config.py

import os
from string import Template


HF_TOKEN = os.getenv('HF_TOKEN')

# model configs
LLM_ID = "google/gemma-3-4b-it" if os.getenv("ENV") == "PROD" else "google/gemma-3-1b-it"
EMBEDDING_MODEL_ID = "google/embeddinggemma-300M"

# embedding parameters
DOMAIN_SELECTION_THRESHOLD = 0.5
DOCUMENT_SELECTION_THRESHOLD = 0.2
DOCUMENT_FOLLOWUP_THRESHOLD = 0.5

# generation parameters
RESPONSE_MAX_NEW_TOKENS = 512
RESPONSE_SAMPLING_TOP_P = 0.95
RESPONSE_SAMPLING_TEMPERATURE = 0.7

# cv document data files
CV_DATA_GCS_BUCKET = "wlee-jsons"
CV_DATA_DIRNAME = "cvdata/"
CV_DATA_FILE_EDUCATION = "education.json"
CV_DATA_FILE_JOBS = "jobs.json"
CV_DATA_FILE_PAPERS = "papers.json"
CV_DATA_FILE_SKILLS = "skills.json"
RESET_QUERY = "restart chat"

# text configs
PERSON_NAME = "William"
PERSON_PRONOUN = "he"
PERSON_PRONOUN_POSSESSIVE = "his"
WELCOME_MESSAGE_TEMPLATE = Template(f"Hi! I'm {PERSON_NAME}'s Career Assistant. I can answer questions you might have regarding {PERSON_PRONOUN_POSSESSIVE} resume. "
                   "To start, tell me which one of the following domains you would like to focus on:\n"
                   "$domains\n"
                   f"At any point in chat, say \'{RESET_QUERY}\' to reset chat and return to current state.")
NO_MATCH_RESPONSE = "I don't have that specific information right now."
EDUCATION_DESCRIPTION = "Academic degrees and relevant coursework"
JOBS_DESCRIPTION = "Work experience"
PAPERS_DESCRIPTION = "Papers and publications"
SKILLS_DESCRIPTION = "Technical skills, such as programming language, framework, and tools"
JOB_TYPES = [
    "Software Engineer",
    "Research Intern",
]

# cache files
KNOWLEDGE_CACHE_FILE = "knowledge.pkl"