Ready to supercharge your AI agents?

Install any skill in seconds. Free to start, no credit card required.

Pytorch Model Trainer · dicklesworthstone | DecimalAI

dicklesworthstone/pytorch-model-trainer

SkillScore100Maintained

Build pytorch model trainer operations. Auto-activating skill for ML Training. Triggers on: pytorch model trainer, pytorch model trainer Part of the ML Training skill category. Use when working with pytorch model trainer functionality. Trigger with phrases like "pytorch model trainer", "pytorch trainer", "pytorch".

by dicklesworthstone

Links this skill to your agents and keeps it current with the author’s updates.

Automatic

AI Engineer pack

1 · One-time — install the CLI (docs ↗)

pip install decimalai

2 · Get the skill — writes .claude/skills/dicklesworthstone-pytorch-model-trainer/SKILL.md

decimalai skills pull dicklesworthstone-pytorch-model-trainer --out .claude/skills/

No CLI? Get the same file — curl · raw file

curl -fsSL --create-dirs -o .claude/skills/dicklesworthstone-pytorch-model-trainer/SKILL.md https://app.decimal.ai/s/dicklesworthstone-pytorch-model-trainer/SKILL.md

Raw file: /s/dicklesworthstone-pytorch-model-trainer/SKILL.md ↗ · machine-readable summary at /s/dicklesworthstone-pytorch-model-trainer.json ↗

In productionGlobal

how this skill behaves on real traffic · last 30 days

No production data for this skill yet — usage numbers appear here once agents start running it.

Anonymized aggregate across every org using this skill — proof from real production use, not a demo.

Does this skill help? · Benchmark: with vs. withoutsign in to run this on your agents

22 test cases from the publisher’s eval.yaml · each run twice on gemini-3.6-flash — with the skill injected, then without · last run 23d ago

Yes — +5% lift · at higher cost+42% tokens0% turns100% of whole cases pass with skill (22/22)

Test case	Without → With	Effect	Δ tokens	Δ turns
case-11	✗→✓	▲ Improved	192%	0%
case-01	✓→✓	＝ Same ✓	62%	0%
case-02	✓→✓	＝ Same ✓	52%	0%
case-03	✓→✓	＝ Same ✓	44%	0%
case-04	✓→✓	＝ Same ✓	29%	0%

Showing 5 of 22 cases

SkillSafety

✓ 3 of 3 passed

Code scan15 of 15 checks clean

0 critical · 0 warning · 0 info — scanner v4 — 8/10/2026

live_secretremote_code_executionreverse_shelldata_exfiltrationssrf_metadataagent_snoopinganti_refusaldestructive_commandsobfuscated_unicodeunicode_homoglyphopaque_blobsuspicious_urltool_overreachtrigger_abuseprompt_injection_phrasing

IntentSafe

LLM judge · does the skill do what it claims

ContentClean

LLM judge · policy & harmful-content review

Source

Imported

dicklesworthstone / pi_agent_rust

tests/ext_conformance/artifacts/plugins-community/skills/07-ml-training/pytorch-model-trainer/SKILL.md

View on GitHub ↗

License MITImported 8/10/2026Version v1

Skill contents

SKILL.md

Pytorch Model Trainer

Overview

This skill provides automated assistance for pytorch model trainer tasks within the ML Training domain.

When to Use

This skill activates automatically when you:

Mention "pytorch model trainer" in your request
Ask about pytorch model trainer patterns or best practices
Need help with machine learning training skills covering data preparation, model training, hyperparameter tuning, and experiment tracking.

