Domains
Forms
Mechanism
Breaks down how the skill works and what it produces, so you can quickly judge whether it fits your scenario and value.
When adapting a VLM to a visual domain or task, configuring frozen-vision-tower LoRA, or debugging a VLM fine-tune that trains without learning.
A validated adapter config (frozen components, LoRA targets, pixel budget) that prevents silent training failures and is directly consumed to generate a runnable VLM SFT script.
Use Cases
A team adapts a general VLM to read medical scans. They follow the skill's frozen-tower default, then escalate to unfreeze last-6 ViT layers with 5–10x lower vision LR after plateau. The pixel-budget checklist prevents downsampling detail. Result: a domain-adapted VLM with validated config, avoiding silent training failures.
An engineer sees normal loss but flat eval after VLM SFT. Using the skill, they run the collator checklist in references and find image-tag/count mismatch — the silent killer. Fixing placeholder-to-media mapping restores learning. The skill saves hours of hyperparameter guessing by pinpointing alignment bugs.
Skill Relationships
Dependency relationships read as "the upper tier points to the lower tier." The current Skill sits in the middle tier — above are Skills that depend on it, below are Skills it depends on.
Tier 1 · These Skills Use Me
Tier 2 · Current Skill
Tier 3 · I Use These Skills
Browse this skill's relationships within its skillset. Click a node to switch the side panel; use the search box to jump to any skill.
Skill File
