David Corvoysier · Oct 23, 2026
Does Your Model Run on AWS Trainium? Look It Up
Markdown · JSON · Original source
David Corvoysier describes a Space that reports whether selected Transformers models run unmodified on AWS Trainium and provides scripts for users to run them. The Space reports each test’s result, number of Trainium cores, wall including compilation, and peak device memory. Users can search models in the Trending or Downloads views and select versions of the Neuron compiler and Transformers. The tests load models on Trainium2 instances, generate 32 tokens, and check the output. For sharded models, a small judge model checks whether the text is fluent English. Corvoysier says a PASS does not establish throughput or performance on longer generations. The tests cover only Transformers models in selected tasks, and only models that fit on at most 16 Trainium2 cores. The results and run scripts are also available in a public dataset. How do the models that pass perform in throughput tests or longer generations?
Headline and summary by Escape Meridian, prepared by Luna from the attributed source. A source check is not independent confirmation of the source’s claims.
Original source
· Article
Source details
Original source date: Oct 23, 2026 (2026-10-23T00:00:00.759Z).
Date basis: Date stated by the publisher.
Added to Escape Meridian: 2026-11-14T09:01:39.938Z. First observed: 2026-10-23T13:18:04.958Z. Retrieved: 2026-10-25T01:01:14.019Z.
Referenced by
Recorded citations and public file attachments. A reference alone does not establish agreement or influence.
No public references recorded here yet.
Limits
The tests cover only Transformers models and are limited to text-generation and image-text-to-text models; the author says other tasks are planned. The author says a PASS says nothing about throughput or longer generations. The tests are limited to models that fit on 4 Trainium2 devices, or 16 cores at most.
Open question
How do models that pass perform in throughput tests or longer generations?
Source-stated
David Corvoysier says the Hugging Face Space reports whether selected Transformers models run unmodified on AWS Trainium and provides a script to run them.
It tells you whether a model runs unmodified on Trainium, and gives you the script to run it yourself.
Source-stated
The Space displays each test’s result, Trainium core count, wall time including compilation, and peak device memory.
Each row gives the result, the number of Trainium cores the model is sharded over, the wall time of the run (compilation included) and the peak device memory.
Source-stated
Tests generate 32 tokens; for sharded models, a small judge model checks whether the text is fluent English.
The test then generates 32 tokens and checks them. On a single core, the tokens are compared with those of a CPU run of the same checkpoint. When the model is sharded, a small judge model checks that the text is fluent English.
Source-stated
A PASS does not establish throughput or performance on longer generations.
A PASS says nothing about throughput, or about longer generations.
Source-stated
The tests cover Transformers models in a limited set of tasks and models that fit on no more than 16 Trainium2 cores.
At the time of writing, only Transformers models are tested, and our selection is limited to text-generation and image-text-to-text models.
Source and contributor context
David Corvoysier · Article author
These are credits stated by the source; they do not establish a profile match or authorship of every passage.
Source media
Media was declared by the source, but no asset passed the supported availability checks. Open the original source to view it.
Terms and references
Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.
Hugging Face
Hugging Face hosts models, datasets and software that let people share, adapt and inspect AI systems.
TIME
TIME is a news magazine publishing reporting on world affairs, business, technology and the workers behind digital services.
Editorial provenance
Version 1. Publication ID: d8e9b400-d7cb-4420-87e2-51a05b931c11.
Prepared by gpt-6-luna; grounding review by gpt-6-luna.
Source check: uncertain by anthropic/claude-opus-5.5 on 2026-11-14T09:01:39.938Z. This checks fidelity to the source, not whether its claims are independently true.
Editorial selection: accept; reasons: material_development.
Observation: dcab80877e6739f381d17c1be5e648df3a605b8ff54929225739dfc58e85d8b3. Source text hash: c5775eefba3fa9a4e41f08b66992247624f3520d835832dc02e83db54a331ab1.
Revision and source-check history
Version 1 · 2026-11-14T09:01:39.938Z
Does Your Model Run on AWS Trainium? Look It Up
David Corvoysier describes a Space that reports whether selected Transformers models run unmodified on AWS Trainium and provides scripts for users to run them. The Space reports each test’s result, number of Trainium cores, wall including compilation, and peak device memory. Users can search models in the Trending or Downloads views and select versions of the Neuron compiler and Transformers. The tests load models on Trainium2 instances, generate 32 tokens, and check the output. For sharded models, a small judge model checks whether the text is fluent English. Corvoysier says a PASS does not establish throughput or performance on longer generations. The tests cover only Transformers models in selected tasks, and only models that fit on at most 16 Trainium2 cores. The results and run scripts are also available in a public dataset. How do the models that pass perform in throughput tests or longer generations?
Original source
· Article
Source details
Original source date: Oct 23, 2026 (2026-10-23T00:00:00.759Z).
Date basis: Date stated by the publisher.
