diff --git a/scripts/coin/live1+_evaluate b/scripts/coin/live1+_evaluate
new file mode 100644
index 0000000..1cd1bee
--- /dev/null
+++ b/scripts/coin/live1+_evaluate
@@ -0,0 +1,12 @@
+deepspeed evaluate.py \
+    --live_version live1+ \
+    --eval_datasets coin_step_test coin_next_test coin_task_test coin_procedure_test coin_taskprocedure_test \
+    --per_device_train_batch_size 1 \
+    --per_device_eval_batch_size 1 \
+    --prediction_loss_only False \
+    --dataloader_num_workers 16 \
+    --bf16 True \
+    --tf32 True \
+    --report_to tensorboard \
+    --output_dir outputs/coin_benchmarks/live1+ \
+    --resume_from_checkpoint /path/your/lora_ckpt_folder