At diffusers 086bf957 the guide recommends a CPU Generator for reproducibility, and the shipped enable_full_determinism() hard-codes CUBLAS_WORKSPACE_CONFIG=':16:8' while the repository's own test helper switches to ':4096:8' for torch >= 2.10 citing...

At diffusers 086bf957 the guide recommends a CPU Generator for reproducibility, and the shipped enable_full_determinism() hard-codes CUBLAS_WORKSPACE_CONFIG=':16:8' while the repository's own test helper switches to ':4096:8' for torch >= 2.10 citing CUBLAS_STATUS_NOT_INITIALIZED. Curator qualifications: Whether the shipped helper fails on torch >= 2.10 is the producer's unexecuted hypothesis; needs an authorised run. 'More stable across machines' is the guide's claim; no cross-platform bitwise guarantee (T3-02). Batch-dependence and late-call env-var effect (F2, F5) not reinspected. Static reading only; no GPU, model or pipeline run. Limitations: Static reading only; no GPU, model or pipeline run. Evidence status: statements located in the cited sources by a correlated desk review (Claude); not executed; not independently reviewed.

{
  "citations": [
    {
      "url": "https://github.com/huggingface/diffusers/blob/086bf9578c0f4acbc66e48cd1e7cc26befd9e10f/docs/source/en/using-diffusers/reusing_seeds.md",
      "note": "reusing_seeds.md lines 21-31; src/diffusers/utils/torch_utils.py lines 360-380; tests/testing_utils.py lines 1430-1435"
    }
  ]
}