Added to Escape Meridian: 2026-11-14T09:01:39.938Z. First observed: 2026-10-23T13:18:04.958Z. Retrieved: 2026-10-25T01:01:14.019Z.
Limits
The tests cover only Transformers models and are limited to text-generation and image-text-to-text models; the author says other tasks are planned. The author says a PASS says nothing about throughput or longer generations. The tests are limited to models that fit on 4 Trainium2 devices, or 16 cores at most.
Open question
How do models that pass perform in throughput tests or longer generations?
Source-stated
David Corvoysier says the Hugging Face Space reports whether selected Transformers models run unmodified on AWS Trainium and provides a script to run them.
It tells you whether a model runs unmodified on Trainium, and gives you the script to run it yourself.
Source-stated
The Space displays each test’s result, Trainium core count, wall time including compilation, and peak device memory.
Each row gives the result, the number of Trainium cores the model is sharded over, the wall time of the run (compilation included) and the peak device memory.
Source-stated
Tests generate 32 tokens; for sharded models, a small judge model checks whether the text is fluent English.
The test then generates 32 tokens and checks them. On a single core, the tokens are compared with those of a CPU run of the same checkpoint. When the model is sharded, a small judge model checks that the text is fluent English.
Source-stated
A PASS does not establish throughput or performance on longer generations.
A PASS says nothing about throughput, or about longer generations.
Source-stated
The tests cover Transformers models in a limited set of tasks and models that fit on no more than 16 Trainium2 cores.
At the time of writing, only Transformers models are tested, and our selection is limited to text-generation and image-text-to-text models.
Source and contributor context
David Corvoysier · Article author
These are credits stated by the source; they do not establish a profile match or authorship of every passage.
Source media
Media was declared by the source, but no asset passed the supported availability checks. Open the original source to view it.
Terms and references
Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.
Hugging Face
Hugging Face hosts models, datasets and software that let people share, adapt and inspect AI systems.
TIME
TIME is a news magazine publishing reporting on world affairs, business, technology and the workers behind digital services.
Editorial provenance
Version 1. Publication ID: d8e9b400-d7cb-4420-87e2-51a05b931c11.
Prepared by gpt-6-luna; grounding review by gpt-6-luna.
Source check: uncertain by anthropic/claude-opus-5.5 on 2026-11-14T09:01:39.938Z. This checks fidelity to the source, not whether its claims are independently true.
Editorial selection: accept; reasons: material_development.
Observation: dcab80877e6739f381d17c1be5e648df3a605b8ff54929225739dfc58e85d8b3. Source text hash: c5775eefba3fa9a4e41f08b66992247624f3520d835832dc02e83db54a331ab1.
uncertain · 2026-11-14T09:01:39.938Z · anthropic/claude-opus-5.5
Publication: d8e9b400-d7cb-4420-87e2-51a05b931c11
attribution — pass: The post is correctly attributed to David Corvoysier on the Hugging Face blog. The Space, the tests and the caveats are presented as his statements, not as established facts.
David Corvoysier dacorvo
summary — uncertain: The summary is accurate but hard to follow without specialist background. Central terms go unexplained: AWS Trainium (Amazon's AI chip), Transformers (Hugging Face's model library), Space, sharded/cores and tokens. It also omits the essential context of why this matters: TorchNeuron, the new native PyTorch backend, lets Transformers run its own code on Trainium, so any Hub model may run, and this Space shows which actually do. The smallest repair is one plain-language sentence covering that context, plus tooltips for these terms. The closing question also duplicates the separate question field.
With it, Transformers runs its own modeling code on Trainium, as it does on any other device, so any Transformers model on the Hub may run. Which ones actually do?
date — pass: The originalPublished value of 2026-10-23 matches the post's stated publication date of October 23, 2026.
Published October 23, 2026
support — pass: All claims are grounded in the source: the result columns, 32-token generation, the judge model for sharded runs, the PASS caveat, the task scope, the 16-core limit and the public dataset. The question follows directly from the author's own caveat and does not assume an answer.
At the time of writing, only models that fit on 4 Trainium2 devices are tested, which means 16 cores at most. A PASS says nothing about throughput, or about longer generations.
media — pass: The selected media is the verified direct link to the original blog post, which is the most direct way to reach the source.
Challenge or correct this reading
Send a correction or an opposing source with this reading’s URL. Disagreement and uncertainty can remain unresolved.
Hugging Face hosts models, datasets and software that let people share, adapt and inspect AI systems.
Explore Hugging FaceTIME is a news magazine publishing reporting on world affairs, business, technology and the workers behind digital services.
Explore TIMEHugging Face hosts models, datasets and software that let people share, adapt and inspect AI systems.
Explore Hugging FaceHugging Face hosts models, datasets and software that let people share, adapt and inspect AI systems.
Explore Hugging FaceTIME is a news magazine publishing reporting on world affairs, business, technology and the workers behind digital services.
Explore TIMEHugging Face hosts models, datasets and software that let people share, adapt and inspect AI systems.
Explore Hugging Face