En este tutorial, implementamos NVIDIA cuML como un framework de machine learning acelerado por GPU y construimos un flujo de trabajo práctico que demuestra cómo RAPIDS puede acelerar tareas familiares de ciencia de datos y machine learning.
Comenzamos configurando el entorno de GPU y examinando cuml.accel, que nos permite acelerar cargas de trabajo existentes de scikit-learn con cambios mínimos en el código, antes de pasar a la API nativa de cuML para interoperabilidad directa con CuPy y cuDF.
Luego comparamos implementaciones en CPU y GPU de PCA, K-Means, búsqueda de vecinos más cercanos, regresión logística, random forests y DBSCAN, utilizando temporización sincronizada para obtener mediciones de rendimiento significativas.
GPT-6 Astra: guía definitiva de esfuerzo de razonamiento Low, Medium, High, XHigh, Max o Ultra
También construimos pipelines de aprendizaje de variedades (manifold learning) y clustering basados en GPU con UMAP, t-SNE, HDBSCAN y métricas de confiabilidad (trustworthiness); exploramos la inferencia de bosques de alto rendimiento con FIL; validamos explicaciones SHAP generadas en GPU; realizamos optimización de hiperparámetros con meta-estimadores de scikit-learn; y finalmente serializamos modelos entrenados mientras examinamos la portabilidad entre entornos GPU y CPU.
import os
import sys
import time
import json
import shutil
import warnings
import subprocess
import importlib
import traceback
warnings.filterwarnings("ignore")
QUICK = False
SEED = 42
SCALE = 0.25 if QUICK else 1.0
N_MAIN = int(200_000 * SCALE)
D_MAIN = 64
N_RF = int(50_000 * SCALE)
D_RF = 32
N_NN_INDEX = int(50_000 * SCALE)
N_NN_QUERY = int(5_000 * SCALE)
N_DBSCAN = int(20_000 * SCALE)
N_MANIFOLD = int(60_000 * SCALE)
N_ACCEL = int(80_000 * SCALE)
RESULTS = []
NOTES = []
def banner(title):
line = "=" * 78
print(f"\n{line}\n {title}\n{line}", flush=True)
def section(title, fn, *args, **kwargs):
banner(title)
t0 = time.perf_counter()
try:
fn(*args, **kwargs)
except Exception:
print(f"[!] Section skipped due to an error:\n{traceback.format_exc()}")
print(f"[section wall time: {time.perf_counter() - t0:.1f}s]", flush=True)
def bootstrap():
if shutil.which("nvidia-smi") is None:
raise SystemExit(
"No NVIDIA GPU found. In Colab: Runtime > Change runtime type > GPU."
)
print(subprocess.run(
["nvidia-smi",
"--query-gpu=name,memory.total,compute_cap,driver_version",
"--format=csv"],
capture_output=True, text=True).stdout)
try:
import cuml
print("cuML already available — skipping install.")
except ImportError:
print("Installing RAPIDS cuML (this takes ~1-3 minutes)...")
pin = ""
try:
import cudf
major_minor = ".".join(cudf.__version__.split("+")[0].split(".")[:2])
pin = f"=={major_minor}.*"
print(f" Pinning to the preinstalled cuDF line: cuml-cu12{pin}")
except Exception:
print(" cuDF not found; installing the latest stable cuml-cu12.")
