ALCHEMI Toolkit is a developer toolkit for accelerating training and inference for AI in chemistry and material science.
ALCHEMI Toolkit is a Python framework that uses GPUs to run atomic simulations and train machine-learned models for chemistry and materials science. Researchers and developers use it for molecular dynamics, geometry relaxation, and model training across one or multiple GPUs. Its catalogue add-ons guide coding agents in using the toolkit’s APIs and repository conventions.
This repository also configures its own agents. See what nvalchemi-toolkit tells them →
Skill Claude Code ✓ vendor
How to write, read, compose, and load atomic data using nvalchemi's composable Zarr-backed storage pipeline (Writer, Reader, Dataset, MultiDataset, DataLoader). Use when saving simulation outputs or trajectories to disk, converting structures (e.g. ASE / extxyz) into Zarr stores, assembling datasets for training or…
Skill Claude Code ✓ vendor
How to use AtomicData and Batch, the core graph-based data structures for representing atomic systems and batching them for GPU computation. Use when building systems from positions, cells, and atomic numbers, converting from ASE Atoms, batching or unbatching structures, reading per-atom vs per-graph tensors, grouping…
Skill Claude Code ✓ vendor
How to run domain-decomposed (multi-GPU) MLIP simulations with DomainParallel — choose between the halo and graph-partition strategies, author a distributionspec so a bring-your-own model runs under domain decomposition, and write a custom dynamics integrator that stays correct across ranks.
Skill Claude Code ✓ vendor
How to configure and run dynamics simulations, compose multi-stage pipelines (FusedStage, DistributedPipeline), use inflight batching, and manage data sinks. Use when writing any simulation script — molecular dynamics (NVE/NVT), structure relaxation or geometry optimization (e.g. FIRE2), equation-of-state or…
Skill Claude Code ✓ vendor
How to use and write dynamics hooks — callbacks that observe or modify batch state at specific points during each simulation step. Use when a simulation needs neighbor-list rebuilds, convergence checks or early stopping, temperature control, per-step logging or trajectory capture, or any custom per-step behavior…
Skill Claude Code ✓ vendor
How to implement a dynamics integrator by subclassing BaseDynamics and overriding preupdate() and postupdate() methods. Use when creating a custom integrator, optimizer, or sampler that the built-in stages do not provide; for configuring existing dynamics, see nvalchemi-dynamics-api.
Skill Claude Code ✓ vendor
How to fine-tune nvalchemi-compatible models with FineTuningStrategy, pretrained checkpoint initialization, module patches, trainable-parameter filters, conservative optimizer defaults, validation, restart checkpoints, and model-agnostic MACE, AIMNet2, custom BaseModelMixin, or PyTorch inputs. Use when adapting a…
Skill Claude Code ✓ vendor
How to use built-in loss functions and implement custom losses using the BaseLossFunction template-method pattern — residual types, per-atom normalization, masking, and graph-balanced reductions. Use when choosing or weighting energy, force, or stress objectives for training or fine-tuning, masking atoms or graphs out…
Skill Claude Code ✓ vendor
How to wrap an arbitrary MLIP (Machine Learning Interatomic Potential) using the BaseModelMixin interface to standardize inputs, outputs, and embeddings. Use when integrating a model such as MACE or AIMNet2 (e.g. MACEWrapper, loading pretrained checkpoints) so dynamics, training, or fine-tuning stages can call it, or…
Skill Claude Code ✓ vendor
How to add observability to nvalchemi dynamics and training workflows using ReportingOrchestrator, RichReporter, TensorBoardReporter, scalar extraction, custom reporter callbacks, and dynamics LoggingHook. Use when showing live progress, writing TensorBoard summaries, preserving dynamics CSV rows, adding rank-safe…
Skill Claude Code ✓ vendor
How to configure nvalchemi training workflows with TrainingStrategy, custom training functions, standalone or composed losses, loss-weight schedules, optimizer and scheduler configs, validation, hooks, restartable checkpoints, model-agnostic inputs, and scaling to multiple GPUs or nodes with DistributedManager and…
Skill Claude Code ✓ vendor
Performance tuning for nvalchemi's Zarr-backed Reader, Dataset, and DataLoader pipeline. Use when configuring AtomicDataZarrReader, Dataset, DataLoader, ZarrWriteConfig, or nvalchemi-io-test for training/inference throughput, especially shuffled access, graph-like random access, fused prefetch, pinned memory…
Skill Claude Code ✓ vendor
How to compute minimum-energy paths (MEPs) and transition-state estimates with nvalchemi.dynamics.mep: building initial paths (interpolation, alignment, IDPP) and running batched nudged elastic band (NEB) with the NEB strategy or by building it manually with hooks. Use when setting up reaction paths between reactant…
Skill Claude Code ✓ vendor
How to distill a large teacher MLIP into a small student with DistillationStrategy — teacher signals and offline dataset labeling, the teacher loss targets, the on-policy segment loop (propagator, replay buffer, mixed loader), checkpoints and restart, and accuracy/stability evaluation with acceptance thresholds. Use…
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: