This is the attachable no-write proof: all open issues plus stale open PRs get a proposed maintainer-facing action, with evidence, but no GitHub write is allowed.
| Scope | Item | Dry-Run Action | Would Do | Why It Helps | GitHub Action Allowed |
|---|---|---|---|---|---|
| all_open_issues | issue #88: expose backend selection and cudnn settings to the end user | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #285: Keyword arguments passed to module's __call__ aren't forwarded to the hooks | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #499: Feature Request: Locally Connected Layer | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #630: Add Peephole connections for LSTMs? | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #634: Feature Request: NegativeSampling and HierarchicalSoftmax loss functions | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #711: Feature Request: Easier to extend base RNN implementation | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #842: BCELoss doesn't accept LongTensor targets | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #934: dataloader parallels over elements vs over batches | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #1178: Batched sparse QR factorizations and solves with cusolver | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #1249: Dice Loss PR | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #1328: Unhelpful CrossEntropyLoss dimension error message | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #1329: [proposed feature] Eve: Improving Stochastic Gradient Descent with Feedback | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #1359: Conjugate Gradient mentioned in docs, but not implemented | review_pr_backed_thread | Review active PRs and linked issues as one maintainer work thread. | Review issues, active PRs, referenced context, checks, and review guidance together instead of reopening each issue separately. | False |
| all_open_issues | issue #1362: Feature Request: noise contrastive estimation/negative sampling | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #1369: "Sparsified" mathematical operations | review_pr_backed_thread | Review active PRs and linked issues as one maintainer work thread. | Review issues, active PRs, referenced context, checks, and review guidance together instead of reopening each issue separately. | False |
| all_open_issues | issue #1410: BatchNorm should use Bessel's correction consistently | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #1442: Optimizers can't be moved to a different GPU | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #1462: Change sparse_mask to take indexing mask, rather than entire sparse tensor | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #1468: avg_pool functions hold input for backward | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #1487: [feature request/proposal] Relax scale_factor for nearest neighbor upsampling | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #1489: Optimizer should track parameter names and not id | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #1505: Cannot find Intel MKL | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #1512: [feature request] Support tensors of different sizes as batch elements in DataLoader | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #1529: [feature request] Caching allocator diagnostics and memory allocation tracing/visualization | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #1591: Pad PackedSequences to original batch length | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #1641: [Feature request] In-place 'max' method for Tensor | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #1642: feature request: On-the-fly Operation Batching in Dynamic Computation Graphs | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #1686: Feature Request: load_state_dict should take filenames | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #1736: Expose optimizer options as attributes when there's a single param group | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #1747: In-place bernoulli_ has more functionality than torch.bernoulli with output parameter | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #1750: Add support for colors (and maybe other attributes) to NVTX API | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #1794: Feature request: reverse_padded_sequence | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #1861: Factorized Output Layer | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #1927: [feature request] time-distributed layers for application of normal layers to sequence data | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #1932: Feature Request: ReLU on LSTMs and GRUs | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #2001: Implement similar PyTorch function as model.summary() in keras? | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #2129: [Feature request] truncated normal initializer(sampler) | review_pr_backed_thread | Review active PRs and linked issues as one maintainer work thread. | Review issues, active PRs, referenced context, checks, and review guidance together instead of reopening each issue separately. | False |
| all_open_issues | issue #2180: Discrepancy in BCEWithLogitsLoss and ClassNLLLoss | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #2312: DataParallel is not compatible with pack_padded_sequence | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #2400: Add CRF Layer | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #2407: Hard-negative mining using __getitem__ directive in Dataset class | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #2466: Autograd test failure on ppc64le | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #2478: ReduceLROnPlateau with a naive Backtracking | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #2482: "Shared memory manager connection has timed out" | human_triage_decision | Decide issue category, owner, and response pattern; no automated GitHub action. | The issue is represented and auditable, but the evidence does not justify an automated action. | False |
| all_open_issues | issue #2512: detach_() variant that affects all past uses too | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #2539: DataLoader converts cuda FloatTensor into cpu DoubleTensor when shape is (n,) | human_triage_decision | Decide issue category, owner, and response pattern; no automated GitHub action. | The issue is represented and auditable, but the evidence does not justify an automated action. | False |
| all_open_issues | issue #2545: Counter-intuitive Patience & Cooldown of ReduceLROnPlateau | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #2569: Add a [strong] hash function for tensor data | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #2575: ImportError: dlopen: cannot load any more object with static TLS | priority_human_review | Review early as release/regression risk; no automated GitHub action. | High priority or blocker signal should stay in a human-owned queue. | False |
| all_open_issues | issue #2576: CUDA multinomial is limited to 2^24 categories | priority_human_review | Review early as release/regression risk; no automated GitHub action. | High priority or blocker signal should stay in a human-owned queue. | False |
| all_open_issues | issue #2628: BatchNorm{1-2-3}d are redundant | review_pr_backed_thread | Review active PRs and linked issues as one maintainer work thread. | Review issues, active PRs, referenced context, checks, and review guidance together instead of reopening each issue separately. | False |
