Skip to content

Regarding reproducing the results #1

@SachinVashisth

Description

@SachinVashisth

Hi

In the ReadMe file, you have provided this command: python QA.py TASK DATASET_NAME NLI_MODEL DATASET_VERSION

But I can't see any file QA.py in the repo. can you please provide that file?

Instead, I ran the following command: python main.py "EntailmentBank" "full_validity" "LLaMa3" "neg"

But I got errors in every instance of the EntailmentBank-Neg version.
Eg:- The first instance in the file results/reasoning/EB_neg_NEW-LLaMa3-05_26.jsonl (provided by the authors) doesn't show any error but I got the following error (when I ran the aforementioned main.py command using LLama3) for the same instance:

"logic_premises": [[], []], "logic_conclusion": ["-1", "-1"], "correspondance": [{}, {}], "errors": ["Logic Transform Error - Step n°: 0 name 'find_biggest_NP_parent' is not defined", "Logic Transform Error - Step n°: 1 name 'find_biggest_NP_parent' is not defined", ["Entailments Error - Step n°: 0 'str' object has no attribute 'get_variables_implications'", "Entailments Error - Step n°: 1 'str' object has no attribute 'get_variables_implications'"]], "entailments_dict": [{}, {}], "predicted_steps": [false, false]

Also, can you please tell how to calculate precision, recall, F0.5, and Somer's D using the json file in the directory results/reasoning/?

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type
    No fields configured for issues without a type.

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions