Shiori is a Deep Research Agent for Machine Learning research papers and blog posts.
# Clone the repository
git clone https://github.com/yourusername/Shiori.git
cd Shiori
# Install dependencies
pip install -r requirements.txt
# Install the package in development mode
pip install -e .After installation, you can use the shiori command directly:
# Fetch research papers (default: from the last week)
shiori research
# Fetch blog posts (default: from the last week)
shiori blog
# Specify a time period (d = 1 day, w = 1 week, m = 1 month)
shiori research --time d
shiori research --time w
shiori research --time m
shiori blog --time d
shiori blog --time w
shiori blog --time m
# Specify a research field (nlp or bioinformatics)
shiori research --field nlp
shiori research --field bioinformatics
shiori blog --field nlp
shiori blog --field bioinformatics
# Specify an output file
shiori research --output custom_papers.md
shiori blog --output custom_blogs.md
# Combine options
shiori research --time w --field nlp --output nlp_papers_weekly.mdShiori currently supports two research fields:
- Large Language Models (LLMs)
- Training/fine-tuning methods
- Efficiency & deployment
- And other NLP-related topics
- Long-read sequencing technology
- Pangenomics and pan-genome graph studies
- Pan-genome graph construction and variant calling pipelines
- Pan-genome graph in clinical research
- And other bioinformatics/computational biology topics
Shiori includes telemetry capabilities to help monitor the agent's performance and behavior. To run the telemetry server:
python -m phoenix.server.main serveThis will start the telemetry server at http://localhost:6006, which helps monitor the agent's performance and behavior.
The output is a markdown file containing:
- For research papers (default:
papers_summary.md): Research papers with title, problem, approach, and evaluation - For blog posts (default:
blog_summary.md): Blog posts from major AI companies with key points