cmd = [sys.executable, "-m", "pip", "install", "-q",
"--extra-index-url=https://pypi.nvidia.com", f"cuml-cu12{pin}"]
print("$ " + " ".join(cmd))
rc = subprocess.run(cmd).returncode
if rc != 0:
raise SystemExit(
"pip install failed. Alternative that always works on Colab:\n"
" !git clone https://github.com/rapidsai/rapidsai-csp-utils.git\n"
" !python rapidsai-csp-utils/colab/pip-install.py"
)
importlib.invalidate_caches()
import cuml
import cupy
print(f"cuml {cuml.__version__}")
print(f"cupy {cupy.__version__}")
try:
import cudf
print(f"cudf {cudf.__version__}")
except Exception:
pass
import sklearn
print(f"sklearn {sklearn.__version__} (cuML requires scikit-learn >= 1.6)")
bootstrap()
import numpy as np
import cupy as cp
import cuml
import matplotlib.pyplot as plt
from cuml.datasets import make_classification as gpu_make_classification
from cuml.datasets import make_blobs as gpu_make_blobs
rng = np.random.RandomState(SEED)
cp.random.seed(SEED)
class Timer:
def __init__(self, label, sync=True):
self.label = label
self.sync = sync
def __enter__(self):
if self.sync:
cp.cuda.runtime.deviceSynchronize()
self.t0 = time.perf_counter()
return self
def __exit__(self, *exc):
if self.sync:
cp.cuda.runtime.deviceSynchronize()
self.dt = time.perf_counter() - self.t0
print(f" {self.label:<44s} {self.dt:8.3f}s")
return False
def to_numpy(a):
if isinstance(a, cp.ndarray):
return cp.asnumpy(a)
if hasattr(a, "to_numpy"):
return a.to_numpy()
return np.asarray(a)
def record(task, cpu_s, gpu_s):
RESULTS.append((task, cpu_s, gpu_s))
if cpu_s and gpu_s:
print(f" -> {task}: {cpu_s / gpu_s:.1f}x speedup\n")
ACCEL_SCRIPT = f'''
import time
import numpy as np
from sklearn.datasets import make_blobs
from sklearn.decomposition import PCA
from sklearn.cluster import KMeans
from sklearn.neighbors import NearestNeighbors
from sklearn.linear_model import Ridge
X, y = make_blobs(n_samples={N_ACCEL}, n_features=32, centers=12, random_state=0)
X = X.astype("float32"); y = y.astype("float32")
t0 = time.perf_counter()
PCA(n_components=8).fit_transform(X)
KMeans(n_clusters=12, n_init=1, random_state=0).fit(X)
NearestNeighbors(n_neighbors=8).fit(X[:{N_ACCEL // 2}]).kneighbors(X[:5000])
Ridge(alpha=1.0).fit(X, y)
Ridge(alpha=1.0, positive=True).fit(X[:5000], y[:5000])
print("MODELTIME %.3f" % (time.perf_counter() - t0))
'''
def demo_accel():
path = "/content/_accel_demo.py" if os.path.isdir("/content") else "_accel_demo.py"
with open(path, "w") as f:
f.write(ACCEL_SCRIPT)
def run(cmd, label):
print(f"\n$ {' '.join(cmd[1:])}")
t0 = time.perf_counter()
p = subprocess.run(cmd, capture_output=True, text=True)
wall = time.perf_counter() - t0
out = p.stdout + p.stderr
model_s = None
for line in out.splitlines():
if line.startswith("MODELTIME"):
model_s = float(line.split()[1])
print(out.strip()[:4000])
print(f"[{label}] model time = {model_s}s | process wall = {wall:.1f}s")
return model_s
cpu_s = run([sys.executable, path], "stock sklearn")
cmd = [sys.executable, "-m", "cuml.accel", "--profile", path]
gpu_s = run(cmd, "cuml.accel")
if gpu_s is None:
gpu_s = run([sys.executable, "-m", "cuml.accel", path], "cuml.accel")
record("cuml.accel (sklearn script, unmodified)", cpu_s, gpu_s)
NOTES.append(
"cuml.accel needed ZERO source changes; the profile table above shows "
"which calls ran on GPU and why Ridge(positive=True) fell back to CPU."
)
Configuramos el entorno del tutorial, definimos tamaños de conjuntos de datos y utilidades de benchmarking, y verificamos que haya una GPU NVIDIA disponible.
Instalamos e inicializamos RAPIDS cuML cuando es necesario, configuramos CuPy y controles de reproducibilidad, y creamos ayudas de temporización sincronizada y seguimiento de resultados.
