finbert-causal-detection-model

FinBERT Causal Detection Model

This repository stores the model artifacts for the FinBERT-based causal detection model.

Model

• Owner: dharmendra-nkr
• Repo: dharmendra-nkr/finbert-causal-detection-model

What’s in this repo

• artifacts/ — model files and helper artifacts. Large raw datasets were removed from the model repo to keep the model repo lightweight.

Removed large files

• artifacts/final_financial_causality_dataset.local.csv (removed from repo)
• artifacts/label_shifted_fin_causality_dataset.csv (removed from repo)

These datasets are kept locally in the project's artifacts/ folder. If you want them hosted publicly, consider uploading them to Google Drive, Zenodo, or an S3 bucket and updating these instructions with the download links.

How to download files from this model repo (example) Python (recommended):

from huggingface_hub import hf_hub_download, snapshot_download

# Download a single file
file_path = hf_hub_download(repo_id="dharmendra-nkr/finbert-causal-detection-model", filename="artifacts/multimodal_model_reconstructed.pkl")

# Or download the full repo snapshot locally
local_dir = snapshot_download(repo_id="dharmendra-nkr/finbert-causal-detection-model")

Shell:

• Download a single file via curl (raw link shown on Hugging Face file page) or use git lfs if large files are tracked via LFS.

Usage

• The repository contains serialized model artifacts (.pkl, .pkl.zip) — load them with your project's model-loading code. These files may be custom pickles; inspect their format and load safely in a controlled environment.

Notes & Recommendations

• Large datasets are not hosted here to avoid size limits. I can add links to external storage if you provide them.
• For future large-model pushes, consider enabling Git LFS for the model repo or using the Hub's large-file upload support.

Contact / Next steps If you want, I can:

• Add download links for the CSV datasets if you provide a hosting location.
• Add a model card (README.md) with more metadata, tags, license, and example code to load and use the model.