Update README.md

README.md

@@ -104,7 +104,7 @@ FLM-101B is trained on a cluster of 24 DGX-A800 GPU (8×80G) servers for less th
 
 #### Software
 
-FLM-101B was trained with Megatron-FLM(open source soon...) which is modified based on Megatron-LM.
+FLM-101B was trained with a codebase adapted from Megatron-LM.
 It uses a 3D(DP+TP+PP) parallelism approach and distributed optimizer.