¿Qué es la Superinteligencia? El horizonte inalcanzable e imposible de controlar
También demostramos cuml.accel ejecutando una carga de trabajo de scikit-learn sin modificar y comparando su ejecución en CPU con la ejecución acelerada por GPU.
def demo_native_api():
from cuml.preprocessing import StandardScaler
from cuml.model_selection import train_test_split
X, y = gpu_make_blobs(n_samples=50_000, n_features=8, centers=5,
random_state=SEED, dtype=np.float32)
print(f"cuml.datasets output lives on device: {type(X).__module__}, "
f"shape={X.shape}, dtype={X.dtype}")
try:
import cudf
df = cudf.DataFrame(X, columns=[f"f{i}" for i in range(X.shape[1])])
back = df.values
ptr_a = X.__cuda_array_interface__["data"][0]
ptr_b = back.__cuda_array_interface__["data"][0]
print(f"CuPy ptr = {hex(ptr_a)}")
print(f"cuDF->CuPy= {hex(ptr_b)}")
print("Same device pointer (true zero-copy)? ", ptr_a == ptr_b)
print("Note: a column-major DataFrame round trip may re-pack; what "
"matters is that no host (CPU) round trip ever happens.")
scaled = StandardScaler().fit_transform(df)
print(f"StandardScaler(cuDF) -> {type(scaled).__name__}")
except Exception as e:
print(f"cuDF interop skipped: {e}")
from cuml.decomposition import PCA
pca = PCA(n_components=3).fit(X)
print(f"default (mirrors input) -> {type(pca.transform(X)).__name__}")
with cuml.using_output_type("numpy"):
print(f"inside using_output_type() -> {type(pca.transform(X)).__name__}")
print(f"after the context manager -> {type(pca.transform(X)).__name__}")
NOTES.append(
"Keep output_type as CuPy/cuDF inside a pipeline; converting to NumPy "
"on every step forces a device->host copy and eats the speedup."
)
Xtr, Xte, ytr, yte = train_test_split(X, y, test_size=0.2, random_state=SEED)
print(f"train_test_split -> {Xtr.shape} / {Xte.shape}, still on device: "
f"{isinstance(Xtr, cp.ndarray)}")
Trabajamos directamente con la API nativa de cuML y exploramos cómo los datos residentes en GPU se mueven entre CuPy, cuDF y componentes de cuML.
Inspeccionamos punteros de dispositivo para entender la interoperabilidad de copia cero y usamos los controles de tipo de salida de cuML para gestionar si los resultados permanecen en la GPU o regresan como arreglos NumPy.
También realizamos una división train-test nativa de GPU para que nuestros datos permanezcan en el dispositivo a lo largo del flujo de trabajo.
def demo_benchmarks():
from sklearn.decomposition import PCA as skPCA
from sklearn.cluster import KMeans as skKMeans, DBSCAN as skDBSCAN
from sklearn.neighbors import NearestNeighbors as skNN
from sklearn.linear_model import LogisticRegression as skLR
from sklearn.ensemble import RandomForestClassifier as skRF
from cuml.decomposition import PCA as cuPCA
from cuml.cluster import KMeans as cuKMeans, DBSCAN as cuDBSCAN
from cuml.neighbors import NearestNeighbors as cuNN
from cuml.linear_model import LogisticRegression as cuLR
from cuml.ensemble import RandomForestClassifier as cuRF
print(f"Generating {N_MAIN:,} x {D_MAIN} on the GPU...")
