src/about.py

from dataclasses import dataclass
from enum import Enum

@dataclass
class Task:
    benchmark: str
    metric: str
    col_name: str
    reference_url: str


# Select your tasks here
# ---------------------------------------------------
class Tasks(Enum):
    # task_key in the json file, metric_key in the json file, name to display in the leaderboard 
    task0 = Task("aiera_transcript_sentiment", "accuracy,none","Sentiment", reference_url="https://huggingface.co/datasets/Aiera/aiera-transcript-sentiment")
    task1 = Task("aiera_ect_sum", "bert_f1,none","Summary", reference_url="https://huggingface.co/datasets/Aiera/aiera-ect-sum")
    task2 = Task("finqa", "exact_match_manual,none","Q&A", reference_url="https://huggingface.co/datasets/Aiera/finqa-verified")
    task3 = Task("aiera_speaker_assign", "accuracy,none", "Speaker ID", reference_url="https://huggingface.co/datasets/Aiera/aiera-speaker-assign")


NUM_FEWSHOT = 0 # Change with your few shot
# ---------------------------------------------------


LEADERBOARD_TITLE_PNG = "assets/aiera-leaderboard-transparent.png"
# Your leaderboard name
TITLE = """<h1 align="center" id="space-title">Aiera Leaderboard</h1>"""

# What does your leaderboard evaluate?
INTRODUCTION_TEXT = """
The Aiera Leaderboard evaluates the performance of LLMs on a number of financial intelligence tasks including:
* Assignments of speakers for event transcript segments and identification of speaker changes. 
* Abstractive summarizations of earnings call transcripts.
* Calculation-based Q&A over financial text. 
* Financial sentiment tagging for transcript segments.

A guide for eval tasks is avaliable on github at [https://github.com/aiera-inc/aiera-benchmark-tasks](https://github.com/aiera-inc/aiera-benchmark-tasks).
"""

# Which evaluations are you running? how can people reproduce what you have?
LLM_BENCHMARKS_TEXT = f"""
## How it works
Models are evaluated on the following tasks
* **aiera_speaker_assign**: Assignments of speakers for event transcript segments and identification of speaker changes. Dataset available on [huggingface](https://huggingface.co/datasets/Aiera/aiera-speaker-assign).
* **aiera-ect-sum**: Abstractive summarizations of earnings call transcripts. Dataset available on [huggingface](https://huggingface.co/datasets/Aiera/aiera-ect-sum).
* **finqa**: Calculation-based Q&A over financial text. Dataset available on [huggingface](https://huggingface.co/datasets/Aiera/finqa-verified).
* **aiera-transcript-sentiment**: Event transcript segments with labels indicating the financial sentiment. Dataset available on [huggingface](https://huggingface.co/datasets/Aiera/aiera-transcript-sentiment).

## Reproducibility
A guide for running the above tasks using EleutherAi's lm-evaluation-harness is avaliable on github at [https://github.com/aiera-inc/aiera-benchmark-tasks](https://github.com/aiera-inc/aiera-benchmark-tasks).

"""

EVALUATION_QUEUE_TEXT = """
Note: The evaluation suite is only able to run on models available via Hugging Face's Serverless Inference API. Unfortunately, that means the models available for execution are limited, but we are working to support more models in the future.

## In case of model failure
If your model is displayed in the `FAILED` category, its execution stopped.
Check you can launch the EleutherAIHarness on your model locally using the guide avaliable on github at [https://github.com/aiera-inc/aiera-benchmark-tasks](https://github.com/aiera-inc/aiera-benchmark-tasks).
Models must be able to accomodate large context windows in order to run this evaluation.

"""

CITATION_BUTTON_LABEL = "Copy the following snippet to cite these results"
CITATION_BUTTON_TEXT = r"""@misc{aiera-finance-leaderboard,
  author = {Jacqueline Garrahan, Bryan Healey},
  title = {Aiera Finance Leaderboard},
  year = {2024},
  publisher = {Aiera},
  howpublished = "\url{https://huggingface.co/spaces/Aiera/aiera-finance-leaderboard}"
}
@software{eval-harness,
  author       = {Gao, Leo and
                  Tow, Jonathan and
                  Biderman, Stella and
                  Black, Sid and
                  DiPofi, Anthony and
                  Foster, Charles and
                  Golding, Laurence and
                  Hsu, Jeffrey and
                  McDonell, Kyle and
                  Muennighoff, Niklas and
                  Phang, Jason and
                  Reynolds, Laria and
                  Tang, Eric and
                  Thite, Anish and
                  Wang, Ben and
                  Wang, Kevin and
                  Zou, Andy},
  title        = {A framework for few-shot language model evaluation},
  month        = sep,
  year         = 2021,
  publisher    = {Zenodo},
  version      = {v0.0.1},
  doi          = {10.5281/zenodo.5371628},
  url          = {https://doi.org/10.5281/zenodo.5371628}
}
"""