distributed-training

Multi-GPU and distributed training patterns with PyTorch DDP. Use when scaling training across GPUs.

Quality

76%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Passed

No findings from the security scan

Fix and improve this skill with Tessl

tessl review fix ./researchclaw/skills/builtin/tooling/distributed-training/SKILL.md

Quality

Content

80%Weight 40%Scale 1-5

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

The content is admirably lean and well-structured for a short best-practices skill, with concrete API fragments on several items. It is held back by a lack of validation checkpoints and by directive-only steps that lack executable code.

Suggestions

Add a brief runnable DDP initialization snippet so the guidance is copy-paste ready rather than fragmentary.

Include a validation/checkpoint step (e.g., verify all ranks reach the barrier or validate checkpoint integrity on rank 0) to introduce a feedback loop into the workflow.

For directive-only items like 'Scale learning rate linearly with world size', add the concrete formula (e.g., lr = base_lr * world_size) to improve actionability.

Dimension	Reasoning	Score
Conciseness	The body is a lean seven-item list with no padding and no explanation of concepts Claude already knows; every token earns its place.	5 / 5
Actionability	Several items give concrete API fragments (dist.init_process_group(backend='nccl'), nn.SyncBatchNorm.convert_sync_batchnorm()), but others are directive-only ('Scale learning rate linearly', 'Use gradient accumulation') and no fully runnable example is provided.	4 / 5
Workflow Clarity	The items are numbered best practices rather than a sequenced process, and there are no validation or verification checkpoints for the batch/checkpoint operations, matching the anchor with sequence present but checkpoints missing.	3 / 5
Progressive Disclosure	Under 50 lines with no external references needed, the single well-organized section with a clear heading and numbered list satisfies the simple-skill exception for progressive disclosure.	5 / 5
	Total	17 / 20 Passed

Description

73%Weight 40%Scale 1-5

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

The description is concise and third-person with an explicit 'Use when' trigger, covering both what and when clearly. Its main weakness is specificity, as it relies on the generic term 'patterns' rather than enumerating concrete capabilities.

Dimension	Reasoning	Score
Specificity	It names the domain and a concrete framework ('Multi-GPU and distributed training patterns with PyTorch DDP') but 'patterns' is generic and no specific actions are enumerated, matching the anchor that lists domain plus 1-2 concrete actions without comprehensive coverage.	3 / 5
Completeness	It states both what ('Multi-GPU and distributed training patterns with PyTorch DDP') and when ('Use when scaling training across GPUs'), matching the anchor where both are present but the 'when' could be slightly more specific.	4 / 5
Trigger Term Quality	It includes natural terms a user would say ('multi-GPU', 'distributed training', 'DDP', 'scaling training across GPUs'), giving good keyword coverage with only minor synonyms missing.	4 / 5
Distinctiveness Conflict Risk	It targets a clear niche (distributed multi-GPU training) with distinct triggers (DDP, multi-GPU, scaling) and minimal overlap risk with other skills.	5 / 5
	Total	16 / 20 Passed

Validation

100%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 16 / 16 Passed

Validation for skill structure

No warnings or errors.

Repository: aiming-lab/AutoResearchClaw
Path: researchclaw/skills/builtin/tooling/distributed-training/SKILL.md
Commit: be4ba47

Reviewed: 5 days ago

Table of Contents

Discovery Implementation Validation

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.