cuml-machine-learning

Name: cuml-machine-learning
Author: langchain-ai

218installs3.7kforks

How to install cuml-machine-learning?

langchain-ai/deepagents/cuml-machine-learning

$npx -y skills add langchain-ai/deepagents --skill cuml-machine-learning

Installs into the current project.

Summary

Use for GPU-accelerated machine learning on tabular data using NVIDIA cuML. Triggers when tasks involve classification, regression, clustering, dimensionality reduction, or model training on datasets.

Files · 1

View on GitHub

SKILL.md

1	# cuML Machine Learning Skill
2
3	GPU-accelerated machine learning using NVIDIA RAPIDS cuML. cuML provides a scikit-learn-compatible API that runs on NVIDIA GPUs, enabling massive speedups on large datasets.
4
5	## When to Use This Skill
6
7	Use this skill when:
8	- Training classification models (predict categories, detect fraud, classify text)
9	- Training regression models (forecast values, predict prices, estimate quantities)
10	- Clustering data (segment customers, group documents, find patterns)
11	- Dimensionality reduction (visualize high-dimensional data, compress features)
12	- Preprocessing and feature engineering on large datasets
13	- Any ML task on datasets with 10K+ rows where GPU acceleration helps
14
15	## Initialization (REQUIRED)
16
17	Always start every script with this boilerplate. It tests actual GPU ML operations.
18
19	```python
20	import pandas as pd
21	import numpy as np
22
23	try:
24	import cudf
25	import cuml
26	# Smoke-test: verify GPU ML works end-to-end
27	_test_data = cudf.DataFrame({'a': [1.0, 2.0, 3.0, 4.0], 'b': [5.0, 6.0, 7.0, 8.0]})
28	_km = cuml.cluster.KMeans(n_clusters=2, n_init=1, random_state=42)
29	_km.fit(_test_data)
30	assert len(_km.labels_) == 4
31	GPU = True
32	except Exception as e:
33	print(f"[GPU] cuml unavailable, falling back to scikit-learn: {e}")
34	GPU = False
35
36	def read_csv(path):
37	return cudf.read_csv(path) if GPU else pd.read_csv(path)
38
39	def to_pd(df):
40	"""Convert cuML/cuDF output to pandas. Use this instead of .to_pandas() directly."""
41	if not GPU:
42	return df
43	try:
44	return df.to_pandas()
45	except Exception as e:
46	print(f"[GPU] .to_pandas() failed, using Arrow fallback: {e}")
47	return df.to_arrow().to_pandas()
48	```
49
50	## Import Patterns
51
52	```python
53	# GPU mode
54	if GPU:
55	from cuml.cluster import KMeans, DBSCAN, HDBSCAN
56	from cuml.ensemble import RandomForestClassifier, RandomForestRegressor
57	from cuml.linear_model import LinearRegression, Ridge, Lasso, LogisticRegression
58	from cuml.neighbors import KNeighborsClassifier, KNeighborsRegressor
59	from cuml.svm import SVC, SVR
60	from cuml.decomposition import PCA, TruncatedSVD
61	from cuml.manifold import UMAP, TSNE
