Concrete-ml

Feature
* Support Expand Equal ONNX op ([`cf3ce49`](https://github.com/zama-ai/concrete-ml/commit/cf3ce492dd9d9a433545bd1b827b211fda660bd8))
* Add rounding feature on cml trees ([`064eb82`](https://github.com/zama-ai/concrete-ml/commit/064eb821a2171ba921ed327deded5b2f1d12e7c6))
* Add multi-output support ([`fef23a9`](https://github.com/zama-ai/concrete-ml/commit/fef23a9c85bbac02dc3a1651fe5808152aed894f))
* Allow QuantizedAdd produces_output_graph ([`0b57c71`](https://github.com/zama-ai/concrete-ml/commit/0b57c71140cca7535d1f5bdc0c088b205e28cbb5))
* Encrypted gemm support - 3d inputs - better rounding control - sgd training test ([`111c7e3`](https://github.com/zama-ai/concrete-ml/commit/111c7e314f12c072e44fc5a4615219a765d1a5ee))

Fix
* Add --no-warnings flag to linkchecker ([`1dc547e`](https://github.com/zama-ai/concrete-ml/commit/1dc547e093e9aa5f7437325e132cdf8732f52f0b))
* Fix wrong assumption in ReduceSum operator's axis parameter ([`1a592d7`](https://github.com/zama-ai/concrete-ml/commit/1a592d751659c0fdec86c88f75d370f87439cd4f))
* Mark flaky tests due to issue in simulation ([`4f67883`](https://github.com/zama-ai/concrete-ml/commit/4f678839f1b2e9b3b53e7175fb6da07cb8932876))
* Update learning rate default value for XGB models ([`e4984d6`](https://github.com/zama-ai/concrete-ml/commit/e4984d6a2085955187ac3797e4ef7adef897de42))

Documentation
* Update api doc ([`d8e9e64`](https://github.com/zama-ai/concrete-ml/commit/d8e9e648055346a888eb53cd7b986141b9aae093))
* Update Apple Silicon install information ([`4c0c02f`](https://github.com/zama-ai/concrete-ml/commit/4c0c02f15672313f641b92df22f107343fbfb647))

Feature
* Add SGD regressor ([`abb143c`](https://github.com/zama-ai/concrete-ml/commit/abb143c0852f7a37a6de916bc119147096c0067b))

Fix
* Fix shape output mismatch for KNNClassifier ([`6de7c6e`](https://github.com/zama-ai/concrete-ml/commit/6de7c6e85f668193d026763530507f5afe8f50f1))

Feature
* Enable import of fitted linear sklearn models ([`771c7ff`](https://github.com/zama-ai/concrete-ml/commit/771c7ffe05ef5c88ae5f3deec68fb1cb0cec80f8))
* Support QAT models in hybrid model ([`526b000`](https://github.com/zama-ai/concrete-ml/commit/526b000cb13291ea4f038c9dd76db151fb8f2729))
* Expose statuses to compile torch ([`8abddf6`](https://github.com/zama-ai/concrete-ml/commit/8abddf6afe84983e952a1ca9d40e90dc05e66e21))
* Add KNN classifier in CML ([`1c33ec8`](https://github.com/zama-ai/concrete-ml/commit/1c33ec8121f44be330387508f82c7b6dd30cf843))
* Add power of two scaling adapter for roundPBS ([`546fac9`](https://github.com/zama-ai/concrete-ml/commit/546fac9cbc060b020224a997c2cd2c4d1a7c67c5))
* Add hybrid FHE models ([`be6aa6e`](https://github.com/zama-ai/concrete-ml/commit/be6aa6e7e3fc963b9fdef4fa73ec0598606fb50e))