Xg, yg = gpu_make_classification(n_samples=N_MAIN, n_features=D_MAIN,
n_informative=32, n_classes=4,
random_state=SEED)
Xg = Xg.astype(cp.float32)
yg = yg.astype(cp.int32)
Xc, yc = cp.asnumpy(Xg), cp.asnumpy(yg)
print(f" device array: {Xg.nbytes / 1e6:.0f} MB\n")
print("PCA (n_components=16)")
with Timer("sklearn", sync=False) as t:
skPCA(n_components=16, random_state=SEED).fit_transform(Xc)
cpu = t.dt
with Timer("cuML") as t:
cuPCA(n_components=16, random_state=SEED).fit_transform(Xg)
record("PCA", cpu, t.dt)
print("KMeans (k=16)")
with Timer("sklearn", sync=False) as t:
skKMeans(n_clusters=16, n_init=1, max_iter=100,
random_state=SEED).fit(Xc)
cpu = t.dt
with Timer("cuML") as t:
cuKMeans(n_clusters=16, n_init=1, max_iter=100,
random_state=SEED).fit(Xg)
record("KMeans", cpu, t.dt)
print(f"NearestNeighbors k=16 ({N_NN_INDEX:,} index / {N_NN_QUERY:,} query)")
idx_g, q_g = Xg[:N_NN_INDEX], Xg[N_NN_INDEX:N_NN_INDEX + N_NN_QUERY]
idx_c, q_c = cp.asnumpy(idx_g), cp.asnumpy(q_g)
with Timer("sklearn (brute)", sync=False) as t:
skNN(n_neighbors=16, algorithm="brute", n_jobs=-1).fit(idx_c).kneighbors(q_c)
cpu = t.dt
with Timer("cuML") as t:
d_gpu, i_gpu = cuNN(n_neighbors=16).fit(idx_g).kneighbors(q_g)
record("NearestNeighbors", cpu, t.dt)
print("LogisticRegression (multinomial, lbfgs/QN)")
with Timer("sklearn", sync=False) as t:
sk_lr = skLR(max_iter=200, n_jobs=-1).fit(Xc, yc)
cpu = t.dt
with Timer("cuML") as t:
cu_lr = cuLR(max_iter=200).fit(Xg, yg)
record("LogisticRegression", cpu, t.dt)
print(f" accuracy sklearn={sk_lr.score(Xc, yc):.4f} "
f"cuML={float((cu_lr.predict(Xg) == yg).mean()):.4f} "
"(different solvers, so small differences are expected)\n")
print(f"RandomForestClassifier (100 trees, depth 12, {N_RF:,} x {D_RF})")
Xr_g, yr_g = gpu_make_classification(n_samples=N_RF, n_features=D_RF,
n_informative=16, n_classes=2,
random_state=SEED)
Xr_g = Xr_g.astype(cp.float32)
yr_g = yr_g.astype(cp.int32)
Xr_c, yr_c = cp.asnumpy(Xr_g), cp.asnumpy(yr_g)
with Timer("sklearn", sync=False) as t:
skRF(n_estimators=100, max_depth=12, n_jobs=-1,
random_state=SEED).fit(Xr_c, yr_c)
cpu = t.dt
with Timer("cuML") as t:
cu_rf = cuRF(n_estimators=100, max_depth=12, n_bins=128,
n_streams=4, random_state=SEED).fit(Xr_g, yr_g)
record("RandomForest (fit)", cpu, t.dt)
globals()["_RF_ARTIFACTS"] = (cu_rf, Xr_g, yr_g, Xr_c, yr_c)
print(f"DBSCAN ({N_DBSCAN:,} x 8)")
Xd_g, _ = gpu_make_blobs(n_samples=N_DBSCAN, n_features=8, centers=6,
cluster_std=0.6, random_state=SEED,
dtype=np.float32)
Xd_c = cp.asnumpy(Xd_g)
with Timer("sklearn", sync=False) as t:
lab_c = skDBSCAN(eps=0.9, min_samples=8, n_jobs=-1).fit_predict(Xd_c)
cpu = t.dt
with Timer("cuML") as t:
lab_g = cuDBSCAN(eps=0.9, min_samples=8).fit_predict(Xd_g)
record("DBSCAN", cpu, t.dt)
print(f" clusters found: sklearn={len(set(lab_c.tolist())) - 1}, "
f"cuML={len(set(cp.asnumpy(lab_g).tolist())) - 1}\n")
Comparamos implementaciones de scikit-learn y cuML de PCA, K-Means, vecinos más cercanos, regresión logística, random forests y DBSCAN.
Generamos conjuntos de datos en la GPU, sincronizamos las operaciones CUDA para una temporización justa y registramos la aceleración que logra cada algoritmo acelerado.
¿Qué es la desalineación de la IA? A qué deberían temer los seres humanos de su desarrollo
También comparamos el comportamiento de los modelos y conservamos el random forest de cuML entrenado para reutilizarlo más adelante en el tutorial.
def demo_manifold():
from cuml.manifold import UMAP, TSNE
from cuml.metrics import trustworthiness
X, y = gpu_make_blobs(n_samples=N_MANIFOLD, n_features=48, centers=8,
cluster_std=1.6, random_state=SEED, dtype=np.float32)
print(f"data: {X.shape}")
embeddings = {}
for n_neighbors, min_dist in [(15, 0.1), (50, 0.0)]:
key = f"UMAP(n_neighbors={n_neighbors}, min_dist={min_dist})"
with Timer(key) as t:
emb = UMAP(n_neighbors=n_neighbors, min_dist=min_dist,
n_components=2, random_state=SEED).fit_transform(X)
sub = slice(0, min(5000, X.shape[0]))
tw = trustworthiness(X[sub], emb[sub], n_neighbors=10)
print(f" trustworthiness = {tw:.4f}")
embeddings[key] = (emb, t.dt, tw)
with Timer("TSNE(method='fft')") as t:
tsne_emb = TSNE(n_components=2, perplexity=30,
random_state=SEED).fit_transform(X)
embeddings["TSNE"] = (tsne_emb, t.dt, float("nan"))
best_key = max([k for k in embeddings if k.startswith("UMAP")],
key=lambda k: embeddings[k][2])
emb = embeddings[best_key][0]
print(f"\nClustering the '{best_key}' embedding with GPU HDBSCAN")
try:
from cuml.cluster import HDBSCAN
with Timer("HDBSCAN") as t:
hdb = HDBSCAN(min_cluster_size=max(int(50 * SCALE), 5),
min_samples=10, prediction_data=True).fit(emb)
labels = cp.asarray(hdb.labels_)
n_clusters = int(labels.max()) + 1
noise = float((labels == -1).mean())
print(f" clusters={n_clusters} noise fraction={noise:.3f}")
try:
from cuml.metrics.cluster import adjusted_rand_score
print(f" adjusted Rand index vs ground truth: "
f"{adjusted_rand_score(y, labels):.4f}")
except Exception as e:
print(f" ARI skipped: {e}")
try:
from cuml.cluster.hdbscan import all_points_membership_vectors
mv = all_points_membership_vectors(hdb)
print(f" soft-cluster membership matrix: {tuple(mv.shape)}")
except Exception as e:
print(f" soft clustering skipped: {e}")
except Exception as e:
print(f" HDBSCAN step skipped: {e}")
fig, axes = plt.subplots(1, 3, figsize=(16, 5))
keys = list(embeddings)[:3]
for ax, k in zip(axes, keys):
e = to_numpy(embeddings[k][0])
c = to_numpy(y)
ax.scatter(e[:, 0], e[:, 1], c=c, s=1.5, cmap="tab10", alpha=0.6)
ax.set_title(f"{k}\n{embeddings[k][1]:.2f}s", fontsize=9)
ax.set_xticks([]); ax.set_yticks([])
plt.suptitle("GPU manifold learning (colored by ground-truth cluster)")
plt.tight_layout(); plt.show()
Construimos un pipeline no supervisado en GPU usando UMAP y t-SNE para reducir datos de alta dimensionalidad a embeddings bidimensionales.
Evaluamos configuraciones de UMAP con la métrica de confiabilidad (trustworthiness), seleccionamos el mejor embedding y aplicamos HDBSCAN para identificar clústeres y puntos de ruido.
