- 02 1月, 2022 1 次提交
-
-
由 Oscar Gustafsson 提交于
There was a direct access to a NumPy configuration which was removed in 1.22. This PR removes the special handling and rely on accessing the configuration in a controlled way to avoid later issues.
-
- 23 12月, 2021 3 次提交
-
-
由 Brandon T. Willard 提交于
-
由 Brandon T. Willard 提交于
-
由 Brandon T. Willard 提交于
-
- 16 12月, 2021 4 次提交
-
-
由 Brandon T. Willard 提交于
-
由 Brandon T. Willard 提交于
-
由 Brandon T. Willard 提交于
-
由 Brandon T. Willard 提交于
These changes remove the stride-based manual computation of the new shape, since those are potentially sensitive to broadcasted arrays with no strides.
-
- 14 12月, 2021 9 次提交
- 13 12月, 2021 6 次提交
-
-
由 Brandon T. Willard 提交于
-
由 Brandon T. Willard 提交于
-
由 Brandon T. Willard 提交于
-
由 Brandon T. Willard 提交于
-
由 Brandon T. Willard 提交于
* Fix class, attribute, and function formatting * Update the relevant glossary terms and references to them * Rename documentation to better reflect their content * Remove redundant documentation * Reorder index so that it starts with the basic, necessary articles * Grammar and wording updates
-
由 Brandon T. Willard 提交于
-
- 11 12月, 2021 2 次提交
-
-
由 Brandon T. Willard 提交于
-
由 Brandon T. Willard 提交于
-
- 10 12月, 2021 3 次提交
-
-
由 Brandon T. Willard 提交于
Closes #692
-
由 Brandon T. Willard 提交于
-
由 Brandon T. Willard 提交于
-
- 08 12月, 2021 3 次提交
-
-
由 Brandon T. Willard 提交于
-
由 Brandon T. Willard 提交于
These changes only affect `PatternSub`, which now no longer allows constraints on non (logic) variable terms in its patterns. Likewise, the `values_eq_approx` and `skip_identities_fn` options are no longer supported.
-
由 Brandon T. Willard 提交于
-
- 07 12月, 2021 5 次提交
-
-
由 Brandon T. Willard 提交于
This commit introduces a `LocalOptTracker` object that performs an MRO-based lookup of `LocalOptimizer`s that track `Op` types.
-
由 Brandon T. Willard 提交于
-
由 Brandon T. Willard 提交于
-
由 Brandon T. Willard 提交于
-
由 Brandon T. Willard 提交于
-
- 05 12月, 2021 4 次提交
-
-
由 Brandon T. Willard 提交于
-
由 Brandon T. Willard 提交于
This improvement is due to the use of `x[indices] += y` instead of `np.add.at(x, indices, y)`, but comes at the cost of not handling duplicate `indices` values.
-
由 Brandon T. Willard 提交于
-
由 Brandon T. Willard 提交于
-