skills.rest
tenstorrenttenstorrentOfficial·4 Agent Skills Included

tt-xla

Run PyTorch and JAX models on Tenstorrent hardware

Compiles and runs PyTorch and JAX models on Tenstorrent AI accelerators through a PJRT backend and TT-MLIR compiler. Diagnoses nightly CI failures, benchmark regressions, graph breaks, dtype mismatches, and missing op fusions without manual log digging. Designs multi-chip sharding strategies and fixes model test failures so hardware bring-up moves faster.
74|32
↗
npx skills add tenstorrent/tt-xla --all -g -y
Available:

Explains the TT-XLA architecture, build system, environment variables, and test commands so the agent can build, test, and debug the PJRT plugin correctly.

All Skills in This Repository (4)

Pure Emerald Level Indicators
📦 In Repo
tenstorrenttenstorrent

excalidraw-diagram

Generate Excalidraw diagram JSON encoding visual arguments for workflows and concepts.

Official
Advanced
📦 In Repo
tenstorrenttenstorrent

analyze-nightly

Summarize GitHub Actions nightly CI failures grouped by ownership area.

Official
Advanced
📦 In Repo
tenstorrenttenstorrent

code-reviewer

Automate structured code-review checks for the tt-xla project.

Official
Advanced
📦 In Repo
tenstorrenttenstorrent

graph-break-analysis

Diagnose graph breaks in TT-XLA compilation logs and generate fix reports.

Official
Advanced

Frequently Asked Questions

FAQPage Schema
How to install tt-xla?▼

Run `npx skills add tenstorrent/tt-xla --all -g -y` in your terminal to install all skills in this suite globally.

What does tt-xla do?▼

It connects PyTorch/XLA and JAX to Tenstorrent AI hardware through a PJRT plugin, compiling models via TT-MLIR for single-chip and multi-chip execution.

How to analyze failing nightly CI runs?▼

Use the analyze-nightly or ci-benchmark-analyzer skills with a GitHub Actions run ID to get a grouped failure report with root causes and performance comparisons.

How to shard a model across multiple Tenstorrent chips?▼

The sharding-model-analysis skill maps your model architecture, picks a CCL-minimizing tensor or sequence parallelism strategy, and implements it with Shardy annotations.

Can it fix model test failures automatically?▼

Yes. Dedicated triage skills diagnose bfloat16 dtype errors, missing training inputs, and forward-output unpacking failures, then apply minimal loader fixes and update test configs.

Related Repositories in Software Engineering

View All in Software Engineering→
openclawopenclaw
388.4k

openclaw

Run a personal AI assistant across your devices and chat apps

Official📦 120 Skills
obraobra
280.0k

superpowers

Gives coding agents a disciplined workflow from idea to merged code

Community📦 15 Skills
facebookfacebook
248.4k

react

AI agent skills for building, testing, and porting React core

Official📦 14 Skills