Results for “checkpoints”

Turn a repeated manual task into a reusable skill you can call by name or schedule. It separates changing inputs from fixed steps, adds approval checkpoints, and tests the workflow against a known result before registration and team sharing.

• Turning a weekly routine into a reusable skill
• Capturing a workflow demonstrated in conversation
• Adding approval checkpoints to business tasks

Check that changes to existing PEFT methods still work with saved training checkpoints. The skill guides compatibility tests, configuration checks, benchmark comparisons, and reporting, including comparisons of model outputs before and after a change.

• Checking compatibility after a forward method rewrite
• Changing state_dict structure while supporting old checkpoints
• Reviewing defaults for new configuration fields

Coordinate feature work, bug fixes, refactoring, and MVP development through shared phases scaled to the change. Related operation skills use this workflow to delegate research, planning, test-first implementation, review, and commits with human approval checkpoints.

• Using a shared process for feature additions and bug fixes
• Scaling development steps to the size of a change
• Organizing MVP development from a design document

Save an agent session’s state, settings, memory, and task progress, then restore it in another conversation. List saved sessions, inspect their details, and remove old ones when managing checkpoints for ongoing work.

• Pause work and resume it in a later conversation
• Save progress checkpoints during long tasks
• Inspect saved sessions and their details

Build and check clinical machine learning workflows with PyHealth, from loading electronic health records, signals, or images to training and evaluation. Includes patient-level data splits, prediction task checks, model checkpoints, and medical code mapping.

• Prototype mortality and readmission prediction models
• Check patient splits and data leakage
• Train and evaluate sleep staging models

Organize PyTorch models and data pipelines with PyTorch Lightning, automate training, save checkpoints, and track experiments. Includes templates and guidance for configuring training across multiple GPUs or TPUs.

• Restructure PyTorch training code into LightningModules
• Build reusable data preparation and loading pipelines
• Configure checkpoints, early stopping, and experiment logs

Coordinate multiple skills to deliver a SaaS prototype, audit web security, build an AI agent, test browser behavior, or design a domain model. Verify each step before continuing and report completed work with supporting evidence.

• Delivering a SaaS minimum viable product
• Running a web application security audit
• Building an AI agent system with checkpoints

Follow Conductor's development plans using test-driven development: write a failing test, implement the change, then improve the code. Manage phase verification, progress updates, and focused git commits as tasks move toward completion.

• Implementing tasks from plan.md
• Following the test-driven development cycle
• Verifying completion at phase checkpoints

Ask a more capable agent for a concrete execution plan for complex work. Request ordered steps, dependencies, checkpoints, and verification methods, then have the requesting agent assess the plan and carry out the work itself.

• Break complex work into actionable steps
• Plan changes across multiple modules
• Define checkpoints and completion criteria

Configure large language model pretraining with torchtitan across multiple GPUs. Covers ways to split training work, launch jobs on computing clusters, enable Float8 numerical processing, and save or resume training through distributed checkpoints.

• Pretraining Llama 3.1 8B on eight GPUs
• Configuring multi-node training with SLURM
• Enabling Float8 training on H100 GPUs

Guides setup and operation of Lambda Labs GPU instances for machine learning training and inference. Covers SSH access, persistent storage, API automation, and multi-GPU or multi-node training, including how to retain checkpoints and terminate idle instances.

• Setting up a GPU instance for LLM fine-tuning
• Keeping datasets and checkpoints between sessions
• Automating instance launch and termination through the API

Build AI agents with LangGraph by connecting processing steps into a graph. Define shared state, conditional paths, tool calls, saved checkpoints, and human review points to create workflows that retain information as they run.

• Building an agent that calls tools in a loop
• Sharing accumulated findings between multiple agents
• Routing requests to different processing paths

Adapt machine learning training to run across multiple GPUs or machines with Ray Train. Covers changes to existing code, comparing training settings with Ray Tune, and saving checkpoints so training can resume after failures.

