Usage and Output

As of now LAiSER can be used a command line tool or from the Jupyter notebook(Google Colab). The steps to setup the tool are as follows:

Google Colab Setup (preferred)

LAiSER's Jupyter notebook is, currently, the fastest way to get started with the tool. You can access the notebook here

• Once the notebook is imported in google colaboratory, connect to a GPU-accelerated runtime(T4 GPU) and run the cells in the notebook.

HuggingFace Setup

• Follow this article to create an account in HuggingFace and activate access tokens to access the models.

Create Colab Secret Keys

• Click the keys (can be found in the below image) button.

• Fill HF_TOKEN in the Name field and your huggingface access token in the value field.

LAiSER package Installation

• Install the laiser package using pip:

  pip install uv
  uv pip install laiser -q

Importing Skill Extractor

• Import Skill Extractor from laiser.

  from laiser.skill_extractor import Skill_Extractor

NOTE

Import pandas and torch libraries to handle data and GPU-accelerated computations.

• Install additional libraries:

   import pandas as pd
   import torch

Initializing Extractor

• To initialize the skill extractor follow these steps:

 se = Skill_Extractor(AI_MODEL_ID="marcsun13/gemma-2-9b-it-GPTQ", HF_TOKEN="<YOUR_HUGGING_FACE_API_TOKEN>", use_gpu=True)

Output

Output Column Descriptions:

• Research ID - Unique ID of each job description.
• Raw Skill - Skill Extracted by the model.
• Level - Skill level mapped basis the description between 1 to 12.
• Knowledge Required - Knowledge required to gain the extarcted skill.
• Task Abilities - Tasks that ca be performed using the extracted skill.
• Skill Tag - Unique ID of taxonomy skill that matches the Raw Skill.
• Correlation Coefficient - Describes the closeness of Raw Skill and Skill Tag.