---
language:
- en
- ru

license: mit

tags:
- Cometrain AutoCode
- Cometrain AlphaML
- moex

datasets:
- financial-sentiment-analysis
- moscow-exchange-market
- financial_phrasebank

widget:
- text: "April 14 (Reuters) - Rio Tinto (RIO.AX), one of the largest Australian mining companies, on Thursday confirmed its exit from the state mining lobby group after raising concerns that its policy on expansion of coal mines did not align with the Paris Climate Agreement."
  example_title: "Rio Tinto Decision (Neutral)"
- text: "LONDON, April 13 (Reuters) - Crypto lender Nexo said it has teamed up with global payments company Mastercard (MA.N) to launch on Wednesday what it calls the world's first crypto-backed payment card."
  example_title: "New Mastercard & Nexo project (Positive)"
- text: "April 14 (Reuters) - The Russian rouble weakened on Thursday, driven by expectations that Russia may relax its temporary capital control measures further, while stocks fell as the country continued what it calls 'a special military operation' in Ukraine."
  example_title: "Crisis in Russia (Negative)"

inference:
 parameters:
   top_p: 0.9
   temperature: 0.5
---
# moexT5
The stocks-news-t5 model was further trained based on Moscow Exchange data obtained with AlgoPack(https://www.moex.com/ru/algopack)

## stocks-news-t5
This model has been automatically fine-tuned and tested as part of the development of the GPT-2-based AutoML framework for accelerated and easy development of NLP enterprise solutions. Fine-tuned [T5](https://huggingface.co/t5-base) allows to analyze financial market news.
Automatically trained on [Financial Sentiment Analysis(2022)](https://www.kaggle.com/datasets/sbhatti/financial-sentiment-analysis) dataset.
## Made with Cometrain AlphaML & AutoCode
This model was automatically fine-tuned using the Cometrain AlphaML framework and tested with CI/CD pipeline made by Cometrain AutoCode
## Cometrain AlphaML command
```shell
$ cometrain create --name stocks-news --model auto --task 'Machine learning model for finance news analysis' --output transformers
```