Spaces:

neurotech

/

Swahili-sentiment-analysis

like 4

Runtime error

Hi @mutiga Thanks for giving feedback, Let me look more into this , getting back to you shortly

Thanks again
Kalebu

So just to provide more information on the model, It is a Bayesian Model which means it only supports binary classification(Negative|Positive) trained on https://github.com/Neurotech-HQ/swahili-sentiment-analysis-dataset, I guess the size of the data was too small for it to well generalize the two classes to real-world sentiment and also we are also working on realising one fine-tuned Roberta-Swahili to really exploit pre-learned Swahili words embeddings, Hope this will perform well

Thank you
Kalebu

Thank you Kalebu,

I have to admit that this is a job well done. Thank you for adding Swahili to the platform. May I ask, is your model private? I tried to port it onto Jupyter with the code below but it threw a key error.

swahili_classifier = pipeline('sentiment analysis', model = "neurotech/Swahili-sentiment-analysis")

KeyError: "Unknown task sentiment analysis, available tasks are ['audio-classification', 'automatic-speech-recognition', 'conversational', 'feature-extraction', 'fill-mask', 'image-classification', 'image-segmentation', 'ner', 'object-detection', 'question-answering', 'sentiment-analysis', 'summarization', 'table-question-answering', 'text-classification', 'text-generation', 'text2text-generation', 'token-classification', 'translation', 'visual-question-answering', 'vqa', 'zero-shot-classification', 'zero-shot-image-classification', 'translation_XX_to_YY']"