tao-train-metric-learning-recognition

Metric-learning recognition (ml-recog) for fine-grained visual recognition. Learns embeddings for retrieval-based matching (e.g., retail product recognition) using triplet / contrastive losses. Use when training, evaluating, exporting, or running inference for a TAO metric-learning recognition model. Trigger phrases include "train metric learning", "ml-recog", "retrieval embeddings", "triplet loss recognition", "fine-grained matching".

Quality

83%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Low

Low-risk findings worth noting

Quality

Content

66%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 body is a dense, well-organized operational reference with concrete spec overrides, parameter tables, and clearly signaled bundle references. It leans verbose in a few policy paragraphs and lacks explicit validation checkpoints for batch/train operations.

Suggestions

Tighten the AutoML Train Action Policy and Dataclass Schemas paragraphs into shorter rules or move the packaging/regeneration mechanics into a reference file to improve conciseness.

Add explicit validation/verification steps (e.g., confirm checkpoint exists, verify dataset paths resolve, validate spec parses) around the train/export/batch workflows to lift workflow clarity.

Replace placeholder strings like "<selected train/AutoML checkpoint>" with a concrete instruction for how the agent should resolve the checkpoint path, or reference the Spec Param / Parent Model Inference section inline.

Dimension	Reasoning	Score
Conciseness	Mostly efficient and dense with actionable specifics, but lengthy prose blocks (e.g., the AutoML Train Action Policy paragraph and Dataclass Schemas paragraph) restate policy and packaging mechanics that could be tightened or pushed to references.	3 / 5
Actionability	Concrete spec_override code blocks with real dataset paths and parameter tables give mostly executable guidance, though checkpoint placeholders like "<selected train/AutoML checkpoint>" and "<PTM when no resume checkpoint>" leave minor gaps.	4 / 5
Workflow Clarity	Sequencing is clear across train/evaluate/inference/deploy with explicit policy routing and dataset requirements, but validation/verification checkpoints for the batch and destructive train/export actions are implicit rather than stated, capping just below 5.	4 / 5
Progressive Disclosure	Good structure with one-level-deep references to real bundle files (tao-deploy-.md, spec_template_.yaml, skill_info.yaml) clearly signaled in the body; minor gaps such as the data-source table and inference-mapping table being inlined rather than referenced keep it just below 5.	4 / 5
	Total	15 / 20 Passed

Description

100%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.

A strong, well-formed description: it states concrete capabilities, an explicit Use-when clause, and a set of natural trigger phrases in third person. It is concise yet comprehensive with minimal overlap risk.

Dimension	Reasoning	Score
Specificity	Names the domain and multiple concrete actions — "Learns embeddings for retrieval-based matching" plus "training, evaluating, exporting, or running inference" — giving comprehensive coverage of what the skill does.	5 / 5
Completeness	Clearly answers "what" (metric-learning recognition, embeddings, triplet/contrastive losses) and "when" with an explicit "Use when..." clause plus concrete trigger phrases.	5 / 5
Trigger Term Quality	Explicit trigger phrases like "train metric learning", "ml-recog", "retrieval embeddings", "triplet loss recognition", "fine-grained matching" are natural synonyms users would actually say.	5 / 5
Distinctiveness Conflict Risk	Narrow niche (TAO metric-learning recognition / ml-recog) with distinctive triggers unlikely to fire for unrelated skills; minimal conflict risk.	5 / 5
	Total	20 / 20 Passed

Validation

87%

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 — 14 / 16 Passed

Validation for skill structure

Criteria	Description	Result
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning
referenced_paths_exist	Referenced path issues: 2 missing	Warning

	Total	14 / 16 Passed

Repository: NVIDIA/skills
Path: skills/tao-train-metric-learning-recognition/SKILL.md
Commit: be3d1a4

Reviewed: 4 minutes 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.