Guide post-training workflows for SFT, alignment, and evaluation.
npx skills add https://github.com/zzy1127/PostTrainAgent --skill posttrain
Please help me install this Agent Skill. Skill: posttrain Source: https://github.com/zzy1127/PostTrainAgent/tree/main/skills/posttrain Command: npx skills add https://github.com/zzy1127/PostTrainAgent --skill posttrain
Post-training engineering guidelines to systematically improve an AI model's capabilities, focusing on robust methodologies and repeatable processes rather than hacks.
Audit current post-training workflow and establish a repeatable evaluation plan for capability improvement.
High-intent search queries and answers about installing and using this skill.
Establish a repeatable post-training workflow by loading a structured context framework for agents that defines data quality checks, compute efficiency, and rigorous evaluation plans to systematically improve model capabilities.
Frameworks for SFT and alignment evaluation provide structured guidelines, guardrails, and process rules to ensure reproducible experiments and safe deployments across ML engineering teams during capability generalization.
Optimize data quality checks for post-training by applying a data strategy framework that enforces robust validation rules, improving generalization and ensuring high-quality inputs for SFT and alignment pipelines.
Post-training engineering requires compute pipeline configurations designed for hardware efficiency, providing scalable guidance for SFT, alignment, and evaluation workflows to ensure robust and repeatable model improvements.
Ensure reproducible post-training experiments by applying structured frameworks with clear guardrails and process guidelines, establishing disciplined evaluation plans that prevent undocumented hacks and ensure safe deployments.
Apply post-training guardrails when systematic capability improvement is needed without altering base architecture, using disciplined SFT and alignment workflows to ensure reproducible experiments and safe deployments across teams.
aloneggAutomate end-to-end ML analysis from raw data to evaluated models.
timurgaleevMonitor post-deploy web apps for defects and performance regressions.
M4n5terStandardize LLM post-training data collection and evaluation across SFT, DPO, and RLHF workflows.
davpatel605-beepConfigure and run reference-free SimPO preference alignment training for LLMs.