Instructions

Provides step-by-step guidance for pytorch model trainer
Follows industry best practices and patterns
Generates production-ready code and configurations
Validates outputs against common standards

Examples

Example: Basic Usage Request: "Help me with pytorch model trainer" Result: Provides step-by-step guidance and generates appropriate configurations

Prerequisites

Relevant development environment configured
Access to necessary tools and services
Basic understanding of ml training concepts

Output

Generated configurations and code
Best practice recommendations
Validation results

Error Handling

| Error | Cause | Solution | |-------|-------|----------| | Configuration invalid | Missing required fields | Check documentation for required parameters | | Tool not found | Dependency not installed | Install required tools per prerequisites | | Permission denied | Insufficient access | Verify credentials and permissions |

Resources

Official documentation for related tools
Best practices guides
Community examples and tutorials

Related Skills

Part of the ML Training skill category. Tags: ml, training, pytorch, tensorflow, sklearn

Illustrative Want to see what dicklesworthstone/pytorch-model-trainer would put in front of your model? Preview the rendered body and check its trigger phrases — no LLM call, no signup.

Latest benchmark · publisher runVerified

8/30/2026, 02:51 UTC · 22 cases

Lift

+5%

100% vs 95%

Agent turns

1.0 vs 1.0

Tokens

+42%

2,180.73 vs 1,534.91

fail → pass · 1pass → pass · 21

Per-case results (22)

Case	Status	Duration (ms)			Turns			Tokens			Tool calls
Case	Status	Without	With	Δ	Without	With	Δ	Without	With	Δ	Without	With	Δ
case-01 I am writing a PyTorch training loop for an image classifier. Should I place `optimizer.zero_grad()` before `loss.backward()` or after `optimizer.step()`?	pass→pass	12,803	15,398	+20%	1	1	0%	1,352	2,189	+62%	0	0	—
case-02 When running the evaluation loop over the validation set, should I leave the PyTorch model in default training state or call `model.eval()`?	pass→pass	8,248	14,727	+79%	1	1	0%	1,410	2,144	+52%	0	0	—
case-03 I want to save memory and compute during validation in PyTorch. Should I execute evaluation inside a `with torch.no_grad():` context block?	pass→pass	13,974	14,972	+7%	1	1	0%	1,540	2,223	+44%	0	0	—
case-04 I need to save my trained PyTorch model checkpoint to disk. Is it better practice to serialize `model.state_dict()` or serialize the entire model object with `torch.save(model, path)`?	pass→pass	15,488	17,803	+15%	1	1	0%	2,023	2,618	+29%	0	0	—
case-05 In PyTorch, should `scheduler.step()` be called before `optimizer.step()` or after `optimizer.step()` inside the training epoch loop?	pass→pass	11,789	12,867	+9%	1	1	0%	1,293	1,987	+54%	0	0	—
case-06 To prevent exploding gradients in PyTorch, I want to use `torch.nn.utils.clip_grad_norm_`. Should gradient clipping occur before or after calling `optimizer.step()`?	pass→pass	6,468	13,576	+110%	1	1	0%	1,275	2,089	+64%	0	0	—
case-07 When transferring PyTorch DataLoader batches to CUDA GPUs, what DataLoader argument should be set to true to enable faster host-to-device memory copy?	pass→pass	4,932	5,725	+16%	1	1	0%	639	1,239	+94%	0	0	—
case-08 When training with `torch.cuda.amp.autocast()`, in what order should `scaler.step(optimizer)` and `scaler.update()` be invoked?	pass→pass	12,456	10,429	-16%	1	1	0%	1,347	1,943	+44%	0	0	—
case-09 I am training a Transformer architecture in PyTorch and want to apply weight decay properly. Should I use standard Adam or AdamW?	pass→pass	10,870	16,189	+49%	1	1	0%	2,011	2,481	+23%	0	0	—
case-10 When tracking total training loss across an epoch in PyTorch, should I accumulate `total_loss += loss` or `total_loss += loss.item()`?	pass→pass	8,212	14,018	+71%	1	1	0%	1,539	1,935	+26%	0	0	—
case-11 After completing a validation loop in PyTorch where `model.eval()` was called, what command must be called before resuming training?	fail→pass	7,556	8,518	+13%	1	1	0%	333	974	+192%	0	0	—
case-12 When using PyTorch `nn.CrossEntropyLoss`, should raw unnormalized logits or softmax probabilities be passed as predictions?	pass→pass	6,702	11,060	+65%	1	1	0%	1,299	1,489	+15%	0	0	—
case-13 My PyTorch training loop on GPU is bottlenecked by CPU data loading speed. What DataLoader parameter should be increased from its default value of 0?	pass→pass	5,373	10,416	+94%	1	1	0%	938	1,335	+42%	0	0	—
case-14 How should PyTorch tensors and models be assigned to GPU or CPU dynamically based on CUDA availability?	pass→pass	16,687	12,118	-27%	1	1	0%	1,911	2,884	+51%	0	0	—
case-15 I am performing gradient accumulation over 4 micro-batches in PyTorch. How often should `optimizer.zero_grad()` be executed?	pass→pass	12,606	9,633	-24%	1	1	0%	1,509	2,149	+42%	0	0	—
case-16 When accumulating gradients over 4 micro-batches in PyTorch, how should the batch loss be scaled before calling `backward()`?	pass→pass	12,938	13,392	+4%	1	1	0%	1,606	1,999	+24%	0	0	—
case-17 To make a PyTorch training run reproducible across executions, what core function sets the global random seed for CPU and GPU tensor operations?	pass→pass	5,519	6,600	+20%	1	1	0%	1,022	1,637	+60%	0	0	—
case-18 When resuming a PyTorch training run from a checkpoint, what state besides `model.state_dict()` must be restored to preserve optimization momentum?	pass→pass	13,662	17,217	+26%	1	1	0%	1,612	2,725	+69%	0	0	—
case-19 What standard Python dunder method must be implemented to index sample items when subclassing `torch.utils.data.Dataset`?	pass→pass	3,274	8,464	+159%	1	1	0%	646	976	+51%	0	0	—
case-20 How do I export an existing PyTorch model to ONNX format using `torch.onnx.export` for optimized production deployment serving?	pass→pass	19,779	21,662	+10%	1	1	0%	2,941	3,951	+34%	0	0	—
case-21 I need to set up a training loop using Keras and TensorFlow 2.x `tf.GradientTape`. How do I compute gradients and apply them with `optimizer.apply_gradients`?	pass→pass	16,286	19,674	+21%	1	1	0%	2,467	3,623	+47%	0	0	—
case-22 How do I use `torch.jit.trace` to serialize a trained model for C++ inference deployment?	pass→pass	20,106	19,605	-2%	1	1	0%	3,055	3,386	+11%	0	0	—

These cases were also screened by the skill’s intent / safety pass.See the Trust review ↗Reproduce this benchmark in Colab ↗

How this skill was measured

DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +5 percentage points is the difference between those two pass rates over the 22 comparable cases.

Model measured: gemini-3.6-flash
Graded by: gemini-3.6-flash
Grading method: judged
Verification: verified
Pass rate without the skill: 95.5%
Pass rate with the skill: 100%
Cases attempted: 22
Cases counted toward lift: 22
Cases that regressed: 0
Skill version measured: v1
Run completed: 8/30/2026, 02:51 UTC

Measured example — same prompt, with vs. without the skill · from the benchmark run · case case-11

Prompt

After completing a validation loop in PyTorch where `model.eval()` was called, what command must be called before resuming training?

Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.

Show both answers

SkillScore blends 4 signals1 of 4 complete

BenchmarkPass rate WITH skill on a verified run100

Live evalNeeds ≥10 evaluated traces in 30 days

AI ratingNeeds ≥10 rated traces

AdoptionNeeds ≥10 serves across ≥2 orgs in 30 days

Efficiency · separate axis (at equal correctness)+0% turns+41% tokens

Shown for transparency; not credited unless a verified run shows no correctness regression.

Backed by 1 of 4 signals — each added signal makes the score more robust.

Verified on gemini-3.6-flash · re-tested 22 days ago

Other measured skills in the registry, with their headline benchmark lift.