By mlc-ai

Query a captured PithTrain Nsight Systems profile to measure compute/communication overlap, locate exposed comm by DualPipeV stage, and inspect per-rank stream behavior.

mlc-ai/pith-train

Capture a Nsight Systems (.nsys-rep) profile of a short PithTrain run for performance analysis.

mlc-ai/pith-train

Authoritative reference for TIRx-lite authoring on TIRx: TIRx-lite APIs, canonical kernels, optimization guidance, PTX ISA, and GPU references.

mlc-ai/TIRx-harness

Validates that code changes do not break training correctness by comparing loss deltas against a base-vs-base run-to-run envelope.

mlc-ai/pith-train

Register, extend, or audit instructions in the table-driven T.ptx dialect (python/tvm/backend/cuda/ptx/table.py), and move the table to a newer PTX ISA version.

mlc-ai/relax

Measures the throughput difference between two branches with force-balanced routing.

mlc-ai/pith-train

Debug and validate runnable TIRx kernels by selecting and executing the applicable correctness tools.

mlc-ai/TIRx-harness

Profile and optimize runnable TIRx kernels by selecting and executing the applicable measurement and analysis tools.

mlc-ai/TIRx-harness

Set up the minimal set of artifacts (tokenized DCLM corpus shard + released HuggingFace checkpoint converted to DCP) required to benchmark, profile, or regression-test a MoE model in PithTrain.

mlc-ai/pith-train

Adds support for a new MoE language model to PithTrain. An agent skill from mlc-ai/pith-train.

mlc-ai/pith-train

Read, analyze, and manage Weights & Biases (wandb) experiment data for PithTrain runs.

mlc-ai/pith-train

Estimate peak GPU memory for a DualPipeV training run. An agent skill from mlc-ai/pith-train.

mlc-ai/pith-train

Reference for launching jobs inside a SLURM allocation via srun (single-node or multi-node).

mlc-ai/pith-train

Add detailed memory profiling prints throughout the training framework.

mlc-ai/pith-train