Luego visualizamos los embeddings resultantes y comparamos sus estructuras usando las etiquetas conocidas de los clústeres de referencia.
def demo_fil():
from sklearn.ensemble import RandomForestClassifier as skRF
n = int(30_000 * SCALE)
Xg, yg = gpu_make_classification(n_samples=n, n_features=24,
n_informative=12, n_classes=2,
random_state=SEED)
Xg = Xg.astype(cp.float32)
Xc, yc = cp.asnumpy(Xg), cp.asnumpy(yg.astype(cp.int32))
print("Training a 200-tree sklearn forest on CPU (the model to be served)...")
sk_model = skRF(n_estimators=200, max_depth=10, n_jobs=-1,
random_state=SEED).fit(Xc, yc)
with Timer("sklearn.predict_proba (CPU)", sync=False) as t:
p_cpu = sk_model.predict_proba(Xc)[:, 1]
cpu = t.dt
fil = None
try:
from cuml.fil import ForestInference
except Exception as e:
print(f"cuml.fil unavailable on this build ({e}); skipping. "
"On newer stacks use the standalone nvForest library instead.")
return
for kwargs in ({"is_classifier": True, "output_type": "numpy"},
{"output_class": True, "output_type": "numpy"},
{}):
try:
fil = ForestInference.load_from_sklearn(sk_model, **kwargs)
print(f"Loaded into FIL with kwargs={kwargs or '{}'}")
break
except Exception as e:
print(f" load_from_sklearn(**{kwargs}) -> {type(e).__name__}: {e}")
if fil is None:
print("Could not load the forest into FIL on this build; skipping.")
return
try:
fil.optimize(batch_size=Xg.shape[0])
print("Ran fil.optimize() to auto-tune layout/chunk size for this batch.")
except Exception:
pass
fil.predict_proba(Xg[:1024])
with Timer("FIL predict_proba (GPU)") as t:
p_gpu = fil.predict_proba(Xg)
record("Forest inference (200 trees)", cpu, t.dt)
p_gpu = to_numpy(p_gpu)
p_gpu = p_gpu[:, 1] if p_gpu.ndim == 2 and p_gpu.shape[1] == 2 else p_gpu.ravel()
print(f" max |prob difference| vs sklearn: {np.abs(p_gpu - p_cpu).max():.2e} "
"(FIL defaults to float32, so ~1e-6 is normal)")
NOTES.append(
"FIL/nvForest is the piece that matters in production: the same trained "
"artifact, served with GPU-class throughput and no retraining."
)
Nos enfocamos en acelerar la inferencia de modelos basados en árboles después del entrenamiento.
Entrenamos un random forest de scikit-learn en la CPU, lo cargamos en la Forest Inference Library de cuML cuando es compatible, y optimizamos la configuración de inferencia para el tamaño de lote actual de la GPU.
El muro de datos de la inteligencia artificial es ya, un hecho completamente real
Comparamos los tiempos de predicción en CPU y GPU y validamos que las probabilidades predichas sigan siendo numéricamente consistentes.
def demo_explainer():
from cuml.linear_model import Ridge
from cuml.explainer import PermutationExplainer
n, d = int(20_000 * SCALE), 12
X = cp.asarray(rng.randn(n, d), dtype=cp.float32)
true_coef = cp.asarray(rng.uniform(-3, 3, size=d), dtype=cp.float32)
y = (X @ true_coef + 0.1 * cp.asarray(rng.randn(n), dtype=cp.float32))
model = Ridge(alpha=1e-3).fit(X, y)
coef = cp.asarray(model.coef_).ravel()
background = X[:200]
to_explain = X[200:220]
with Timer("PermutationExplainer (GPU)") as t:
expl = PermutationExplainer(model=model.predict, data=background,
random_state=SEED)
shap_values = expl.shap_values(to_explain)
shap_values = cp.asarray(shap_values)
analytic = (to_explain - background.mean(axis=0)) * coef
err = float(cp.abs(shap_values - analytic).max())
print(f" max |SHAP - analytical linear SHAP| = {err:.4f}")
print(" (permutation SHAP is sampling-based, so a small residual is "
