This is a fine-tuned version of Qwen/Qwen2.5-7B-Instruct trained on high-rigor CAT (Common Admission Test) mock examination datasets, including Quantitative Aptitude, DILR matrix sets, and VARC passages.

Model Highlights

Base Model: Qwen/Qwen2.5-7B-Instruct
Fine-Tuning Method: QLoRA (Rank 16, Alpha 32)
Training Loss: 1.0868
Target Rigor: SimCAT / AIMCAT Level
Outputs: Strict JSON question structures with step-by-step mathematical derivations and distractor options.

Usage with Transformers

python

from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "GaneshKhovalan/Qwen2.5-7B-CAT-Instruct-v1"
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(model_id, device_map="auto")

Developed for CAT Infinite AI Platform.