| all_open_issues | issue #2629: [feature request] add pairwise ranking loss | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #2691: Feature Request: Support grad of grad in fused RNNs | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #2703: Have ppc64le docker images? | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #2724: DataLoader gets stuck after model initialization | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #2732: support grid_sample with batch=1 but supprting batch affine parameters | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #2739: Proposal: simplify overloaded Tensor function signatures | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #2849: Learning rate scheduler have different APIs | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #2854: Add the new lr_scheduler which called poly | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #3076: BN slows down double-backprop enormously | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #3152: Sparse tensor .new(size) can be confusing | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #3176: Add safety checks in `index_add`/`scatter_add` | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #3281: DataLoader "casting" non statndard objects to lists | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #3356: Data sampling seems to be more complicated than necessary | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #3364: type of torch.bernoulli and torch.multinomial inconsistent | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #3390: High CPU use by clock_gettime syscall | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #3396: CUDA topk is slow for some input sizes | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #3468: improve performance of common CPU clone / contiguous calls with HPTT | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #3473: Feature Request: Distributed send arbitrary objects | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #3600: making .cuda() falls back to an identity function when gpu is not available | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #3619: Multiprocessing with torch.solve hangs | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #3625: Proposal: combine requires_grad and retain_grad() | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #3667: Exposing CuDNN benchmark strategy selection | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #3697: Deprecate inplace argument in torch.nn.functional | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #3790: Add SGDR, SGDW, AdamW and AdamWR | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #3791: Make pytest stop printing docstrings in its default diagnostic output | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #3818: Memory leak when doing backward with grad as yourself | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #3823: Fuse bias to CuDNN convolution | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #3863: Considerable slowdown in Adam.step after a number of epochs with multiple losses | human_triage_decision | Decide issue category, owner, and response pattern; no automated GitHub action. | The issue is represented and auditable, but the evidence does not justify an automated action. | False |
| all_open_issues | issue #3867: [Feature Request] Implement "same" padding for convolution operations? | human_triage_decision | Decide issue category, owner, and response pattern; no automated GitHub action. | The issue is represented and auditable, but the evidence does not justify an automated action. | False |
| all_open_issues | issue #3877: Wrap Cephes library for mathematical special functions | review_pr_backed_thread | Review active PRs and linked issues as one maintainer work thread. | Review issues, active PRs, referenced context, checks, and review guidance together instead of reopening each issue separately. | False |
| all_open_issues | issue #3898: Sparse matrices in dataloader error | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #3904: Implement DE in pytorch.optim | human_triage_decision | Decide issue category, owner, and response pattern; no automated GitHub action. | The issue is represented and auditable, but the evidence does not justify an automated action. | False |
| all_open_issues | issue #3920: [docs] Tensor.new is not documented | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #3931: Suppress hidden state output of RNNs? | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #3990: Raise an error when using magma built against wrong version of cuda | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #4073: Feature request: Correlation module | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #4085: GridSampler behaviours | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #4102: Make the generator tools data model more explicit | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #4123: Use the int64 version of MKL calls | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #4132: x.grad should be 0 but get NaN after x/0 | human_triage_decision | Decide issue category, owner, and response pattern; no automated GitHub action. | The issue is represented and auditable, but the evidence does not justify an automated action. | False |
| all_open_issues | issue #4145: [Proposal] Consistent `batch_first` effect for RNN modules | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #4148: ATen explicitly differentiated native function resolution hazard (call is ambiguous) | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #4159: Variable outputs of stochastic functions should never require grad | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #4181: Fused RNN refactor plan | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #4186: Feature Request: CPU performance optimization with MKL-DNN | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #4241: single-gpu works but multi-gpu hangs | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #4247: Feature request: sparse matrix max(axis) | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #4334: torch.cuda.device_count() returns 1 using 4 TitanX setup. | human_triage_decision | Decide issue category, owner, and response pattern; no automated GitHub action. | The issue is represented and auditable, but the evidence does not justify an automated action. | False |
| all_open_issues | issue #4358: Invoking MKL in multiprocessing with importing torch causes blocking | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #4392: Cache CuDNN benchmark selection, turn it on by default, use it across PyTorch runs | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #4400: Installation Optimise For Chinese Users Who Behind the Wall | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #4406: DistributedDataParallel doesn't converge well when using MPI | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #4474: [docs] Return value type hint to be specified in docs for torch.is_grad_enabled (i.e. -> bool) | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #4501: Bind in Python _backward ATen functions | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #4530: descriptor 'add' of 'torch._C._VariableBase' object needs an argument | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #4542: Carefully audit contiguity requirements of code | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #4551: Met 'cudnnDestroyDropoutDescriptor' while run multiply gpu-based models in multiply processes | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #4564: Mixed Tensor/TensorList arguments in ATen functions with explicit derivatives | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #4574: `from` keyword in `random_` gives error | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #4622: Protect user from No module named _C import error | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #4632: CUDNN_STATUS_INTERNAL_ERROR when training with conv3d | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #4636: Better header hygiene in ATen | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #4660: [Feature proposal] Add MC-derived optimizers | human_triage_decision | Decide issue category, owner, and response pattern; no automated GitHub action. | The issue is represented and auditable, but the evidence does not justify an automated action. | False |
