Qwen-2.5-3B-Resume-Evaluator

An objective, rubric-based AI Resume Screener and Evaluator adapter fine-tuned on Qwen/Qwen2.5-3B-Instruct using QLoRA.

Key Capabilities

• Calibrated Scoring (0–100): Comprehensive fit rating designed to prevent score inflation.
• Categorical Breakdown: Technical Skills, Experience Relevance, Education & Certifications, and Leadership & Impact.
• Red Flag & Discrepancy Detection: Penalizes buzzword stuffing without metrics, title inflation, and chronological conflicts.
• Strict JSON Output: 100% strict JSON schema conformance without markdown drift.

Quickstart Inference (8-bit INT8 ~3.25 GB VRAM)

GitHub: "https://github.com/ChiragPrasad2006/Resume-Evaluator-3B"

import json, torch
from transformers import AutoModelForCausalLM, AutoTokenizer, BitsAndBytesConfig
from peft import PeftModel

bnb_config = BitsAndBytesConfig(load_in_8bit=True)
model = AutoModelForCausalLM.from_pretrained(
    "Qwen/Qwen2.5-3B-Instruct",
    quantization_config=bnb_config,
    device_map="cuda:0",
    torch_dtype=torch.float16,
)
model = PeftModel.from_pretrained(model, "Kami0867/Qwen-2.5-3B-Resume-Evaluator")
model.eval()