explosion spaCy Help-model-advice Discussions
Help: Model Advice Discussions
Understanding a model's predictions and improving accuracy
-
You must be logged in to vote Is it possible to use GPT2 model in spacy?
enhancementFeature requests and improvements -
You must be logged in to vote Can't find GPU for model training on Paperspace P4000
installInstallation issues gpuUsing spaCy on GPU -
You must be logged in to vote Wrong
feat / taggerpos_for "BATATA"Feature: Part-of-speech tagger perf / accuracyPerformance: accuracy -
You must be logged in to vote puzzled by results processing a project guternberg text
feat / nerFeature: Named Entity Recognizer -
You must be logged in to vote Training transformer model goes from score 0.97 to ZERO
bugBugs and behaviour differing from documentation feat / nerFeature: Named Entity Recognizer perf / memoryPerformance: memory use feat / trainingFeature: Training utils, Example, Corpus and converters feat / transformerFeature: Transformer -
You must be logged in to vote Review of legal documents in PDF - suggestion needed.
usageGeneral spaCy usage -
You must be logged in to vote Textcat model has extremely low score on predictions - too little sample (19 categories, 5400 rows).
feat / textcatFeature: Text Classifier perf / accuracyPerformance: accuracy -
You must be logged in to vote NER on whole paragraphs, instead of just keywords
feat / nerFeature: Named Entity Recognizer feat / spancatFeature: Span Categorizer -
You must be logged in to vote Tok2vec loss is growing in TextCategorizer
lang / frFrench language data and models trainingTraining and updating models feat / textcatFeature: Text Classifier feat / cliFeature: Command-line interface feat / tok2vecFeature: Token-to-vector layer and pretraining -
You must be logged in to vote Inconsistent model training across different computers
feat / nerFeature: Named Entity Recognizer reproducibilityConsistency, reproducibility, determinism, and randomness -
You must be logged in to vote Training a model for multiple tasks on a dataset with missing values
trainingTraining and updating models feat / nerFeature: Named Entity Recognizer feat / taggerFeature: Part-of-speech tagger feat / lemmatizerFeature: Rule-based and lookup lemmatization feat / transformerFeature: Transformer -
You must be logged in to vote Performance by named entity category of spacy/en_core_web_trf
docsDocumentation and website perf / accuracyPerformance: accuracy -
You must be logged in to vote Huge losses while Training Spacy Custom Model
feat / nerFeature: Named Entity Recognizer feat / trainingFeature: Training utils, Example, Corpus and converters -
You must be logged in to vote LOSS is skyrocketing when trying to train a NER classifier with transformers
feat / nerFeature: Named Entity Recognizer lang / ukUkrainian language data and models feat / transformerFeature: Transformer -
You must be logged in to vote Need help understanding results from text_categorizer
third-partyThird-party packages and services feat / textcatFeature: Text Classifier -
You must be logged in to vote Cant figure out why my model is not accurate
trainingTraining and updating models more-info-neededThis issue needs more information perf / accuracyPerformance: accuracy -
You must be logged in to vote Does it make sense to change the # hidden layers of a pretrained transformer?
feat / trainingFeature: Training utils, Example, Corpus and converters feat / transformerFeature: Transformer -
You must be logged in to vote Training NER model with a frozen transformer
feat / nerFeature: Named Entity Recognizer feat / trainingFeature: Training utils, Example, Corpus and converters feat / transformerFeature: Transformer -
You must be logged in to vote Model training using M1 metal gpu.
gpuUsing spaCy on GPU plat / m1Apple M1 architecture support -
You must be logged in to vote NER for partially annotated documents
feat / nerFeature: Named Entity Recognizer -
You must be logged in to vote Differences between Spacy3 and Spacy2 in NER
feat / nerFeature: Named Entity Recognizer v2spaCy v2.x -
You must be logged in to vote POS tagging inconsistent with length for french transformer model
feat / taggerFeature: Part-of-speech tagger perf / accuracyPerformance: accuracy -
You must be logged in to vote Extending training of an existing SpaCy model for NER
feat / nerFeature: Named Entity Recognizer -
You must be logged in to vote SpanCategorizer confidence scores are equal for all spans.
feat / pipelineFeature: Processing pipeline and components feat / tok2vecFeature: Token-to-vector layer and pretraining -
You must be logged in to vote Norwegian tagging is wrong (both large and small model)?
lang / nbNorwegian (Bokmål) language data and models
