skills.rest
rdromer2
rdromer2COMMUNITY
Explore all skills by this author →
GitHub

debug:tensorflow

Diagnose TensorFlow debugging issues across shapes, memory, NaN/Inf losses, and GPU visibility.

Updated Aug 23, 2026
Search Tags:#debugging#memory#gpu#tensorflow#nan#profiler#tf.debugging
One-click install
npx skills add https://github.com/rdromer2/mambavision_tf_kernel --skill debug-tensorflow-rdromer2
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: debug:tensorflow
Source: https://github.com/rdromer2/mambavision_tf_kernel/tree/main/.agents/skills/debug-tensorflow
Command: npx skills add https://github.com/rdromer2/mambavision_tf_kernel --skill debug-tensorflow-rdromer2

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

TensorFlow debugging is complex and error-prone, covering shape mismatches, memory issues, NaN/Inf losses, gradient problems, and GPU visibility in diverse environments.

Core Features & Use Cases

  • Systematic diagnosis workflow for common TF errors.
  • Tools and techniques: tf.debugging, TensorBoard Debugger V2, eager execution, memory profiling.
  • Use cases include addressing shape errors, OOM during training, NaN losses, and GPU visibility issues in local or containerized environments.

Quick Start

Run a minimal reproduction, enable eager execution, and add targeted tf.debugging checks to isolate the root cause.

Frequently Asked Questions about debug:tensorflow

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

FAQPage Schema
How do I debug NaN losses in TensorFlow training?▼

Debug NaN losses in TensorFlow by running a minimal reproduction, enabling eager execution, and inserting targeted tf.debugging checks to isolate the exact tensor operations producing invalid values.

What is the best way to resolve TensorFlow GPU visibility issues in containers?▼

Resolve TensorFlow GPU visibility issues in containerized environments by verifying compatible CUDA drivers are installed and applying a systematic diagnosis workflow to check device availability before training.

How do I fix out of memory errors during TensorFlow model training?▼

Fix out of memory errors during TensorFlow training by using memory profiling tools to identify allocation spikes and scoping the workflow to isolate the specific layer or batch size causing the exhaustion.

Does the TensorFlow Debugger V2 work with eager execution?▼

Yes, the TensorBoard Debugger V2 works with eager execution to trace tensor values and graph structures, allowing developers to systematically diagnose shape mismatches and gradient problems end-to-end.

Do I need specific Python or CUDA versions to use tf.debugging tools?▼

Yes, TensorFlow debugging requires Python 3.8 through 3.12 and compatible CUDA drivers when GPU debugging is involved, ensuring the environment supports the TensorFlow 2.x framework features.

Related Skills

Explore More Software Engineering Skills→
yangwhaleyangwhale
9

ray-cluster-installer

Install, configure, and validate Ray clusters on NVIDIA GPUs.

Community
Advanced
AndyAnh174AndyAnh174

python-performance-optimization

Profile Python code to identify CPU and memory bottlenecks.

Community
Advanced
Clay-HHKClay-HHK
2

tensorboard

Visualize training metrics and debug models with TensorBoard event logs.

Community
Intermediate
michaelrizvimichaelrizvi
11

pytorch-debug

Trace PyTorch tensors to diagnose runtime errors and propose fixes.

Community
Intermediate