"expected; the pattern must match)")
base = float(model.predict(background).mean())
recon = cp.asarray(shap_values).sum(axis=1) + base
actual = cp.asarray(model.predict(to_explain)).ravel()
print(f" additivity residual (mean |sum(phi)+base - f(x)|) = "
f"{float(cp.abs(recon - actual).mean()):.4f}")
imp = to_numpy(cp.abs(shap_values).mean(axis=0))
order = np.argsort(imp)[::-1]
plt.figure(figsize=(8, 3.5))
plt.bar(range(d), imp[order], color="#76b900")
plt.xticks(range(d), [f"f{i}" for i in order])
plt.ylabel("mean |SHAP|")
plt.title("GPU SHAP feature importance (cuml.explainer.PermutationExplainer)")
plt.tight_layout(); plt.show()
def demo_hpo():
from sklearn.model_selection import RandomizedSearchCV
from cuml.ensemble import RandomForestClassifier as cuRF
n = int(60_000 * SCALE)
X, y = gpu_make_classification(n_samples=n, n_features=24, n_informative=14,
n_classes=3, random_state=SEED)
X = cp.asnumpy(X.astype(cp.float32))
y = cp.asnumpy(y.astype(cp.int32))
param_dist = {
"n_estimators": [50, 100, 200],
"max_depth": [8, 12, 16],
"max_features": [0.3, 0.5, 0.8],
"n_bins": [64, 128, 256],
}
search = RandomizedSearchCV(
cuRF(random_state=SEED, n_streams=1),
param_distributions=param_dist,
n_iter=8, cv=3, n_jobs=1, random_state=SEED, verbose=0,
)
with Timer("RandomizedSearchCV over cuML RF (8 x 3 fits)", sync=True) as t:
search.fit(X, y)
print(f" best CV accuracy: {search.best_score_:.4f}")
print(f" best params : {json.dumps(search.best_params_)}")
NOTES.append(
"Because each fit is seconds instead of minutes, you can afford a real "
"search space instead of one hand-tuned guess."
)
Usamos el explicador de permutación basado en GPU de cuML para calcular valores SHAP de un modelo de regresión Ridge y validamos esas explicaciones contra la solución lineal analítica.
Probamos la aditividad de SHAP y visualizamos la importancia de las características para confirmar que las atribuciones calculadas se comportan como se espera.
¿Cómo RTX Spark y Windows podrían convertir los agentes de IA en una herramienta cotidiana?
También combinamos estimadores de cuML con RandomizedSearchCV de scikit-learn para realizar optimización de hiperparámetros con validación cruzada mientras ajustamos el modelo en la GPU.
def demo_persistence():
import pickle
art = globals().get("_RF_ARTIFACTS")
if art is None:
from cuml.ensemble import RandomForestClassifier as cuRF
Xg, yg = gpu_make_classification(n_samples=int(20_000 * SCALE),
n_features=16, n_classes=2,
random_state=SEED)
Xg = Xg.astype(cp.float32); yg = yg.astype(cp.int32)
model = cuRF(n_estimators=50, max_depth=10, random_state=SEED).fit(Xg, yg)
else:
model, Xg, yg, _, _ = art
before = to_numpy(model.predict(Xg[:1000]))
path = "/content/cuml_rf.pkl" if os.path.isdir("/content") else "cuml_rf.pkl"
with open(path, "wb") as f:
pickle.dump(model, f)
size_mb = os.path.getsize(path) / 1e6
with open(path, "rb") as f:
restored = pickle.load(f)
after = to_numpy(restored.predict(Xg[:1000]))
print(f" pickled model: {size_mb:.2f} MB at {path}")
print(f" predictions identical after round trip: {np.array_equal(before, after)}")
print(" cuML uses cloudpickle internally, so models trained under "
"cuml.accel can be loaded and used by plain scikit-learn on a "
"CPU-only machine.")
print(" SECURITY: never unpickle a model file from an untrusted source.")
def demo_summary():
rows = [(t, c, g) for (t, c, g) in RESULTS if c and g]
if not rows:
print("No comparable timings were collected.")