Fix
* Fix confusing print in CNN tutorial of advanced-examples ([`9136c47`](https://github.com/zama-ai/concrete-ml/commit/9136c47e5a8c5dbfaa0b7f2b736da35ba77bdbdd))
* Fix path parsing and default in hybrid serving ([`afd049a`](https://github.com/zama-ai/concrete-ml/commit/afd049ae5394fc2ba37ac6230cfece5996ffc373))
* Fix flaky padding test ([`6aaf5f0`](https://github.com/zama-ai/concrete-ml/commit/6aaf5f03868dc58e13fa9f04147b6ad8354baae0))
* Fix issues with OMP library ([`2b61846`](https://github.com/zama-ai/concrete-ml/commit/2b61846fd8da94ccc84757d5bb358a66c1f4b0e9))
* Make sure structured pruning and unstructured pruning work well together ([`ada18ab`](https://github.com/zama-ai/concrete-ml/commit/ada18abc54121cf21008fcaaac2f21f783f0fe7e))
* Fix structured pruning crash not caught by test ([`cafd8d1`](https://github.com/zama-ai/concrete-ml/commit/cafd8d140e0077b2843136925187d80c9588aa0c))
* Fix bad top1 accuracy in cifar_brevitas_training use case ([`f0a984e`](https://github.com/zama-ai/concrete-ml/commit/f0a984e7031506f0730014e2daddaca6bc3b6fa9))
* Fix flaky double_fit test ([`3da6408`](https://github.com/zama-ai/concrete-ml/commit/3da6408cc1df1ee382fca7d299ee8b9f0fa342ef))
* Remove workaround for simulating linear models ([`3f622bc`](https://github.com/zama-ai/concrete-ml/commit/3f622bc51d67bd1407d0577bb45b44f8d053c1e1))
* Re-compute quantization params when re-fitting linear models ([`3bad62e`](https://github.com/zama-ai/concrete-ml/commit/3bad62e7490493ee999e6f01c7316f4b91cbcb28))

Documentation
* Fix and improve credit scoring use case example ([`e4db376`](https://github.com/zama-ai/concrete-ml/commit/e4db376757af57b382de18dbf7fa55dbbc69be06))
* Update contribution part ([`f2822d1`](https://github.com/zama-ai/concrete-ml/commit/f2822d1a9db6b4f2590c68aaf2d30b2bd8aac3a7))
* Document KNN, PoT, Hybrid models ([`68a0b4c`](https://github.com/zama-ai/concrete-ml/commit/68a0b4ca7fd906fa14e6652453d27a6485158c6c))
* Update mnist CNN ([`f80c90b`](https://github.com/zama-ai/concrete-ml/commit/f80c90bd98afb85cc5a089b1c60fbf609a3817b3))
* Update mnist Fully Connected example with PoT + rounding ([`6e3d003`](https://github.com/zama-ai/concrete-ml/commit/6e3d003c8f11780faafb45222404e8ebdcc7c576))
* Update cifar_brevitas_training accuracy using representative calibration set ([`39480ef`](https://github.com/zama-ai/concrete-ml/commit/39480ef07c322172f4d9894e11a79876f029cb1e))
* Correct n_bits markdown value in the LLM use case notebook ([`0cf1174`](https://github.com/zama-ai/concrete-ml/commit/0cf117460e8c6f647629e2ff2ad4c1004e2dc53c))

Feature
* Add hybrid FHE models ([`be6aa6e`](https://github.com/zama-ai/concrete-ml/commit/be6aa6e7e3fc963b9fdef4fa73ec0598606fb50e))

Fix
* Flaky test double_fit ([`3da6408`](https://github.com/zama-ai/concrete-ml/commit/3da6408cc1df1ee382fca7d299ee8b9f0fa342ef))
* Re-compute quantization parameters when re-fitting linear models ([`3bad62e`](https://github.com/zama-ai/concrete-ml/commit/3bad62e7490493ee999e6f01c7316f4b91cbcb28))

Documentation
* Add link to deployment use case examples ([`f9333b3`](https://github.com/zama-ai/concrete-ml/commit/f9333b3e66312f94e0386956af9f4b39907781b9))
* Correct mention of n_bits in LLM use case notebook text ([`0cf1174`](https://github.com/zama-ai/concrete-ml/commit/0cf117460e8c6f647629e2ff2ad4c1004e2dc53c))