Automatic update of fbcode/onnx to 79a7e0df7e86e0f32e7a05f563b24a566540c18b

[houseroad]

Summary:
Previous import was 807c62cf7e4c96ce49040bcf073b7e4a054f28a5

Included changes:

• 79a7e0df: Fix copy paste error of Min op test case ("torch.utils.data.DataLoader" and ".data.cpu().numpy().copy()" goes wrong with no cuda environment. #2640)
• 4cd2538d: Add a release pipeline for Windows python packages (fix indices for data_parallel and add parameter gradient tests #2632)
• e8b33a5a: Adding UnfoldToDepth op [1.7 Release] (Update torch.tril's documentation to reflect named argument changes #2616)
• c2a8d525: update docs (Conv3D on CPU segfaults #2627)
• 22752354: Generate tests for MeanSquareDistance and SoftmaxCrossEntropyLoss (How to handle exploding/vanishing gradient in Pytorch and negative loss values #2623)
• 602bd622: Add section about external tensor data to IR.md (allow single non-tuple sequence to trigger advanced indexing #2323)
• 165c3f3b: Add integer support to Clip (Fix check_no_size_average tests. #2532)
• a5fabf87: Add operators LessOrEqual and GreaterOrEqual (as functions) (make logsoftmax documenatation readable #2606)
• f1dcdafc: Fix input document of quantized operators (Add dynamic infer shape for nn.Linear #2117)
• 43af9b69: Add reference impl for sequence ops (Remove unnecessary moves in convolution autograd (master) #2380)
• aa50aa12: Print value case of TypeProto more friendly (Add calls to contiguous for cudnn affine_grid #2422)
• 2e67bfc3: Fix issue Bugs in torch.std and torch.var #2436 (Improve Windows Compatibility(for lib/THPP) #2447)
• d27ffc6b: Add support for integer tensors to Min and Max (Missing conda dependency for source install #2608)
• 5cc668af: Update IR.md to describe training extension. (Why does pytorch need the sentences in a minibatch to be sorted? #2615) <G. Ramalingam>
• 8c5bf9d4: Generate node backend tests for celu operator (invoking cuda() takes a long time #2607)
• 7b65287e: Change dtype of dd_da in gradient test to float32 (nn.Embedding error when using max_norm #2620)
• e91739f2: Introduce SoftmaxCrossentropy as a loss function (Cannot backpropagate through DataParallel #2573)
• b008ed3a: Support gathernd with batch_dim mode ([WIP] Add support for colors in NVTX #2585)
• d2fe4f22: Introduce MeanSquaredError as Loss Function (THMapAllocatorContext_ only define once in #if #2570)
• 10b812a6: Add support for default attributes within FunctionExpandHelper (Match NumPy ellipsis semantics for Advanced Indexing (v0.2.0) #2588) <Ewa Tusień>
• 3368834c: adding version update content. (Inconsistent behavior of the max function. #2609)
• 8873cb02: Adding Inverse Op (Make CPU and CUDA share bernoulli yaml declaration #2578)

Test Plan: ci

Reviewed By: hl475

Differential Revision: D21471424

[facebook-github-bot]

This pull request was exported from Phabricator. Differential Revision: D21471424

[dr-ci]

💊 CI failures summary and remediations

As of commit ba426dc (more details on the Dr. CI page):

---

💚 💚 Looks good so far! There are no failures yet. 💚 💚

---

This comment was automatically generated by Dr. CI (expand for details).

Follow this link to opt-out of these comments for your Pull Requests.

Please report bugs/suggestions on the GitHub issue tracker.

See how this bot performed.

This comment has been revised 7 times.

[facebook-github-bot]

This pull request was exported from Phabricator. Differential Revision: D21471424

[facebook-github-bot]

This pull request was exported from Phabricator. Differential Revision: D21471424

[facebook-github-bot]

This pull request has been merged in 8181711.