| all_open_issues | issue #4703: Very slow on CPU | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #4716: Compilation issue: problem with GPU capability check | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #4731: Consider disallowing Variables that require grad in NCCL/comm functions | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #4758: Clang color diagnostics don't work with ninja | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #4825: [feature request]Support AVX512F intrinstics to vectorize operations | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #4829: [Feature Request] clip_grad_norm for sparse gradients | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #4906: Rebuild from no-CUDA to CUDA leads to: error: #error "Expected GLOO_USE_CUDA to be defined" | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #4927: [Feature Request]would PackedSequence support unsorted sequences? | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #4930: [feature request] Type-1 Multi-layer bidirectional RNN | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #4952: [Feature request] Optimize autograd/ATen when a gradient is clearly zero | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #4954: [docs] Docs website search finds duplicates and produces bad snippets | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #4958: [Feature Request] Extract glimpses from a batch of images (as in tf.image.extract_glimpse) | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #4959: Speed up data loading for `TensorDataset` if the underlying dataset supports index by a list of indices | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #4987: MultiGPU hangs Titan Xp in multiprocessing/queue.py | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #5070: NVIDIA_DRIVER_CAPABILITIES env variable is missing in pytorch docker images | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #5096: [feature request]Add an env variable to cover different pathes when testing code with openmp | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #5106: Saving model with runtime code changes | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #5157: BCELoss - weight parameter shape incorrect | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #5159: TakeBackward taking a significant portion of backward time | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #5161: ASAN detected leaks on python -c 'import torch' | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #5212: Weird error message in torch.split_size_or_sections | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #5231: [feature request] Stratified splits in random_split function | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #5266: [feature request] warnings for functions with unspecified dim arguments | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #5272: Unsafe out= keyword argument with tensors sharing storage | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #5280: BatchNorm1d raises RuntimeError (CUDNN_STATUS_BAD_PARAM) on 3D input. | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #5326: TestMultiprocessing.test_fd_sharing hangs with ASAN | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #5353: Bugs: Score Function approach in REINFORCE for PONG | priority_human_review | Review early as release/regression risk; no automated GitHub action. | High priority or blocker signal should stay in a human-owned queue. | False |
| all_open_issues | issue #5385: Handle python_arg_parser dtype constants better | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #5388: Perf regression: indexing 1-d tensor | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #5405: scatter_add_ should support scalar source (including Python scalar) | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #5434: RuntimeError: $ Torch: not enough memory: you tried to allocate 72GB. Buy new RAM! | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #5435: Gaussian Sampling | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #5489: torch.jit.trace(network, data) fails if data is an OrderedDict | human_triage_decision | Decide issue category, owner, and response pattern; no automated GitHub action. | The issue is represented and auditable, but the evidence does not justify an automated action. | False |
| all_open_issues | issue #5524: Redo torch.nn.functional docstring strategy | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #5528: TestNN.test_data_parallel takes 10G of memory | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #5565: [feature request] Different interpolation algos for 'grid_sample' function | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #5580: [feature request] F.interpolate to support integral data types: bool, int8, int32, int16, int64 ||| support uint8 on CUDA | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #5737: [feature request] Callback on learning rate drop in torch.optim.lr_scheduler.ReduceLROnPlateau | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #5740: [feature request] scalar input to scatter_add_ like scatter_ | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #5790: Add hookable weights | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #5857: Change THCudaCheck to suggest that device-side asserts likely mean that you have out of bound indices | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #5900: Build Fails on Gentoo with CUDA 9.1, GCC 6.4, Python 3.5 | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #5912: Conv-RNN combination slow in backward pass | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #5924: Install doesn't work with spaces in directory | batch_stale_issue_sweep | Review in its collapse group for close/reproduce/docs-update decision; do not post automatically. | Old or low-urgency issue can be handled in a component sweep instead of rediscovering context alone. | False |
| all_open_issues | issue #5953: LBFGS always give nan results, why | human_triage_decision | Decide issue category, owner, and response pattern; no automated GitHub action. | The issue is represented and auditable, but the evidence does not justify an automated action. | False |
| all_open_issues | issue #6007: Clean up the extra copies that occur in the execution engine and other places | human_triage_decision | Decide issue category, owner, and response pattern; no automated GitHub action. | The issue is represented and auditable, but the evidence does not justify an automated action. | False |