62	from cuml.preprocessing import StandardScaler, MinMaxScaler, LabelEncoder
63	from cuml.model_selection import train_test_split
64	from cuml.metrics import accuracy_score, r2_score, mean_squared_error
65	# CPU fallback
66	else:
67	from sklearn.cluster import KMeans, DBSCAN, HDBSCAN
68	from sklearn.ensemble import RandomForestClassifier, RandomForestRegressor
69	from sklearn.linear_model import LinearRegression, Ridge, Lasso, LogisticRegression
70	from sklearn.neighbors import KNeighborsClassifier, KNeighborsRegressor
71	from sklearn.svm import SVC, SVR
72	from sklearn.decomposition import PCA, TruncatedSVD
73	from sklearn.manifold import TSNE
74	from sklearn.preprocessing import StandardScaler, MinMaxScaler, LabelEncoder
75	from sklearn.model_selection import train_test_split
76	from sklearn.metrics import accuracy_score, r2_score, mean_squared_error
77	# UMAP not in sklearn — skip or pip install umap-learn
78	```
79
80	## Quick Reference
81
82	### Train/Test Split (Start Here)
83
84	```python
85	X = df[["feature1", "feature2", "feature3"]].astype("float32")
86	y = df["target"]
87
88	X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)
89	```
90
91	### Classification
92
93	```python
94	model = RandomForestClassifier(n_estimators=100, max_depth=10, random_state=42)
95	model.fit(X_train, y_train)
96
97	predictions = model.predict(X_test)
98	accuracy = float(accuracy_score(to_pd(y_test), to_pd(predictions)))
99	print(f"Accuracy: {accuracy:.4f}")
100
101	# Feature importances (tree models only)
102	importances = to_pd(model.feature_importances_)
103	for name, imp in zip(feature_names, importances):
104	print(f" {name}: {imp:.4f}")
105	```
106
107	### Regression
108
109	```python
110	model = Ridge(alpha=1.0)
111	model.fit(X_train, y_train)
112
113	predictions = model.predict(X_test)
114	r2 = float(r2_score(to_pd(y_test), to_pd(predictions)))
115	mse = float(mean_squared_error(to_pd(y_test), to_pd(predictions)))
116	print(f"R² Score: {r2:.4f}")
117	print(f"MSE: {mse:.4f}")
118
119	# Coefficients
120	coeffs = to_pd(model.coef_)
121	print(f"Intercept: {float(model.intercept_):.4f}")
122	```
123
124	### Clustering (KMeans)
125
126	```python
127	X = df[["feature1", "feature2"]].astype("float32")
128
129	model = KMeans(n_clusters=4, n_init=10, random_state=42)
130	model.fit(X)
131
132	labels = to_pd(model.labels_)
133	centroids = to_pd(model.cluster_centers_)
134	inertia = float(model.inertia_)
135
136	print(f"Inertia: {inertia:.2f}")
137	print(f"Cluster sizes: {labels.value_counts().sort_index().to_dict()}")
138	print(f"Centroids:\n{centroids}")
139	```
140
141	### Dimensionality Reduction (PCA)
142
143	```python
144	scaler = StandardScaler()
145	X_scaled = scaler.fit_transform(X.astype("float32"))
146
147	pca = PCA(n_components=3)
148	X_

