Update README.md

README.md

@@ -5,7 +5,18 @@ base_model: [deepseek-ai/DeepSeek-Coder-V2-Instruct]
 #### Custom quantizations of deepseek-coder-v2-instruct optimized for cpu inference.
 
 ### Theis ones uses GGML TYPE IQ_4_XS in combination with q8_0 so it runs fast with minimal loss and takes advantage of int8 optimizations on most nevwer server cpus.
-### While it requirement custom code to make it is standard compatible with plain llama.cpp
+### While it requiremed custom code to make it is standard compatible with plain llama.cpp
+
+>[!TIP]
+>The following 4bit version is the one I use myself, it gets 17tps on 64 arm cores
+>You don't need to consolidates the files anymore, just point llama-cli to the first one and it'll handle the rest fine
+>Then to run just do
+>```c++
+>./llama-cli --temp 0.4 -m deepseek_coder_v2_cpu_iq4xm.gguf-00001-of-00004.gguf -c 32000 -co -cnv -i -f bigprompt.txt
+>```
+>
+
+
 
 ```verilog
 deepseek_coder_v2_cpu_iq4xm.gguf-00001-of-00004.gguf