Spaces:
Running
Running
File size: 3,237 Bytes
f9fd05d |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 |
from sentence_transformers import SentenceTransformer
from faiss import IndexFlatIP
import numpy as np
from huggingface_hub import InferenceClient
import json
import os
class RAGInterviewAgent:
def __init__(self, job_role, cv_summary):
self.job_role = job_role
self.cv_summary = cv_summary
self.current_q = 0
# Initialize models
self.embedder = SentenceTransformer('all-MiniLM-L6-v2')
self.llm = InferenceClient(model="meta-llama/Meta-Llama-3-8B-Instruct")
# Setup FAISS index
self.index = IndexFlatIP(384) # Embedding dimension
self._setup_knowledge_base()
# Generate questions
self.questions = self._generate_questions()
def _setup_knowledge_base(self):
"""Load job requirements into vector store"""
reqs = {
"Software Engineer": [
"Explain SOLID principles",
"How would you optimize a slow SQL query?",
"Describe your experience with unit testing"
],
"Data Scientist": [
"Explain bias-variance tradeoff",
"How would you handle missing data?",
"Describe a machine learning project you worked on"
]
}
self.knowledge = reqs[self.job_role]
embeddings = self.embedder.encode(self.knowledge)
self.index.add(embeddings)
def _generate_questions(self):
"""Generate questions using RAG"""
questions = []
# Get most relevant requirements
cv_embed = self.embedder.encode(self.cv_summary["text"])
_, indices = self.index.search(np.array([cv_embed]), 3)
for i in indices[0]:
prompt = f"""
Generate an interview question for a {self.job_role} position
based on this requirement: {self.knowledge[i]}
Candidate's CV summary: {self.cv_summary['text'][:1000]}
"""
response = self.llm.text_generation(
prompt,
max_new_tokens=100,
temperature=0.7
)
questions.append({
"text": response.strip(),
"base_requirement": self.knowledge[i]
})
return questions
def get_current_question(self):
return self.questions[self.current_q]
def next_question(self):
self.current_q += 1
def evaluate_answer(self, answer):
"""Evaluate answer against knowledge base"""
# Get relevant requirement
req = self.questions[self.current_q]["base_requirement"]
prompt = f"""
Evaluate this interview answer on a scale of 1-10:
Requirement: {req}
Answer: {answer}
Provide JSON output with: score, feedback
"""
response = self.llm.text_generation(
prompt,
max_new_tokens=200,
temperature=0.3
)
try:
return json.loads(response.split("{")[1].split("}")[0] + "}")
except:
return {"score": 5, "feedback": "Evaluation failed"} |