update

app.py

@@ -11,18 +11,14 @@ llama_model_name = "TheBloke/dolphin-2.2.1-AshhLimaRP-Mistral-7B-GGUF"
 print("! INITING DONE !")
 
 # Preparing things to work
-title = "llama.cpp API"
-desc = '''<h1>Hello, world!</h1>
-This is showcase how to make own server with Llama2 model.<br>
-I'm using here 7b model just for example. Also here's only CPU power.<br>
-But you can use GPU power as well!<br><br>
-<h1>How to GPU?</h1>
-Change <code>`CMAKE_ARGS="-DLLAMA_BLAS=ON -DLLAMA_BLAS_VENDOR=OpenBLAS`</code> in Dockerfile on <code>`CMAKE_ARGS="-DLLAMA_CUBLAS=on"`</code>. Also you can try <code>`DLLAMA_CLBLAST`</code> or <code>`DLLAMA_METAL`</code>.<br><br>
-<h1>How to test it on own machine?</h1>
-You can install Docker, build image and run it. I made <code>`run-docker.sh`</code> for ya. To stop container run <code>`docker ps`</code>, find name of container and run <code>`docker stop _dockerContainerName_`</code><br>
-Or you can once follow steps in Dockerfile and try it on your machine, not in Docker.<br>
-<br>''' + f"Memory used: {psutil.virtual_memory()[2]}<br>" + '''
-Powered by <a href="https://github.com/abetlen/llama-cpp-python">llama-cpp-python</a> and <a href="https://www.gradio.app/">Gradio</a>.<br><br>'''
+title = "Demo for 7B Models - Quantized"
+desc = '''<h1>Demo for 7B Models - Quantized</h1>
+Quantized to run in the free tier hosting. 
+Have a quick way to test models or share them with others without hassle.
+It runs slow, as it's on cpu. Usable for basic tests.
+It uses quantized models in gguf-Format and llama.cpp to run them.
+<br>''' + f"DEBUG: Memory used: {psutil.virtual_memory()[2]}<br>" + '''
+Powered by ...'''
 
 # Loading prompt
 prompt = ""

requirements.txt

@@ -1,3 +1,3 @@
 psutil
 gradio
-llama_cpp
+llama-cpp-python