Daily edition · AI Tools
TILens turns technical updates into a focused daily brief: official releases, trusted reporting, and practitioner analysis, deduplicated and organized by topic.
Changes since langchain-openrouter==0.2.7 release(openrouter): 0.2.8 (#39658) chore(model-profiles): refresh model profile data (#39646) chore(model-profiles): refresh model profile data (#39625) fix(openrouter):…
Changes since langchain-core==1.5.4 release(core): 1.5.5 (#39655) fix(core): make abatch_iterate consistent with batch_iterate for None and zero size (#39367) fix(core): respect pydantic aliases when validating tool…
Changes since langchain-openai==1.5.0 release(openai): 1.5.1 (#39653) fix(openai): preserve streamed encrypted reasoning (#39635) chore(infra): support langsmith gateway in CI (#39651)
[Testcase Refactoring] Add GENERIC hw_classification to dynamic_spec …
[Testcase Refactoring]Add hw_classification to test_compatibility.py …
[XPU][Refactor] Refactor test decoupling and device-agnostic in test_…
[XPU] Add bf16 tolerance override for LinearCrossEntropyLoss test_sav…
Pull Request resolved: #189536 Approved by: https://github.com/fffrog, https://github.com/albanD
Pull Request resolved: #189536 Approved by: https://github.com/fffrog, https://github.com/albanD
[XPU][Test]Port device agnostic cases in 7 test files to Intel GPU (#…
Adds a second way for mark_kernels to discover which nodes a scope contains, on top of the shared subscriber-callback registry: a RESOURCE / GRAPHNODE_CREATED handler records each node against whatever scope is open…
Adds a second way for mark_kernels to discover which nodes a scope contains, on top of the shared subscriber-callback registry: a RESOURCE / GRAPHNODE_CREATED handler records each node against whatever scope is open…
This PR is auto-generated nightly by this action. Update the pinned vision hash. Pull Request resolved: #193463 Approved by: https://github.com/pytorchbot
This PR is auto-generated nightly by this action. Update the pinned vision hash. Pull Request resolved: #193463 Approved by: https://github.com/pytorchbot
Fixes /Users/malfet/git/pytorch/pytorch/aten/src/ATen/native/mps/kernels/Convolution.metal:295:55: warning: comparison of integers of different signs: 'int' and 'const constant uint32_t' (aka 'const constant unsigned…
Scalar clamp kernels converted bounds directly to the input dtype, causing out-of-range values to wrap or raise an exception, which is mathematically incorrect, as such bounds should become a no-ops. Apply the same rule…
This PR is auto-generated nightly by this action. Update the pinned vllm hash. Pull Request resolved: #193461 Approved by: https://github.com/pytorchbot
[Test] Make test_complex_tensor.py device-agnostic for OOT backends (…
[Test] Make test_complex_tensor.py device-agnostic for OOT backends (…
Not SOL as it's reusing existing infra for GroupNorm but achieves comparable memory B/W on H100. Still a perf. win as the current implementation forces outputs in channels-first/contiguous which is problematic for…
As title Test passed Approved by: https://github.com/EikanWang Pull Request resolved: #193399 Approved by: https://github.com/EikanWang, https://github.com/atalman
[RAWTHROW][1a] Allow-by-rule policy; delete six exclude patterns (#19…
[RAWTHROW][0] Lexically-aware linter (no exclude_patterns change) (#1…
[Inductor] Query DeviceInterface for device identity and capability (…
[Inductor] Query DeviceInterface for device identity and capability (…
[ghstack-poisoned]
test/dynamo/test_after_aot.py
Adds an unstable.yml arm that mirrors trunk's linux-jammy-cuda13.0-py3.10-gcc11 "default" shards but runs with no in-process pytest reruns and exactly one retry in a fresh process, so the two policies can be compared on…
adjust test case