People who liked this also viewed

6 picks

Type

cuml-machine-learning

218installs3.7kforks

View on GitHub

How to install cuml-machine-learning?

langchain-ai/deepagents/cuml-machine-learning

$npx -y skills add langchain-ai/deepagents --skill cuml-machine-learning

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# cuML Machine Learning Skill
2
3	GPU-accelerated machine learning using NVIDIA RAPIDS cuML. cuML provides a scikit-learn-compatible API that runs on NVIDIA GPUs, enabling massive speedups on large datasets.
4
5	## When to Use This Skill
6
7	Use this skill when:
8	- Training classification models (predict categories, detect fraud, classify text)
9	- Training regression models (forecast values, predict prices, estimate quantities)
10	- Clustering data (segment customers, group documents, find patterns)
11	- Dimensionality reduction (visualize high-dimensional data, compress features)
12	- Preprocessing and feature engineering on large datasets
13	- Any ML task on datasets with 10K+ rows where GPU acceleration helps
14
15	## Initialization (REQUIRED)
16
17	Always start every script with this boilerplate. It tests actual GPU ML operations.
18
19	```python
20	import pandas as pd
21	import numpy as np
22
23	try:
24	import cudf
25	import cuml
26	# Smoke-test: verify GPU ML works end-to-end
27	_test_data = cudf.DataFrame({'a': [1.0, 2.0, 3.0, 4.0], 'b': [5.0, 6.0, 7.0, 8.0]})
28	_km = cuml.cluster.KMeans(n_clusters=2, n_init=1, random_state=42)
29	_km.fit(_test_data)
30	assert len(_km.labels_) == 4
31	GPU = True
32	except Exception as e:
33	print(f"[GPU] cuml unavailable, falling back to scikit-learn: {e}")
34	GPU = False
35
36	def read_csv(path):
37	return cudf.read_csv(path) if GPU else pd.read_csv(path)
38
39	def to_pd(df):
40	"""Convert cuML/cuDF output to pandas. Use this instead of .to_pandas() directly."""
41	if not GPU:
42	return df
43	try:
44	return df.to_pandas()
45	except Exception as e:
46	print(f"[GPU] .to_pandas() failed, using Arrow fallback: {e}")
47	return df.to_arrow().to_pandas()
48	```
49
50	## Import Patterns
51
52	```python
53	# GPU mode
54	if GPU:
55	from cuml.cluster import KMeans, DBSCAN, HDBSCAN
56	from cuml.ensemble import RandomForestClassifier, RandomForestRegressor
57	from cuml.linear_model import LinearRegression, Ridge, Lasso, LogisticRegression
58	from cuml.neighbors import KNeighborsClassifier, KNeighborsRegressor
59	from cuml.svm import SVC, SVR
60	from cuml.decomposition import PCA, TruncatedSVD
61	from cuml.manifold import UMAP, TSNE
62	from cuml.preprocessing import StandardScaler, MinMaxScaler, LabelEncoder
63	from cuml.model_selection import train_test_split
64	from cuml.metrics import accuracy_score, r2_score, mean_squared_error
65	# CPU fallback
66	else:
67	from sklearn.cluster import KMeans, DBSCAN, HDBSCAN
68	from sklearn.ensemble import RandomForestClassifier, RandomForestRegressor
69	from sklearn.linear_model import LinearRegression, Ridge, Lasso, LogisticRegression
70	from sklearn.neighbors import KNeighborsClassifier, KNeighborsRegressor
71	from sklearn.svm import SVC, SVR
72	from sklearn.decomposition import PCA, TruncatedSVD
73	from sklearn.manifold import TSNE
74	from sklearn.preprocessing import StandardScaler, MinMaxScaler, LabelEncoder
75	from sklearn.model_selection import train_test_split
76	from sklearn.metrics import accuracy_score, r2_score, mean_squared_error
77	# UMAP not in sklearn — skip or pip install umap-learn
78	```
79
80	## Quick Reference
81
82	### Train/Test Split (Start Here)
83
84	```python
85	X = df[["feature1", "feature2", "feature3"]].astype("float32")
86	y = df["target"]
87
88	X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)
89	```
90
91	### Classification
92
93	```python
94	model = RandomForestClassifier(n_estimators=100, max_depth=10, random_state=42)
95	model.fit(X_train, y_train)
96
97	predictions = model.predict(X_test)
98	accuracy = float(accuracy_score(to_pd(y_test), to_pd(predictions)))
99	print(f"Accuracy: {accuracy:.4f}")
100
101	# Feature importances (tree models only)
102	importances = to_pd(model.feature_importances_)
103	for name, imp in zip(feature_names, importances):
104	print(f" {name}: {imp:.4f}")
105	```
106
107	### Regression
108
109	```python
110	model = Ridge(alpha=1.0)
111	model.fit(X_train, y_train)
112
113	predictions = model.predict(X_test)
114	r2 = float(r2_score(to_pd(y_test), to_pd(predictions)))
115	mse = float(mean_squared_error(to_pd(y_test), to_pd(predictions)))
116	print(f"R² Score: {r2:.4f}")
117	print(f"MSE: {mse:.4f}")
118
119	# Coefficients
120	coeffs = to_pd(model.coef_)
121	print(f"Intercept: {float(model.intercept_):.4f}")
122	```
123
124	### Clustering (KMeans)
125
126	```python
127	X = df[["feature1", "feature2"]].astype("float32")
128
129	model = KMeans(n_clusters=4, n_init=10, random_state=42)
130	model.fit(X)
131
132	labels = to_pd(model.labels_)
133	centroids = to_pd(model.cluster_centers_)
134	inertia = float(model.inertia_)
135
136	print(f"Inertia: {inertia:.2f}")
137	print(f"Cluster sizes: {labels.value_counts().sort_index().to_dict()}")
138	print(f"Centroids:\n{centroids}")
139	```
140
141	### Dimensionality Reduction (PCA)
142
143	```python
144	scaler = StandardScaler()
145	X_scaled = scaler.fit_transform(X.astype("float32"))
146
147	pca = PCA(n_components=3)
148	X_

People who liked this also viewed

6 picks

Type