mixed-precision-trainer
Mixed Precision Trainer - Auto-activating skill for ML Training. Triggers on: mixed precision trainer, mixed precision trainer Part of the ML Training skill category.
Install with Tessl CLI
npx tessl i github:jeremylongshore/claude-code-plugins-plus-skills --skill mixed-precision-trainerOverall
score
19%
Does it follow best practices?
Validation for skill structure
Activation
7%This description is severely underdeveloped, functioning more as a label than a useful description. It provides no information about what the skill actually does, lists redundant trigger terms, and lacks any guidance on when Claude should select this skill. The description would be nearly useless for skill selection among multiple ML-related skills.
Suggestions
Add specific actions the skill performs, e.g., 'Configures automatic mixed precision (AMP) training, converts model layers to fp16/bf16, manages loss scaling, and optimizes GPU memory usage'
Add a 'Use when...' clause with natural trigger scenarios, e.g., 'Use when the user mentions mixed precision, fp16, half precision, AMP, reducing training memory, or speeding up model training'
Include common user-facing terms and file types, e.g., 'PyTorch AMP, TensorFlow mixed precision, .pt models, training optimization, GPU memory issues'
| Dimension | Reasoning | Score |
|---|---|---|
Specificity | The description only names the skill ('Mixed Precision Trainer') without describing any concrete actions. There are no verbs or specific capabilities listed - it doesn't explain what the skill actually does. | 1 / 3 |
Completeness | The description fails to answer 'what does this do' (no actions described) and 'when should Claude use it' (no explicit use cases or scenarios). The 'Triggers on' line is not a proper 'Use when...' clause with meaningful guidance. | 1 / 3 |
Trigger Term Quality | The triggers listed are redundant ('mixed precision trainer' repeated twice) and represent technical jargon rather than natural user language. Missing common variations like 'fp16', 'half precision', 'amp', 'automatic mixed precision', or 'reduce memory usage'. | 1 / 3 |
Distinctiveness Conflict Risk | While 'mixed precision' is a specific ML concept that provides some distinctiveness, the lack of detail about what this skill does versus other ML Training skills creates potential overlap. The category mention 'ML Training skill category' suggests there may be related skills it could conflict with. | 2 / 3 |
Total | 5 / 12 Passed |
Implementation
0%This skill content is essentially a placeholder template with no actual instructional value. It contains zero concrete guidance on mixed precision training—no code examples, no specific techniques (like torch.cuda.amp, gradient scaling, loss scaling), no workflows, and no actionable information. The content explains what the skill claims to do rather than teaching Claude how to do it.
Suggestions
Add executable code examples showing mixed precision training setup (e.g., using torch.cuda.amp.autocast and GradScaler in PyTorch)
Include a clear workflow with steps: 1) wrap forward pass in autocast, 2) scale loss, 3) backward pass, 4) unscale and step optimizer, with validation checkpoints
Remove all generic boilerplate ('provides automated assistance', 'follows best practices') and replace with specific techniques and common pitfalls
Add concrete examples of when to use FP16 vs BF16, which layers to keep in FP32, and how to debug NaN/Inf issues
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The content is padded with generic boilerplate that provides no actual information. Phrases like 'provides automated assistance' and 'follows industry best practices' are meaningless filler that Claude doesn't need. | 1 / 3 |
Actionability | No concrete code, commands, or specific guidance is provided. The skill describes what it claims to do but never shows how to actually implement mixed precision training. | 1 / 3 |
Workflow Clarity | No workflow, steps, or process is defined. The content only lists vague capabilities without any sequence of actions or validation checkpoints. | 1 / 3 |
Progressive Disclosure | The content is a monolithic block of generic text with no structure pointing to detailed materials, examples, or references. No useful organization for discovery. | 1 / 3 |
Total | 4 / 12 Passed |
Validation
69%Validation — 11 / 16 Passed
Validation for skill structure
| Criteria | Description | Result |
|---|---|---|
description_trigger_hint | Description may be missing an explicit 'when to use' trigger hint (e.g., 'Use when...') | Warning |
allowed_tools_field | 'allowed-tools' contains unusual tool name(s) | Warning |
metadata_version | 'metadata' field is not a dictionary | Warning |
frontmatter_unknown_keys | Unknown frontmatter key(s) found; consider removing or moving to metadata | Warning |
body_steps | No step-by-step structure detected (no ordered list); consider adding a simple workflow | Warning |
Total | 11 / 16 Passed | |
Reviewed
Table of Contents
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.