Concrete-ml

Feature
* Add structured pruning to QNNs ([`70bff38`](https://github.com/zama-ai/concrete-ml/commit/70bff38cc5873e4cd483148dcd49716b8cb22f0d))
* Add accumulator rounding ([`8ee9267`](https://github.com/zama-ai/concrete-ml/commit/8ee92674f26b66747c33dbc311f9b5a6289508e0))
* Add bitwidth and value range report per layer ([`dd37f4e`](https://github.com/zama-ai/concrete-ml/commit/dd37f4e3394c287cd3a37312977ae9c97d495dbc))
* Extend deployment features (AWS and docker) ([`77e2d80`](https://github.com/zama-ai/concrete-ml/commit/77e2d80ab3cf0b2f491c128a6ced87d0fb620282))
* Add sentiment analysis deployment use-case ([`96c9158`](https://github.com/zama-ai/concrete-ml/commit/96c9158106af0b337cf3e9287201a2c03677bd7a))
* Support ONNX operators Gather, Slice, Shape, ConstantOfShape ([`667e9ae`](https://github.com/zama-ai/concrete-ml/commit/667e9ae51d4fe51a948686c5f6eb05c18dd60a85))
* Include quant and dequant steps in QuantizedModule's forward method ([`55369ed`](https://github.com/zama-ai/concrete-ml/commit/55369ed67ed1d2e01c6bf3ca4baeb925c30f0213))
* Add scikit-learn model serialization ([`ae7658b`](https://github.com/zama-ai/concrete-ml/commit/ae7658bf40166f6feca780212d7d8bdf74530e9d))
* Add cifar-10 8-bit model deployment ([`9177058`](https://github.com/zama-ai/concrete-ml/commit/91770584c9928878a121f1a66555e0dca88b92e6))
* Support pandas and list inputs in predict and compile methods for NNs ([`6a5e619`](https://github.com/zama-ai/concrete-ml/commit/6a5e619f51997e8992783880b4796f9ec3c4684c))
* Add example of model deployment to use-cases ([`be7bcb0`](https://github.com/zama-ai/concrete-ml/commit/be7bcb02a074228c9bfe512cf1959ddcb65ea3c3))
* Support pandas, list and torch for trees and linear models ([`8156bc9`](https://github.com/zama-ai/concrete-ml/commit/8156bc9359e0d849219f1ddd86b5536c5778821f))
* Simplify the API by removing n_bits for compile_brevitas_qat ([`d081212`](https://github.com/zama-ai/concrete-ml/commit/d08121274bd051b43bca6b276e8a86ef0bcabaab))

Fix
* Fix ci packaging ([`bdda1da`](https://github.com/zama-ai/concrete-ml/commit/bdda1dac45fe5f84f580351a03a171dc03194e21))
* Fix deploy_to_aws for python 3.10 ([`7665809`](https://github.com/zama-ai/concrete-ml/commit/7665809ab347fdfb13801350754257687d6aa7e1))
* Non-quantized NN constant folding bug ([`bdc04c4`](https://github.com/zama-ai/concrete-ml/commit/bdc04c4db23c4206eabcb7a3ddc61489758021bc))
* Make the client-server API support Tweedie models ([`0de7398`](https://github.com/zama-ai/concrete-ml/commit/0de739876fe8cc28b0b794abb904128744ee27c4))
* QNN API improvements and pruning fix ([`33c4bf0`](https://github.com/zama-ai/concrete-ml/commit/33c4bf03d4a8b9ff9c7e7e6e22810f9b8f80b0b1))
* Set specific dependency versions ([`9729dc6`](https://github.com/zama-ai/concrete-ml/commit/9729dc6c9a4224bf3eba2c5ed9a5f354be845c16))
* Flaky client server ([`3eb86c1`](https://github.com/zama-ai/concrete-ml/commit/3eb86c14025af8960720d016173d7ca9ac48bee3))
* Fixing issues with pytest and macOS ([`a0c22fa`](https://github.com/zama-ai/concrete-ml/commit/a0c22fa0074cb11ead3cb2ab41bb3b1c9be89921))

Documentation
* Add tree experiments ([`e1c0ce0`](https://github.com/zama-ai/concrete-ml/commit/e1c0ce0ad99d9a61947e84807c62ab2af70f797d))
* Add chapter on optimization and simulation ([`5454620`](https://github.com/zama-ai/concrete-ml/commit/54546203e48486444a30b17396fe5ed0b026db60))
* Update simulation ([`d298459`](https://github.com/zama-ai/concrete-ml/commit/d298459b05f1f6fb6b7c5efbcab58be5835bc77d))
* Good quantization configurations for target accumulator bit-widths ([`ef6355f`](https://github.com/zama-ai/concrete-ml/commit/ef6355f07f521266194d8e602eef658fe94cd68c))
* Add rounding documentation ([`b7f56d5`](https://github.com/zama-ai/concrete-ml/commit/b7f56d507f57be3524f040f8a5e06e99da6e1b71))
* Add an example that separates encryption, FHE execution and decryption ([`d4681fb`](https://github.com/zama-ai/concrete-ml/commit/d4681fbba877e4ad09d9dc621e896a0e7e17e9b2))

