awslabs · AditModi · May 5, 2022 · May 5, 2022 · May 5, 2022 · May 5, 2022
diff --git a/docs/wiki/1_Motivation.md → docs/1_Motivation.md b/docs/wiki/1_Motivation.md → docs/1_Motivation.md
diff --git a/docs/wiki/1_getting_start/Build-on-Conda.md → docs/1_getting_start/Build-on-Conda.md b/docs/wiki/1_getting_start/Build-on-Conda.md → docs/1_getting_start/Build-on-Conda.md
diff --git a/.../1_getting_start/Build-on-Ubuntu-18.04.md → .../1_getting_start/Build-on-Ubuntu-18.04.md b/.../1_getting_start/Build-on-Ubuntu-18.04.md → .../1_getting_start/Build-on-Ubuntu-18.04.md
diff --git a/docs/wiki/1_getting_start/Build-on-macOS.md → docs/1_getting_start/Build-on-macOS.md b/docs/wiki/1_getting_start/Build-on-macOS.md → docs/1_getting_start/Build-on-macOS.md
diff --git a/docs/wiki/2_user_guide/AMP.md → docs/2_user_guide/AMP.md b/docs/wiki/2_user_guide/AMP.md → docs/2_user_guide/AMP.md
diff --git a/...wiki/2_user_guide/Distributed-Training.md → docs/2_user_guide/Distributed-Training.md b/...wiki/2_user_guide/Distributed-Training.md → docs/2_user_guide/Distributed-Training.md
diff --git a/docs/wiki/2_user_guide/Train-Model.md → docs/2_user_guide/Train-Model.md b/docs/wiki/2_user_guide/Train-Model.md → docs/2_user_guide/Train-Model.md
diff --git a/.../wiki/2_user_guide/Train-PyTorch-Model.md → docs/2_user_guide/Train-PyTorch-Model.md b/.../wiki/2_user_guide/Train-PyTorch-Model.md → docs/2_user_guide/Train-PyTorch-Model.md
diff --git a/docs/wiki/3_dev_guide/Add-Operator.md → docs/3_dev_guide/Add-Operator.md b/docs/wiki/3_dev_guide/Add-Operator.md → docs/3_dev_guide/Add-Operator.md
diff --git a/docs/wiki/3_dev_guide/Add-Pass.md → docs/3_dev_guide/Add-Pass.md b/docs/wiki/3_dev_guide/Add-Pass.md → docs/3_dev_guide/Add-Pass.md
diff --git a/docs/wiki/3_dev_guide/Analyze-Model.md → docs/3_dev_guide/Analyze-Model.md b/docs/wiki/3_dev_guide/Analyze-Model.md → docs/3_dev_guide/Analyze-Model.md
diff --git a/...iki/3_dev_guide/Distribution-Mechanism.md → docs/3_dev_guide/Distribution-Mechanism.md b/...iki/3_dev_guide/Distribution-Mechanism.md → docs/3_dev_guide/Distribution-Mechanism.md
@@ -241,4 +241,4 @@ It is recommended for communication operators to invoke `RequestDistributed()`.
 | ---  | ---         | ---              | ---     |
 | Void | No | Yes | Dummy for test purposes and standalone mode |
 | MPI | No | No | Help NCCL to initialize |
-| NCCL | Yes | Yes | Exchange data between GPUs efficiently |
+| NCCL | Yes | Yes | Exchange data between GPUs efficiently |
diff --git a/docs/wiki/3_dev_guide/IR-Format.md → docs/3_dev_guide/IR-Format.md b/docs/wiki/3_dev_guide/IR-Format.md → docs/3_dev_guide/IR-Format.md
diff --git a/docs/wiki/3_dev_guide/Memory-Pool.md → docs/3_dev_guide/Memory-Pool.md b/docs/wiki/3_dev_guide/Memory-Pool.md → docs/3_dev_guide/Memory-Pool.md
diff --git a/docs/wiki/3_dev_guide/Op-Dialect.md → docs/3_dev_guide/Op-Dialect.md b/docs/wiki/3_dev_guide/Op-Dialect.md → docs/3_dev_guide/Op-Dialect.md
diff --git a/docs/wiki/3_dev_guide/Profile-Model.md → docs/3_dev_guide/Profile-Model.md b/docs/wiki/3_dev_guide/Profile-Model.md → docs/3_dev_guide/Profile-Model.md
diff --git a/...wiki/3_dev_guide/pass/AutoDataParallel.md → docs/3_dev_guide/pass/AutoDataParallel.md b/...wiki/3_dev_guide/pass/AutoDataParallel.md → docs/3_dev_guide/pass/AutoDataParallel.md
diff --git a/..._guide/pass/Multi-CUDA-Stream-Schedule.md → ..._guide/pass/Multi-CUDA-Stream-Schedule.md b/..._guide/pass/Multi-CUDA-Stream-Schedule.md → ..._guide/pass/Multi-CUDA-Stream-Schedule.md
diff --git a/...iki/3_dev_guide/pass/Rematerialization.md → docs/3_dev_guide/pass/Rematerialization.md b/...iki/3_dev_guide/pass/Rematerialization.md → docs/3_dev_guide/pass/Rematerialization.md
@@ -95,4 +95,4 @@ foreach operator in RAF IR:
             memory consumption -= sizeof(T) 
 ```
 
-The score of each tensor is computed according to its size and the total latency required to recompute it. The latency of each operator is obtained through profiling by default. As mentioned earlier, you can set `"raf.remat.use_gflops_cost" = False` to use the faster but less-accurate GFLOPS-baed cost instead. 
+The score of each tensor is computed according to its size and the total latency required to recompute it. The latency of each operator is obtained through profiling by default. As mentioned earlier, you can set `"raf.remat.use_gflops_cost" = False` to use the faster but less-accurate GFLOPS-baed cost instead. 
diff --git a/...rib_guide/Development-And-Pull-Request.md → ...rib_guide/Development-And-Pull-Request.md b/...rib_guide/Development-And-Pull-Request.md → ...rib_guide/Development-And-Pull-Request.md
diff --git a/docs/wiki/4_contrib_guide/RFC-Collections.md → docs/4_contrib_guide/RFC-Collections.md b/docs/wiki/4_contrib_guide/RFC-Collections.md → docs/4_contrib_guide/RFC-Collections.md
diff --git a/docs/wiki/gen_readme.py → docs/gen_readme.py b/docs/wiki/gen_readme.py → docs/gen_readme.py
diff --git a/docs/wiki/README.md → docs/index.md b/docs/wiki/README.md → docs/index.md