becomes `axis` * `tf.reduce_mean`: `reduction_indices` becomes `axis` *
`tf.reduce_min`: `reduction_indices` becomes `axis` * `tf.reduce_prod`:
`reduction_indices` becomes `axis` * `tf.reduce_sum`: `reduction_indices`
becomes `axis` * `tf.reverse_sequence`: `batch_dim` becomes `batch_axis`,
`seq_dim` becomes `seq_axis` * `tf.sparse_concat`: `concat_dim` becomes `axis` *
`tf.sparse_reduce_sum`: `reduction_axes` becomes `axis` *

a3b3 a2b3 a1b3 a0b3 + // a4b4 a3b4 a2b4 a1b4 a0b4 = // ---------- // r8 r7 r6 r5 r4 r3 r2 r1 r0 // // We can then use the reduction identity (a * 2²⁵⁵ + b = a * 19 + b) to // reduce the limbs that would overflow 255 bits. r5 * 2²⁵⁵ becomes 19 * r5, // r6 * 2³⁰⁶ becomes 19 * r6 * 2⁵¹, etc. // // Reduction can be carried out simultaneously to multiplication. For // example, we do not compute r5: whenever the result of a multiplication // belongs to r5, like a1b4, we multiply it by 19 and add the result...

a3b3 a2b3 a1b3 a0b3 + // a4b4 a3b4 a2b4 a1b4 a0b4 = // ---------- // r8 r7 r6 r5 r4 r3 r2 r1 r0 // // We can then use the reduction identity (a * 2²⁵⁵ + b = a * 19 + b) to // reduce the limbs that would overflow 255 bits. r5 * 2²⁵⁵ becomes 19 * r5, // r6 * 2³⁰⁶ becomes 19 * r6 * 2⁵¹, etc. // // Reduction can be carried out simultaneously to multiplication. For // example, we do not compute r5: whenever the result of a multiplication // belongs to r5, like a1b4, we multiply it by 19 and add the result...

## 0.112.4

This release is mainly a big internal refactor to enable adding support for Pydantic models for `Form` fields, but that feature comes in the next release.

This release shouldn't affect apps using FastAPI in any way. You don't even have to upgrade to this version yet. It's just a checkpoint. 🤓

### Refactors