skills.rest
Clay-HHK
Clay-HHKCOMMUNITY
Explore all skills by this author →
GitHub

tensorboard

Visualize training metrics, model graphs, and performance profiles with TensorBoard.

1|Updated Mar 5, 2026
Search Tags:#mlops#profiling#visualization#pytorch#experiment tracking#tensorboard#training metrics
One-click install
npx skills add https://github.com/Clay-HHK/claude-skills --skill tensorboard-clay-hhk
Or copy as Structured Prompt for Agentâ–¼
Please help me install this Agent Skill.
Skill: tensorboard
Source: https://github.com/Clay-HHK/claude-skills/tree/main/tensorboard
Command: npx skills add https://github.com/Clay-HHK/claude-skills --skill tensorboard-clay-hhk

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires tensorboard, torch, tensorflow, and includes references (resource) components.

What problem does it solve? Training machine learning models without visibility into metrics, weight distributions, and performance bottlenecks makes debugging slow and experiment comparison difficult. This Skill provides complete guidance for instrumenting PyTorch and TensorFlow training loops with TensorBoard logging. ## Core Features & Use Cases - Metric & Experiment Tracking: Log scalars, images, histograms, text, and PR curves, then compare multiple runs side-by-side in the TensorBoard dashboard. - Model Debugging: Visualize model graphs, weight and gradient histograms, and project high-dimensional embeddings with PCA, t-SNE, or UMAP. - Performance Profiling: Profile CPU/GPU utilization, memory allocation, and kernel execution to identify training bottlenecks. - Use Case: While fine-tuning a ResNet model, log train/validation loss per epoch, weight histograms every 5 epochs, and sample predictions every 10 epochs, then launch tensorboard --logdir=runs to compare against previous experiments. ## Quick Start Ask the AI to add TensorBoard logging to your PyTorch training loop using SummaryWriter with scalar, histogram, and image logging.

Frequently Asked Questions about tensorboard

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I use TensorBoard with PyTorch?â–¼

Import SummaryWriter from torch.utils.tensorboard, create a writer with a log directory, and call add_scalar, add_histogram, or add_image during training. Launch the dashboard with tensorboard --logdir=runs and open localhost:6006.

How do I compare multiple training runs in TensorBoard?â–¼

Write each experiment to its own subdirectory under a common log directory, such as runs/exp1 and runs/exp2. Launch TensorBoard on the parent directory and all runs appear together, with toggles and regex filters for comparison.

Does TensorBoard work with TensorFlow and Keras?â–¼

Yes, TensorFlow includes TensorBoard natively. Pass tf.keras.callbacks.TensorBoard to model.fit with options like histogram_freq, write_graph, and profile_batch to log metrics, graphs, and profiling data automatically.

How do I profile GPU performance with TensorBoard?â–¼

Use torch.profiler with the tensorboard_trace_handler for PyTorch, or set profile_batch in the Keras TensorBoard callback. The Profile tab shows GPU utilization, kernel execution times, memory allocation, and operation traces.

Why are my TensorBoard log files so large?â–¼

Logging every batch creates excessive event files. Log batch metrics only every 100 steps and epoch-level metrics once per epoch, and reduce histogram frequency to every few epochs to keep log sizes manageable.

Related Skills

Explore More Data & Analytics Skills→
i-bebsii-bebsi

weights-and-biases

Track ML experiments, hyperparameter sweeps, and model artifacts with Weights & Biases.

Community
Advanced
kamelmhkamelmh

fine-tuning-expert

Prepare and tune large language models with LoRA, QLoRA, and PEFT.

Community
Advanced
cabeznocabezno

weights-and-biases

Log ML experiment metrics and manage runs, artifacts, and sweeps in W&B.

Community
Intermediate
shuff57shuff57

weights-and-biases

Automate ML experiment logging and model management with Weights & Biases.

Community
Advanced