Jupyter Lab Servers¶
AlphAI integrates directly with American Data Science's remote Jupyter Lab servers.
You can start servers, stop servers, upload files to your remote file system, load GPU profiling data, and even run code programmatically!
Using these tools require authentication.
American Data Science Client¶
Let's instantiate the AlphAI object to profile our tensor run with PyTorch.
import os
from dotenv import load_dotenv
# Make a `.env` file that contains the following line
# ALPHAI_API_KEY=<your-api-key>
load_dotenv()
import os
from alphai import AlphAI
aai = AlphAI(
# Don't need this line if you ran load_dotenv()
api_key=os.environ.get("ALPHAI_API_KEY"),
)
Start your remote Jupyter Lab servers¶
Starting up your servers with AlphAI will use the configurations and environment runtime that were used the last time you started the server from your Dashboard.
If it's a new server or you've never started it, it will use the default AI environment runtime.
Command Agent Alph¶
response = aai.alph(server_name="default", messages="I need a python file that analyzes a generic csv file.", engine="gpt3")
Run code remotely¶
Alphai allows you to run and "deploy" your code given a string or file path. Your server will automatically start a kernel and run your code remotely.
All servers will also run a tunnel on port 5000, so you could even check out your running servers and apps hosted directly from the server!
hf_code = """from transformers import pipeline
import gradio as gr
gr.close_all()
pipe = pipeline("translation", model="Helsinki-NLP/opus-mt-en-de")
demo = gr.Interface.from_pipeline(pipe)
demo.launch(server_port=5000, inline=False)"""
aai.run_code(hf_code, clear_other_kernels=True)