CtrlK
BlogDocsLog inGet started
Tessl Logo

torchtitan

Pretrain LLMs at scale with PyTorch 4D parallelism.

56

Quality

66%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

SecuritybySnyk

High

Do not use without reviewing

Fix and improve this skill with Tessl

tessl review fix ./optional-skills/mlops/torchtitan/SKILL.md
SKILL.md
Quality
Evals
Security

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 highly actionable and well-structured with verified progressive-disclosure references, but the multi-step training workflows lack explicit validation checkpoints for these batch operations, capping workflow clarity at the midpoint.

Suggestions

Add explicit validation checkpoints to each workflow, e.g. after launch verify 'loss is decreasing' / 'all ranks initialized' before marking the launch step complete.

Tighten the inline 8B TOML in Workflow 1 by referencing only the fields being overridden rather than the full config block.

DimensionReasoningScore

Conciseness

The body is efficient with executable snippets and checklists rather than concept explanations; minor instances like 'achieving 65%+ speedups' and the full inline 8B TOML block could be trimmed slightly.

4 / 5

Actionability

Provides copy-paste-ready bash, TOML, and Python across the common cases (8B/70B/405B, multi-node SLURM, Float8, 4D parallelism) with concrete commands and config overrides.

5 / 5

Workflow Clarity

Each workflow has an explicit numbered checklist and clear sequence, but these are batch/distributed training operations with no explicit verify-before-proceeding validation checkpoints, so per the rubric cap workflow clarity is held at 3.

3 / 5

Progressive Disclosure

Clear overview with well-signaled one-level-deep references to real files (fsdp.md, float8.md, checkpoint.md, custom-models.md, all present in references/), with advanced detail appropriately split out.

5 / 5

Total

17

/

20

Passed

Description

53%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 names a clear, distinct niche, but it omits an explicit 'when to use' trigger clause and lacks synonym coverage, capping completeness and trigger-term quality at the midpoint.

Suggestions

Add a 'Use when...' clause naming concrete triggers, e.g. 'Use when pretraining LLMs from scratch, configuring FSDP/TP/PP/CP parallelism, or doing distributed training with PyTorch.'

Broaden trigger terms with natural synonyms like 'distributed training', 'training from scratch', and file/config terms users actually say.

DimensionReasoningScore

Specificity

Names the domain (LLM pretraining) and one concrete action plus method ('Pretrain LLMs at scale with PyTorch 4D parallelism'), but lists only a single composite action rather than multiple specific capabilities.

3 / 5

Completeness

Provides a clear 'what' but no 'Use when...' or equivalent trigger clause; per the rubric, a missing explicit trigger guidance caps completeness at 3.

3 / 5

Trigger Term Quality

Contains natural phrases like 'Pretrain LLMs', 'at scale', and 'PyTorch 4D parallelism', but omits common synonyms users would say such as 'distributed training', 'FSDP', or 'training from scratch'.

3 / 5

Distinctiveness Conflict Risk

The 'pretrain' + '4D parallelism' + 'PyTorch-native' niche is mostly distinct with only minor overlap risk against fine-tuning or generic distributed-training skills.

4 / 5

Total

13

/

20

Passed

Validation

75%

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

Validation12 / 16 Passed

Validation for skill structure

CriteriaDescriptionResult

metadata_version

'metadata.version' is missing

Warning

metadata_field

'metadata' should map string keys to string values

Warning

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

12

/

16

Passed

Repository
NousResearch/hermes-agent
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.