Run2 extract pdf content
[COLM'26] SkillLearnBench is the first benchmark for evaluating continual learning methods that automatically generate agent skills.
npx -y skills add cxcscmu/SkillLearnBench --skill run2_extract_pdf_contentAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
What its author says it does
Copied from the file, not written here
Extracts text and mathematical descriptions from a PDF file to identify specific algorithm parameters and loss functions. This is used to ensure the implementation matches the theoretical definition in the paper.
SKILL.md
1.2 KB, as published. Nobody here has run it
- Install necessary extraction tools:
pip install pdfplumber. - Use a Python script to parse
/root/SimPO/paper.pdf. - Target sections involving "SimPO", "Loss Function", "Length Normalization", and "Reward Margin".
- Specifically look for:
- The formula for $\mathcal{L}_{\text{SimPO}}$.
- The definition of the reward margin ($\gamma$) and the scaling factor ($\beta$).
- How length normalization is applied to log probabilities (e.g., dividing by sequence length).
- Save the extracted text or key formulas to a temporary file or variable for reference during implementation.
import pdfplumber
def extract_simpo_details(pdf_path):
with pdfplumber.open(pdf_path) as pdf:
full_text = ""
for page in pdf.pages:
full_text += page.extract_text() + "\n"
return full_text
# Example usage to find specific parameters
# text = extract_simpo_details("/root/SimPO/paper.pdf")
# print(text)