• Adapting existing PyTorch code for multiple GPUs
• Training HuggingFace models across multiple machines
• Comparing learning rates and batch sizes with Ray Tune

Manage shell commands and AI agent tasks through a shared gbrain job queue. Track progress, steer agents, pause or resume work, and coordinate parallel tasks, with guidance for long operations, completion reporting, and reusable stage checkpoints.

• Run scripts as monitored background jobs
• Coordinate parallel research by AI agents
• Redirect, pause, or resume an agent task

Guides adapting OpenPI robot models to custom datasets and serving them to produce actions from observations. Covers JAX and PyTorch training, checkpoint conversion, and troubleshooting for normalization statistics and GPU memory.

• Fine-tune OpenPI models on custom robot datasets
• Convert JAX checkpoints to PyTorch
• Serve policies for ALOHA, DROID, or LIBERO

Guides setup and operation of Lambda Labs GPU instances for machine learning training and inference. Covers SSH access, persistent storage, API automation, and multi-GPU or multi-node training, including how to retain checkpoints and terminate idle instances.

• Setting up a GPU instance for LLM fine-tuning
• Keeping datasets and checkpoints between sessions
• Automating instance launch and termination through the API

Configure large language model pretraining with torchtitan across multiple GPUs. Covers ways to split training work, launch jobs on computing clusters, enable Float8 numerical processing, and save or resume training through distributed checkpoints.

• Pretraining Llama 3.1 8B on eight GPUs
• Configuring multi-node training with SLURM
• Enabling Float8 training on H100 GPUs

Manage Hugging Face Hub through the hf command line tool. Download and upload models or datasets, organize cloud storage, manage authentication and local caches, run scheduled jobs, and deploy or maintain model inference services.

• Download or publish models and datasets
• Sync training checkpoints to cloud storage
• Inspect and clean the local cache

Execute a development phase by grouping plans into waves, sets of tasks that can run together based on dependencies. Collect results, verify completion, and update progress, with options to run a selected wave or work sequentially with checkpoints.

• Execute all plans in a development phase
• Run one wave to manage usage limits
• Execute only plans that address verification gaps

Design and run bounded procedures in environments where current conditions are uncertain. The skill separates agent judgment from precise tool actions, verifies each effect, and records checkpoints so interrupted work can resume from observed evidence.

• Designing procedures for environments with varying conditions
• Resuming interrupted operations from verified evidence
• Deciding whether an agent or ordinary code should control a procedure

Review and improve SKILL.md files using nine evaluation criteria and test prompts. Independent agents compare original and revised versions, while user checkpoints, version history, and result cards help track which changes are retained or rolled back.

• Assess skill quality without editing files
• Improve instructions in a specific SKILL.md
• Prioritize weaker skills for revision

Turn articles, notes, or code history into talk preparation materials, including angles, titles, scripts, slide drafts, and revision sheets. Coordinate the stages with user checkpoints, or run an individual stage again.

• Prepare a talk from articles or notes
• Build an experience report supported by Git history
• Develop a script after choosing an angle and title

Break large features into independently verifiable, human-reviewable slices under a feature directory in specs/. Use for risky or multi-step feature work that needs upfront questioning, API seams, browser-playable checkpoints, HTML visualizations, screenshot gates, staged implementation plans, recursive fog-of-war reslicing, or proactive research into reference implementations/best practices before slicing. Pairs with your project's verification harness and screenshot gates (the browser checkpoints), [refactor-clean](../refactor-clean/SKILL.md) (review the materialized spec so the plan describes one-owner architecture, not the feature bolted on), [screenshot-critique](../../visual/screenshot-critique/SKILL.md) and [compare-screenshots](../../visual/compare-screenshots/SKILL.md) (the visual gates), and a code-review pass (audit each slice before it lands).

The English summary is being prepared. Showing the original description.

Expert blueprint for player mortality loops: progress-index checkpoints, soul graves, ghost/spirit layers, I-frame restitution, world-progress persistence, and death analytics. Use when implementing respawn, shrine checkpoints, corpse-run retrieval, or second-chance mechanics. Keywords revival, respawn, checkpoint, soul grave, ghost mode, I-frames, death analytics, progress index.

The English summary is being prepared. Showing the original description.