RoBERTa-large (mirror)

Byte-identical mirror of FacebookAI/roberta-large at revision 722cf37b1afa9454edce342e7895e588b6ff1d59 (original release: https://github.com/facebookresearch/fairseq) for the Cuvis.AI plugins. No fine-tuning, conversion or re-serialization: the files below are the upstream release as published, with the upstream licence text in LICENSE.

Why this mirror exists

Cuvis.AI provisions model weights once into a shared cache (download-model download <name> from cuvis-ai-core) and then runs its pipelines in an offline, token-free runtime that only reads that cache. Hosting the exact upstream files under the cubert-gmbh organisation makes that provisioning reproducible (commit-pinned and sha256-verified) and takes the per-user Hugging Face account and token out of the users' path.

Files and provenance

File	Upstream source	Upstream revision / id	sha256	Size
model.safetensors	FacebookAI/roberta-large / model.safetensors	722cf37b1afa9454edce342e7895e588b6ff1d59	047c85f0b96269cd62e6f732644f067004eebd95af5b5d35965ae2528f13bf38	1.42 GB
config.json	FacebookAI/roberta-large / config.json	722cf37b1afa9454edce342e7895e588b6ff1d59	82ba49810e6441735e696033ac2512ee09da555507b2917ac3865b202d592cc3	0.5 kB
vocab.json	FacebookAI/roberta-large / vocab.json	722cf37b1afa9454edce342e7895e588b6ff1d59	9e7f63c2d15d666b52e21d250d2e513b87c9b713cfa6987a82ed89e5e6e50655	898.8 kB
merges.txt	FacebookAI/roberta-large / merges.txt	722cf37b1afa9454edce342e7895e588b6ff1d59	1ce1664773c50f3e0cc8842619a93edc4624525b728b188a9e0be33b7726adc5	456.3 kB
tokenizer.json	FacebookAI/roberta-large / tokenizer.json	722cf37b1afa9454edce342e7895e588b6ff1d59	847bbeab6174d66a88898f729d52fa8d355fafe1bea101cf960dd404581df70e	1.4 MB
tokenizer_config.json	FacebookAI/roberta-large / tokenizer_config.json	722cf37b1afa9454edce342e7895e588b6ff1d59	994f46754c5bf4014f1aa92d34b1374319c3a6b3f702105cd5b742beaecd18ce	0.0 kB
LICENSE	https://raw.githubusercontent.com/facebookresearch/fairseq/main/LICENSE	n/a (URL; sha256 pinned)	52412d7bc7ce4157ea628bbaacb8829e0a9cb3c58f57f99176126bc8cf2bfc85	1.1 kB

Mirrored on 2026-10-09 by Cubert GmbH from the sources above. The sha256 values are the upstream values; tools/mirror_weights.py check in cuvis-ai-core re-verifies this repository against them and against the upstream licence text.

Licence

MIT License. The LICENSE file is the upstream licence text taken verbatim from https://github.com/facebookresearch/fairseq (where RoBERTa was released; the Hugging Face repository declares license: mit and ships no LICENSE file) at the mirror date. Copyright remains with the upstream authors; Cubert GmbH claims no rights in these files.

Usage with Cuvis.AI

uv run download-model download roberta_large

download-model (from cuvis-ai-core) provisions the file into the shared Hugging Face cache and verifies its sha256. From cuvis-ai-core 0.16.0 on, ModelWeights.resolve("<name>") returns the cached path and the cuvis-ai-steervit nodes resolve their weights through it, so a pipeline needs no manual step once the weights are provisioned; earlier plugin releases still fetch from their original upstream sources.