Feature
* Support 20+ bits linear models ([`4f112ca`](https://github.com/zama-ai/concrete-ml/commit/4f112caf899ce3e7943f2363eca1abf4828f6f9c))
* Add python 3.10 support ([`aede49b`](https://github.com/zama-ai/concrete-ml/commit/aede49b4935a473b34ed4fc4c41572cba2815f48))
* Add a CIFAR-10 CNN with 8-bit accumulators and show p_error search ([`35715e2`](https://github.com/zama-ai/concrete-ml/commit/35715e24daa6bf157d4af189a003eaf937500396))
* Add tutorials for transfer learning for CIFAR-10/100. ([`42405c5`](https://github.com/zama-ai/concrete-ml/commit/42405c5a34a49d0caeab6013a1197bb7bb60a528))
* Add CIFAR-10 VGG CNN with split clear/FHE compilation. ([`637c272`](https://github.com/zama-ai/concrete-ml/commit/637c272aa4efd8b40530ad77354fe79712e73ecc))
* Change the license ([`a52d917`](https://github.com/zama-ai/concrete-ml/commit/a52d9179958244a428d955c7c1a7a00789b512ca))
* Add support for global_p_error ([`b54fcac`](https://github.com/zama-ai/concrete-ml/commit/b54fcac3ca3b1e4fbf1376345a76f2f336ac555b))

Fix
* Flaky FHE vs VirtualLib overflow ([`1780cd5`](https://github.com/zama-ai/concrete-ml/commit/1780cd5d0a680056ea632fd9dcc0033dd5521c1d))
* Ensure all operations in QNNs are done in FP64 ([`52e87b7`](https://github.com/zama-ai/concrete-ml/commit/52e87b7d979dd7ed93f0d9181432e621e8e93512))
* Raise error when model results are mismatched between Concrete-ML and VL ([`b7fa8c1`](https://github.com/zama-ai/concrete-ml/commit/b7fa8c1115ac9b78dc9645a1104b18e826aa61ca))
* Set specific dependency versions ([`f2dfc3e`](https://github.com/zama-ai/concrete-ml/commit/f2dfc3eed5e64adf1ac52984cbc8fe34ef85fddb))
* Flaky client server API ([`1495214`](https://github.com/zama-ai/concrete-ml/commit/1495214fce24b6e12a0ccc2679a6c9c1272c1408))
* Issues with pytest and macOS ([`5196c68`](https://github.com/zama-ai/concrete-ml/commit/5196c68bd266e9f699e687aa3d0f26eda3a70137))

Documentation
* Add a showcase of use-cases and tutorials ([`36adc09`](https://github.com/zama-ai/concrete-ml/commit/36adc09a35491f2f4133f55302fbf6be88136749))
* Add global_p_error ([`b6b4d7a`](https://github.com/zama-ai/concrete-ml/commit/b6b4d7a3158f6ffd7d7cbb744f87b46f54057842))
* Add CIFAR-10/100 examples for the fine-tuning approach. ([`45a4f66`](https://github.com/zama-ai/concrete-ml/commit/45a4f66ec3713c2c9eddc1ed964118ffefd50ebd))
* Fully connected NN on MNIST using 16b in VL ([`f0be5f3`](https://github.com/zama-ai/concrete-ml/commit/f0be5f31f003536b3558c691aacb1e824e45ba56))
* Provide an image filtering demo app ([`fd11f25`](https://github.com/zama-ai/concrete-ml/commit/fd11f25cecedd8ddd38db508d188eec1a4003e61))

Feature
* Extend python 3.7.14 support to 3.7.1 ([`eb212bf`](https://github.com/zama-ai/concrete-ml-internal/commit/eb212bf267b0899cda9aa94f840405ed89c2ea27))

Fix
* Fixing an issue with LinearRegression ([`ebd06b4`](https://github.com/zama-ai/concrete-ml-internal/commit/ebd06b46e08100520808801f39044e0633167f19))

Feature
* Python 3.7 support ([`fef90d1`](https://github.com/zama-ai/concrete-ml-internal/commit/fef90d1d77f5ca27923a157d1faf29c38e643e0e))
* Remove constraints in numpy_reduce_sum ([`89668bf`](https://github.com/zama-ai/concrete-ml-internal/commit/89668bf3f974a5d503055eedd52ac65e36a4ed18))
* Check if a network imported with import_qat=True is quantized ([`24e8f88`](https://github.com/zama-ai/concrete-ml-internal/commit/24e8f885627ab223758e997a7cf36eff16ea2dbe))

Documentation
* Move titanic notebook to use_case_examples and cleaned SentimentClassification notebook ([`14502f0`](https://github.com/zama-ai/concrete-ml-internal/commit/14502f0cae2c24786814ebe1e9572c013d64c5b6))