*   Added a distributed runtime using GRPC
*   Move skflow to `contrib/learn`
*   Better linear optimizer in `contrib/linear_optimizer`
*   Random forest implementation in `contrib/tensor_forest`
*   CTC loss and decoders in `contrib/ctc`
*   Basic support for `half` data type
*   Better support for loading user ops (see examples in `contrib/`)
*   Allow use of (non-blocking) Eigen threadpool with

pkg log/syslog (freebsd-riscv64), method (*Writer) Debug(string) error #53466
pkg log/syslog (freebsd-riscv64), method (*Writer) Emerg(string) error #53466
pkg log/syslog (freebsd-riscv64), method (*Writer) Err(string) error #53466
pkg log/syslog (freebsd-riscv64), method (*Writer) Info(string) error #53466
pkg log/syslog (freebsd-riscv64), method (*Writer) Notice(string) error #53466
pkg log/syslog (freebsd-riscv64), method (*Writer) Warning(string) error #53466

pkg log/syslog (freebsd-arm64), method (*Writer) Emerg(string) error
pkg log/syslog (freebsd-arm64), method (*Writer) Err(string) error
pkg log/syslog (freebsd-arm64), method (*Writer) Info(string) error
pkg log/syslog (freebsd-arm64), method (*Writer) Notice(string) error
pkg log/syslog (freebsd-arm64), method (*Writer) Warning(string) error
pkg log/syslog (freebsd-arm64), method (*Writer) Write([]uint8) (int, error)

## 0.101.0

### Features

* ✨ Enable Pydantic's serialization mode for responses, add support for Pydantic's `computed_field`, better OpenAPI for response models, proper required attributes, better generated clients. PR [#10011](https://github.com/tiangolo/fastapi/pull/10011) by [@tiangolo](https://github.com/tiangolo).

### Refactors

Section 10.1.2.2. // K = HMAC (K, V || 0x00 || provided_data) h := hmac.New(hash, K) h.Write(d.V) h.Write([]byte{0x00}) h.Write(entropy) h.Write(nonce) switch s := s.(type) { case plainPersonalization: h.Write(s) case blockAlignedPersonal: l := len(d.V) + 1 + len(entropy) + len(nonce) for _, b := range s { pad000(h, l) h.Write(b) l = len(b) } } K = h.Sum(K[:0]) // V = HMAC (K, V) h = hmac.New(hash, K) h.Write(d.V) d.V = h.Sum(d.V[:0]) // K = HMAC (K, V || 0x01 || provided_data). h.Reset() h.Write(d.V)...

Section 10.1.2.2. // K = HMAC (K, V || 0x00 || provided_data) h := hmac.New(hash, K) h.Write(d.V) h.Write([]byte{0x00}) h.Write(entropy) h.Write(nonce) switch s := s.(type) { case plainPersonalization: h.Write(s) case blockAlignedPersonal: l := len(d.V) + 1 + len(entropy) + len(nonce) for _, b := range s { pad000(h, l) h.Write(b) l = len(b) } } K = h.Sum(K[:0]) // V = HMAC (K, V) h = hmac.New(hash, K) h.Write(d.V) d.V = h.Sum(d.V[:0]) // K = HMAC (K, V || 0x01 || provided_data). h.Reset() h.Write(d.V)...