-
Notifications
You must be signed in to change notification settings - Fork 635
All issues
Issue creation is restricted in this repository
Issues
is:issue state:open
is:issue state:open
Search results
test_triton_mxfp8_dim1_randn[FLOOR-256-128]: exact e8m0 scale mismatch (12/1024) on ROCm MI350
ci-failureAutomated report of CI failure on mainAutomated report of CI failure on mainStatus: Open.#4953 In pytorch/ao;- Status: Open.#4949 In pytorch/ao;
- Status: Open.#4942 In pytorch/ao;
- Status: Open.#4940 In pytorch/ao;
- Status: Open.#4938 In pytorch/ao;
- Status: Open.#4929 In pytorch/ao;
[Bug] AdamW8bit picks a different state precision after sharding the same DTensor parameter
bugSomething isn't workingSomething isn't workingStatus: Open.#4912 In pytorch/ao;pt2e observers silently discard the imaginary part of complex tensors
bugSomething isn't workingSomething isn't workingStatus: Open.#4910 In pytorch/ao;- Status: Open.#4866 In pytorch/ao;
- Status: Open.#4865 In pytorch/ao;
test_dtensor.py (float8 TP compile): aten.abs.default NotImplemented on Float8TrainingTensor after removing allow-in-graph autograd wrappers
ci-failureAutomated report of CI failure on mainAutomated report of CI failure on mainStatus: Open.#4864 In pytorch/ao;- Status: Open.#4848 In pytorch/ao;