Add eval config for QWEN2_5 model using 0.5B variant (#2230)

Add eval config for QWEN2_5 model using 0.5B variant (#2230) #6558