Update README

Mention how to compile the version of llama.cpp that works until the PR is merged upstream.
Also give command for verifying the md5sum

README.md

@@ -9,7 +9,20 @@ license: cc-by-nc-4.0
 <!-- description start -->
 ## Description
 
-This repo contains llama.cpp GGUF format model files for [Command-R 35B v1.0](https://huggingface.co/CohereForAI/c4ai-command-r-v01).
+This repo contains llama.cpp GGUF format model files for
+[Command-R 35B v1.0](https://huggingface.co/CohereForAI/c4ai-command-r-v01).
+
+Note: you need to clone llama.cpp and compile until the
+[PR6033](https://github.com/ggerganov/llama.cpp/pull/6033) is merged upstream:
+```
+git clone https://github.com/acanis/llama.cpp.git
+cd llama.cpp
+mkdir build
+cd build
+cmake .. -DLLAMA_CUBLAS=ON
+cmake --build . --config Release -- -j16
+cd ..
+```
 
 ## F16 files are split and require joining
 
@@ -36,3 +49,8 @@ Then you can remove the split files to save space:
 del c4ai-command-r-v01-f16.gguf-split-a c4ai-command-r-v01-f16.gguf-split-b
 ```
 
+You can optionally confirm the checksum of merged c4ai-command-r-v01-f16.gguf
+with the md5sum file:
+```
+md5sum -c md5sum
+```