*   Added `sparse_gradient` option (default=false) that makes the gradient of these functions/ops sparse (`IndexedSlices`) instead of dense (`Tensor`), using new `SparseSegmentSum/Mean/SqrtNGradV2` ops.

*   `tf.nn.embedding_lookup_sparse`

    *   Optimized this function for some cases by fusing internal operations.

*   `tf.saved_model.SaveOptions`

Setting options in `include_router` still makes sense in some cases, for example, to override or increase configurations from a third party router included in an app. But in a router that is part of a bigger application, it would probably make more sense to add those settings when creating the `APIRouter`.

**In `FastAPI`**

not a // whole number of bytes, which is only the case for P-521. if c.curve == p521 && c.N[0]&0b1111_1110 == 0 { key[0] &= 0b0000_0001 } privateKey, err := NewPrivateKey(c, key) if err != nil { continue } // A "Pairwise Consistency Test" makes no sense if we just generated the // public key from an ephemeral private key. Moreover, there is no way to // check it aside from redoing the exact same computation again. SP 800-56A // Rev. 3, Section 5.6.2.1.4 acknowledges that, and doesn't require it. //...

not a // whole number of bytes, which is only the case for P-521. if c.curve == p521 && c.N[0]&0b1111_1110 == 0 { key[0] &= 0b0000_0001 } privateKey, err := NewPrivateKey(c, key) if err != nil { continue } // A "Pairwise Consistency Test" makes no sense if we just generated the // public key from an ephemeral private key. Moreover, there is no way to // check it aside from redoing the exact same computation again. SP 800-56A // Rev. 3, Section 5.6.2.1.4 acknowledges that, and doesn't require it. //...