metadata

license: apache-2.0
datasets:
  - winddude/finacial_pharsebank_66agree_split
  - financial_phrasebank
language:
  - en
metrics:
  - accuracy
model-index:
  - name: financial-sentiment-analysis
    results:
      - task:
          name: Text Classification
          type: text-classification
        dataset:
          name: financial_phrasebank
          type: financial_phrasebank
          args: sentences_66agree
        metrics:
          - name: Accuracy
            type: accuracy
            value: 0.84
pipeline_tag: text-classification
tags:
  - finance
  - sentiment

Mamba Finacial Headline Sentiment

Score 0.84 on accuracy for the finacial phrasebank dataset. A completely huggingface capitable implementation of sequence classification with mamba using: https://github.com/getorca/mamba_for_sequence_classification.

Inference:

from transformers import pipeline


model_path = 'winddude/mamba_finacial_phrasebank_sentiment'

classifier = pipeline("text-classification", model=model_path, trust_remote_code=True)

text = "Finnish retail software developer Aldata Solution Oyj reported a net loss of 11.7 mln euro $ 17.2 mln for 2007 versus a net profit of 2.5 mln euro $ 3.7 mln for 2006 ."

classifier(text)

gives: [{'label': 'NEGATIVE', 'score': 0.8793253302574158}]