lm-sim 0.0.4

Python package to compute similarity between LMs

LM-Similarity

lm-sim is a Python module for computing similarity between Language Models and is distributed under the MIT license.

Installation

Dependencies

lm-sim requries:

• Python (>=3.9)
• Numpy (>= 1.19.5)

User installation

If you already have a working installation of NumPy, the easiest way to install lm-sim is using pip:

pip install lm-sim

Example Usage

Currently we support the calcualtion of 3 similarity metrics in the context of MCQ datasets:

• Goels $k$ (discrete)
• Goels $k_p$ (probabilistic)
• Error Consistency

Goels $k$ and Goels $k_p$

Below is a simple example on how to compute Goels $k_p$. The input has be to formatted as follows:

• output_a: list[np.array], containing the softmax output probabilties of model a
• output_b: list[np.array], containing the softmax output probabilties of model b
• gt: list[int], containing the index of the ground truth

from lmsim.metrics import Goels_k

goels_k = Goels_k()
goels_k.compute_k(output_a, output_b, gt)

For a discrete computation (when output probabilities are not availble) set the flag prob=False and the input must be formatted as one-hot vectors:

• output_a: list[np.array], one-hot vector of model a
• output_b: list[np.array], one-hot vector of model b

from lmsim.metrics import Goels_k

goels_k = Goels_k(prob=False)
goels_k.compute_k(output_a, output_b, gt)

Error Consistency

from lmsim.metrics import EC

ec = EC(prob=False)
ec.compute_k(output_a, output_b, gt)

Implementation supports both softmax output probabilties or one-hot vector as input.