research groups at Hewlett Packard (1985-1995) and Google (2006-present).