القائمة
Use when writing or reviewing a PyTorch training loop. Covers correct train/eval modes, gradient handling, mixed precision, checkpointing, reproducibility, and device management.
التثبيت باستخدام Codex أو Claude انسخ هذا Prompt والصقه في Codex أو Claude أو مساعد آخر ليراجع صفحة Skill ويثبّتها لك.
استنادا إلى تصنيف SOC المهني
| name | pytorch-training-loop |
| description | Use when writing or reviewing a PyTorch training loop. Covers correct train/eval modes, gradient handling, mixed precision, checkpointing, reproducibility, and device management. |
A correct PyTorch loop has a precise sequence of operations. Getting the order or the modes wrong produces silent bugs (no gradients, dropout active at eval, leaked compute graphs). This skill encodes the canonical, production-ready loop.
import torch
from torch.amp import autocast, GradScaler
device = "cuda" if torch.cuda.is_available() else "cpu"
model.to(device)
optimizer = torch.optim.AdamW(model.parameters(), lr=3e-4, weight_decay=0.01)
scaler = GradScaler(enabled=(device == "cuda"))
best_val = float("inf")
for epoch in range(num_epochs):
# ---- TRAIN ----
model.train()
for x, y in train_loader:
x, y = x.to(device, non_blocking=True), y.to(device, non_blocking=True)
optimizer.zero_grad(set_to_none=True)
with autocast(device_type=device, enabled=(device == "cuda")):
out = model(x)
loss = criterion(out, y)
scaler.scale(loss).backward()
scaler.unscale_(optimizer)
torch.nn.utils.clip_grad_norm_(model.parameters(), max_norm=1.0)
scaler.step(optimizer)
scaler.update()
# ---- VALIDATE ----
model.eval()
val_loss = 0.0
with torch.no_grad():
for x, y in val_loader:
x, y = x.to(device), y.to(device)
val_loss += criterion(model(x), y).item() * x.size(0)
val_loss /= len(val_loader.dataset)
# ---- CHECKPOINT BEST ----
if val_loss < best_val:
best_val = val_loss
torch.save({"model": model.state_dict(),
"optimizer": optimizer.state_dict(),
"epoch": epoch}, "best.pt")
model.train() before training, model.eval() before validation/inference (toggles dropout & batchnorm).optimizer.zero_grad() every step — gradients accumulate otherwise.torch.no_grad() (or inference_mode()) to save memory.loss.item(), not loss — keeping tensors leaks the graph and OOMs.import torch, numpy as np, random
def seed_everything(seed=42):
random.seed(seed); np.random.seed(seed)
torch.manual_seed(seed); torch.cuda.manual_seed_all(seed)
torch.backends.cudnn.deterministic = True
torch.backends.cudnn.benchmark = False
zero_grad → gradients pile up, training diverges.eval() never called → dropout/batchnorm corrupt validation metrics.total_loss += loss (tensor, not .item()) → memory explosion.num_workers=0 on large data → CPU-bound; raise workers + pin_memory=True.ml-debugging skill.A checkpointed model + seed config that experiment-tracking logs and model-serving loads for inference.
Use when preparing raw data for modeling — handling missing values, duplicates, inconsistent types, outliers, and bad categorical values. Emphasizes fitting all imputation on train-only to avoid leakage.
Use when running ML experiments that need to be compared, reproduced, or shared. Covers MLflow/Weights & Biases logging, what to track, run organization, and model registry basics.
Use when starting on a new dataset, before modeling, or when asked to "explore", "profile", "understand", or "summarize" data. Covers structured EDA, distributions, correlations, target leakage checks, and visualization.
Use when creating, encoding, scaling, or selecting features for ML models. Covers categorical encoding, numeric transforms, datetime/text/aggregation features, and leakage-safe target encoding.
Use when optimizing model hyperparameters. Covers search strategy (random vs Bayesian/Optuna), leakage-safe tuning inside CV, search-space design, early stopping, and budget management.
Use when the target is rare (fraud, churn, disease, anomalies). Covers correct metrics, resampling (SMOTE/undersampling), class weights, threshold tuning, and avoiding the accuracy trap and resampling leakage.