`j-ddlZddlmZddlmZddlZddlmZddlm Z ddl m Z dd l m Z mZmZmZmZmZmZej(d d d d  e ddd dd e ddeddddeddddeddddedddd edd!d"d#fd$ej*d%ed&ed'eed(eed)eed*ed+efd,Zd-Zy).N)Path)Optional)msg)pretrain) load_config)ArgOptapp import_codeparse_config_overrides setup_gpushow_validation_errorrT)allow_extra_argsignore_unknown_options)context_settings.zPath to config fileF)helpexistsdir_okay allow_dashz+Directory to write weights to on each epoch)rz--codez-czNPath to Python file with additional code (registered functions) to be importedz --resume-pathz-rz;Path to pretrained weights from which to resume pretrainingz--epoch-resumez-erzuThe epoch to resume counting from when using --resume-path. Prevents unintended overwriting of existing weight files.z--gpu-idz-gzGPU ID or -1 for CPUz --skip-lastz-LzSkip saving model-last.binctx config_path output_dir code_path resume_path epoch_resumeuse_gpu skip_lastc t|j}t|t||||t |t j d|t|5t||d} ddd j} | jdst jdd|js*|jd t jd || j|d z t jd t!| ||||d|t jdy#1swYxYw)a Pre-train the 'token-to-vector' (tok2vec) layer of pipeline components, using an approximate language-modelling objective. Two objective types are available, vector-based and character-based. In the vector-based objective, we load word vectors that have been trained using a word2vec-style distributional similarity algorithm, and train a component like a CNN, BiLSTM, etc to predict vectors which match the pretrained ones. The weights are saved to a directory after each epoch. You can then pass a path to one of these pretrained weights files to the 'spacy train' command. This technique may be especially helpful if you have little labelled data. However, it's still quite experimental, so your mileage may vary. To load the weights back in during 'spacy train', you need to ensure all settings are the same between pretraining and training. Ideally, this is done by using the same config file for both commands. DOCS: https://spacy.io/api/cli#pretrain zLoading config from: F) overrides interpolateN pretrainingz/The [pretraining] block in your config is emptyr exitsT)parentszCreated output directory: z config.cfgz)Saved config file in the output directory)rrrsilentr zSuccessfully finished pretrain)rargsr verify_cli_argsrrinforrr#getfailrmkdirgoodto_diskr) rrrrrrrr config_overrides raw_configconfigs \/home/spazio/kingdom_calls/work5/transenv/lib/python3.12/site-packages/spacy/cli/pretrain.py pretrain_clir5sp.chh7 K[,G gHH$[M 23 { +  #3   # # %F ::m $ B!L    & -j\:;zL01HH 89 !HH -.1  s D77EcJ|rt|dk7r(|jstjd|d|jrM|j Dcgc]}|c}r/|rtj ddntj dd||j rtjd d tjd t|}|s|stjd d y|s"|d krtjd|dd yyyycc}w)N-zConfig file not foundr r%zOutput directory is not empty.z}If you're resuming a run in this directory, the old weights for the consecutive epochs will be overwritten with the new ones.zOutput directory is not empty. zuIt is better to use an empty directory or refer to a new output path, then the new directory will be created for you.zI--resume-path should be a weights file, but {resume_path} is a directory.Tz model\d+\.binz]You have to use the --epoch-resume setting when using a renamed weight file for --resume-pathrz=The argument --epoch-resume has to be greater or equal to 0. z is invalid) strrrr-iterdirwarnis_dirresearch)rrrrp model_names r4r*r*ns 3{+s2;;M;M;O (+Q?:+=+=+?@a@  HH0T  HH1B       HH[ YY/[1AB , HHo  q 0 HHOP\~]hi !1 As D )r<pathlibrtypingrtyperwasabirtraining.pretrainrutilr_utilr r r r rrrcommandContextintboolr5r*r4rMsK  ( *.$O 'uQU3%RS #  ] ! #&  J # #&  E # r:t2HI }d)E5R/ R/R/  R/~R/$R/&3-'R/23R/45R/ R/j"rL