README.md

All experiments were conducted using two NVIDIA V100 GPUs or via DeepSeek-V3 API calls .

The following Python libraries and packages were employed in our implementation: torch==2.1.0 transformers==4.34.1 datasets==2.14.6 accelerate==0.33.0 trl==0.9.6 peft==0.12.0 openai==1.38.0

For running:

python Inference_in_context_learning.py --dataset TempReason_l3 --model deepseek-v3 --ICL --CoT

For Evaluation:

python Evaluation.py --dataset TempReason_l3 --model deepseek-v3 --ICL --CoT --type AdapTime_