File size: 3,975 Bytes
b7be4fe | 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 | from fastapi import FastAPI, HTTPException
from pydantic import BaseModel
from transformers import AutoModelForCausalLM, AutoTokenizer
from TextToSpeech import text_to_speech
import torch
import base64
# =========================================
# ENUM MAPPINGS (Match Backend Enums)
# =========================================
SESSION_TYPES = {
1: "technical",
2: "softskills"
}
TRACKS = {
19: "generalprogramming"
}
# =========================================
# LOAD MODEL ONCE (Global)
# =========================================
MODEL_PATH = "Fayza38/Question_and_Answer"
tokenizer = AutoTokenizer.from_pretrained(MODEL_PATH)
model = AutoModelForCausalLM.from_pretrained(
MODEL_PATH,
torch_dtype=torch.float32,
device_map="cpu"
)
app = FastAPI()
# =========================================
# REQUEST MODEL
# =========================================
class QuestionRequest(BaseModel):
sessionType: int
difficultyLevel: int | None = None
trackName: int
# =========================================
# HELPER: GENERATE TEXT USING QWEN TEMPLATE
# =========================================
def generate_from_model(prompt: str):
messages = [
{"role": "system", "content": "You are a professional interview question generator."},
{"role": "user", "content": prompt}
]
formatted_prompt = tokenizer.apply_chat_template(
messages,
tokenize=False,
add_generation_prompt=True
)
inputs = tokenizer(formatted_prompt, return_tensors="pt")
with torch.no_grad():
outputs = model.generate(
**inputs,
max_new_tokens=1200,
temperature=0.7
)
decoded = tokenizer.decode(outputs[0], skip_special_tokens=True)
return decoded
# =========================================
# PARSE Q/A FORMAT
# =========================================
def parse_qa_blocks(text: str):
blocks = text.split("\n\n")
results = []
for block in blocks:
if "Q:" in block and "A:" in block:
parts = block.split("A:")
question = parts[0].replace("Q:", "").strip()
answer = parts[1].strip()
results.append((question, answer))
return results
# =========================================
# MAIN ENDPOINT
# =========================================
@app.post("/generate-questions")
def generate_questions(request: QuestionRequest):
if request.sessionType not in SESSION_TYPES:
raise HTTPException(status_code=400, detail="Invalid session type")
session_type = SESSION_TYPES[request.sessionType]
# ---------------- SOFT SKILLS ----------------
if session_type == "softskills":
prompt = """
Generate 10 behavioral interview questions.
Format exactly as:
Q: ...
A: ...
"""
# ---------------- TECHNICAL ----------------
elif session_type == "technical":
if request.trackName not in TRACKS:
raise HTTPException(status_code=400, detail="Track not supported")
difficulty = request.difficultyLevel or 1
prompt = f"""
Generate 10 General Programming interview questions.
Difficulty level: {difficulty}
Format exactly as:
Q: ...
A: ...
"""
else:
raise HTTPException(status_code=400, detail="Invalid session type")
# -------- Generate once --------
raw_output = generate_from_model(prompt)
qa_pairs = parse_qa_blocks(raw_output)
if len(qa_pairs) == 0:
raise HTTPException(status_code=500, detail="Model failed to generate valid Q/A format")
response = []
for idx, (question, answer) in enumerate(qa_pairs[:10], 1):
response.append({
"questionText": question,
"questionId": idx,
"questionIdealAnswer": answer
})
return response |