Accelerate ML workflows with zero code rewrites on Google Cloud

Thanks for reading, @Ariana_Rose! I just replaced the “Benchmarks” link with an updated documentation page that works.

Sample Code presented as follows %load_ext cudf.pandas

%load_ext cuml.accel import sys

from pathlib import Path

sys.path.insert(0, str(Path.cwd().parent / ‘src’))

import pandas as pd

from sklearn.ensemble import RandomForestRegressor

from sklearn.metrics import mean_squared_error, r2_score

from sklearn.model_selection import train_test_split

from train import FEATURES, build_features, make_synthetic

Data Preparation

raw = make_synthetic(1_000_000, seed=42)

data = build_features(raw, target=‘fare_amount’)

X_train, X_test, y_train, y_test = train_test_split(

data\[FEATURES\], data\['fare_amount'\], test_size=0.2, random_state=42

)

data.shape model = RandomForestRegressor(n_estimators=200, max_depth=18, n_jobs=-1, random_state=42)

model.fit(X_train, y_train)

pred = model.predict(X_test)

{‘rmse’: mean_squared_error(y_test, pred) ** 0.5, ‘r2’: r2_score(y_test, pred)}

%%cudf.pandas.profile

profiled = build_features(raw, target=‘fare_amount’)

@this code snippet helped me a lot

Nice Sample

Steep learning curve, need a few read to understand. but Thank you.

Nice work @jeffnelson . Thanks for the information

i am new in this field please i want to learn all and i do not know to what to do

How exciting

How Amazing

Thanks

Going through this lab now. What stands out to me is that the real win isn’t just “GPU = faster” — it’s being able to profile the workflow and see exactly where CPU fallbacks are creating bottlenecks. That makes the optimization process much more practical. Excited to apply this to some larger simulation workloads I’m working on.