return
w = max(len(r[0]) for r in rows)
print(f"{'task'.ljust(w)} {'CPU (s)':>9} {'GPU (s)':>9} {'speedup':>9}")
print("-" * (w + 32))
for t, c, g in rows:
print(f"{t.ljust(w)} {c:9.3f} {g:9.3f} {c / g:8.1f}x")
labels = [r[0] for r in rows][::-1]
speeds = [r[1] / r[2] for r in rows][::-1]
plt.figure(figsize=(9, 0.55 * len(labels) + 2))
bars = plt.barh(labels, speeds, color="#76b900")
for b, s in zip(bars, speeds):
plt.text(b.get_width() * 1.02, b.get_y() + b.get_height() / 2,
f"{s:.1f}x", va="center", fontsize=9)
plt.axvline(1.0, color="grey", ls="--", lw=1)
plt.xscale("log")
plt.xlabel("speedup vs CPU (log scale, higher is better)")
plt.title(f"cuML {cuml.__version__} on this Colab GPU")
plt.tight_layout(); plt.show()
print("\nTakeaways")
for i, n in enumerate(NOTES, 1):
print(f" {i}. {n}")
print("""
Caveats worth internalizing:
* Speedups are size-dependent. Under ~10k rows, PCIe transfer and kernel
launch overhead usually dominate, and the CPU wins. Benchmark YOUR shapes.
* Always deviceSynchronize() before stopping a timer, or you time nothing.
* cuML matches scikit-learn's API, not its exact numerics: different solvers,
float32 defaults, and non-deterministic reductions produce small deltas.
* Multi-GPU / multi-node: swap cuml.X for cuml.dask.X with a LocalCUDACluster.
Where to go next:
* cuml.accel compatibility matrix : https://docs.nvidia.com/cuml/stable/cuml-accel/compatibility/
* Profiling accelerated code : %%cuml.accel.profile and %%cuml.accel.line_profile
* Multi-GPU guide : https://docs.nvidia.com/cuml/stable/dask_multigpu_guide/
* Walkthrough notebooks : https://github.com/NVIDIA/cuml/tree/main/notebooks
""")
_t_all = time.perf_counter()
section("1. cuml.accel — zero code change acceleration of stock scikit-learn",
demo_accel)
section("2. Native cuML API: cuDF/CuPy interop, zero-copy, output types",
demo_native_api)
section("3. CPU vs GPU benchmark harness", demo_benchmarks)
section("4. UMAP -> trustworthiness -> HDBSCAN pipeline", demo_manifold)
section("5. High-throughput forest inference (FIL / nvForest)", demo_fil)
section("6. GPU SHAP with cuml.explainer, validated analytically", demo_explainer)
section("7. Hyperparameter search over cuML estimators", demo_hpo)
section("8. Serialization and GPU -> CPU portability", demo_persistence)
section("9. Summary", demo_summary)
print(f"\nTotal tutorial wall time: {time.perf_counter() - _t_all:.1f}s")
Serializamos un random forest de cuML entrenado con pickle, lo restauramos y verificamos que sus predicciones permanezcan sin cambios después del viaje de ida y vuelta.
Agregamos los resultados de temporización de CPU y GPU recolectados a lo largo del tutorial y visualizamos las aceleraciones resultantes en un gráfico logarítmico.
Finalmente, ejecutamos cada sección del tutorial en secuencia, imprimimos las conclusiones prácticas acumuladas y reportamos el tiempo total de ejecución del flujo de trabajo completo.
Conclusión
Implementamos una comprensión integral de cómo NVIDIA cuML integra la aceleración por GPU tanto en flujos de trabajo existentes de scikit-learn como en pipelines de machine learning completamente nativos de GPU.
Comparamos el rendimiento computacional en varios algoritmos centrales, gestionamos datos residentes en dispositivo de manera eficiente con CuPy y cuDF, evaluamos representaciones no supervisadas y calidad de clustering, aceleramos la inferencia de modelos de árboles, y generamos explicaciones SHAP interpretables directamente en la GPU.
Gemini CLI en Linux, macOS y Windows (configuración del Agente)
También mostramos que utilidades familiares de scikit-learn como RandomizedSearchCV pueden funcionar junto a estimadores de cuML, preservando patrones establecidos de desarrollo de machine learning mientras se benefician de la ejecución en GPU.
Descubre más desde CIBERED
Suscríbete y recibe las últimas entradas en tu correo electrónico.
