Update README.md

README.md

@@ -1,2 +1,10 @@
 
-A Moe model built on top of Qwen1.5-7B-Chat, Qwen1.5-7B and Crystalcareai/CrystalQwen-1.5-7B, then qlora on WizardLM_evol_instruct_70k via mlx
+A Moe model built on top of Qwen1.5-7B-Chat, Qwen1.5-7B and Crystalcareai/CrystalQwen-1.5-7B.
+
+```
+pip install mlx-moe
+
+python -m mlx_moe.generate --model mzbac/qwen-1.5-2x3-hf-4bit-mlx --prompt "how back prop works"
+
+```
+