Spaces:
Running
Running
Commit
·
a5ef5d4
1
Parent(s):
e543d33
Update quart_app.py
Browse files- quart_app.py +3 -3
quart_app.py
CHANGED
@@ -63,10 +63,10 @@ This is showcase how to make own server with Llama2 model.<br>
|
|
63 |
I'm using here 7b model just for example. Also here's only CPU power.<br>
|
64 |
But you can use GPU power as well!<br><br>
|
65 |
<h1>How to GPU?</h1>
|
66 |
-
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>
|
67 |
-
Powered by <a href="https://github.com/abetlen/llama-cpp-python">llama-cpp-python</a>, <a href="https://quart.palletsprojects.com/">Quart</a> and <a href="https://www.uvicorn.org/">Uvicorn</a>.<br><br>
|
68 |
<h1>How to test it on own machine?</h1>
|
69 |
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>
|
70 |
Or you can once follow steps in Dockerfile and try it on your machine, not in Docker.<br>
|
71 |
<br>''' + f"Memory used: {psutil.virtual_memory()[2]}<br>" + '''
|
72 |
-
<script>document.write("<b>URL of space:</b> "+window.location.href);</script>'''
|
|
|
|
63 |
I'm using here 7b model just for example. Also here's only CPU power.<br>
|
64 |
But you can use GPU power as well!<br><br>
|
65 |
<h1>How to GPU?</h1>
|
66 |
+
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>
|
|
|
67 |
<h1>How to test it on own machine?</h1>
|
68 |
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>
|
69 |
Or you can once follow steps in Dockerfile and try it on your machine, not in Docker.<br>
|
70 |
<br>''' + f"Memory used: {psutil.virtual_memory()[2]}<br>" + '''
|
71 |
+
<script>document.write("<b>URL of space:</b> "+window.location.href);</script>''' + '''
|
72 |
+
Powered by <a href="https://github.com/abetlen/llama-cpp-python">llama-cpp-python</a>, <a href="https://quart.palletsprojects.com/">Quart</a> and <a href="https://www.uvicorn.org/">Uvicorn</a>.<br><br>'''
|