351 Commits

Author	SHA1	Message	Date
Umut	702375f929	feat: consider p_error in virtual circuits	2022-11-18 12:45:28 +01:00
Umut	25b9a59e21	feat: introduce tag extension, create integer range getter for graphs, allow filtering in integer bit width getter	2022-11-18 12:33:03 +01:00
Umut	ccd3f9af6a	feat: introduce auto rounders	2022-11-08 14:17:22 +01:00
Umut	eb601f5948	chore: fix one of the common random test failures	2022-11-08 14:17:22 +01:00
Umut	c552a955c0	feat: introduce round bit pattern extension for virtual circuits	2022-11-03 11:52:41 +01:00
Umut	cb9cbb05ab	feat: introduce maxpool extension for virtual circuits	2022-11-02 16:12:00 +01:00
Umut	3de24a8369	feat: support multiplication between two encrypted values in virtual circuits	2022-10-28 12:07:01 +03:00
Umut	f0014dc05f	feat: improve fusing related error messages	2022-10-28 12:06:54 +03:00
Umut	0827cb9360	fix: correctly transfer sign information in direct definition	2022-10-25 14:19:15 +03:00
Umut	a2624086a2	feat: support axes kwarg for np.transpose	2022-10-18 11:58:50 +03:00
youben11	5f07a72e5c	fix: use pads arg for torch evaluation ... as we weren't supporting padding, we neglected to use them in the plain evaluation, but this result in a confusing error message for the user when trying to use padding. This fixes the problem by using padding properly during evaluation, and leaves the error up to the compiler.	2022-10-17 14:01:38 +03:00
Umut	d6ba76318f	feat: add support for np.expand_dims	2022-10-17 10:52:01 +03:00
Umut	ea34aeba09	feat: introduce 'global_p_error' configuration	2022-10-13 17:27:18 +03:00
Umut	21a0923e2e	feat: support up to 16-bit table lookups	2022-10-13 16:00:10 +03:00
Umut	66c707cd69	feat: introduce circuit decorator to directly define circuits	2022-10-13 15:18:18 +03:00
youben11	7901d8868c	fix: adhere to ONNX spec of kernel_shape ... ONNX spec: "kernel_shape : list of ints The shape of the convolution kernel. If not present, should be inferred from input W." We were taking the number of input/output feature maps into account, which we realized we should not.	2022-10-05 14:55:54 +01:00
Umut	593abc6d9a	feat: add various properties about the circuit	2022-09-21 13:11:04 +03:00
Umut	b8a91cac62	chore: ease signed integer restriction on high bit-widths	2022-09-08 11:45:46 +03:00
Umut	847aaae0ab	chore: fix a test which was failing on python 3.9 due to changed python semantics	2022-08-29 16:29:28 +03:00
Umut	05282285a3	feat: allow saving servers via MLIR	2022-08-29 13:04:22 +03:00
Umut	7415dd07e1	feat: support unsigned levelled operations on large bit-widths	2022-08-29 13:04:04 +03:00
Umut	ef7e0d762f	feat: support assignments to tensors	2022-08-23 14:56:36 +03:00
Umut	48014ed60a	feat: allow broadcast_to operation on scalars	2022-08-23 14:56:36 +03:00
Umut	710ee3408d	docs: fix documentation of static indexing execution test module	2022-08-23 14:56:36 +03:00
Umut	46e275d233	refactor: use kwargs to store index during static indexing	2022-08-23 14:56:36 +03:00
Umut	3aef4cd932	chore: remove pygraphviz	2022-08-18 11:55:33 +03:00
Umut	caebde9d51	chore: fix grammar mistakes on virtual circuits	2022-08-18 10:32:04 +03:00
youben11	0aadb4ac43	feat: support grouped conv2D	2022-08-16 16:28:37 +01:00
Umut	039a632c72	feat: support all kind of subtraction operations	2022-08-16 10:04:47 +03:00
Umut	09dde3013e	feat: support passing plain arrays to encrypt	2022-08-12 19:10:42 +03:00
Umut	6c6e657b6e	fix: resolve integration issues with the new compiler	2022-08-12 17:18:46 +03:00
Umut	d50b2c1547	feat: detect invalid values during bounds measurement	2022-07-27 09:49:02 +03:00
Umut	e398a4fbd0	feat: add support for np.broadcast_to	2022-07-25 13:18:48 +03:00
Umut	a0f167ac79	fix: reduce number of retries to 1 for tlu tests	2022-07-20 10:46:02 +03:00
Umut	f5842fb7ad	test: iteration of tracers	2022-07-19 15:15:21 +03:00
Umut	a60891292b	feat: cache constant operations during MLIR conversion	2022-07-18 18:22:23 +03:00
Umut	078512d55d	feat: create array extension	2022-07-12 14:30:08 +03:00
Umut	83f18e262a	feat: add option to print optimizer output	2022-07-11 19:24:42 +03:00
Umut	5a065769bb	fix: allow generator inputsets again	2022-07-08 15:16:07 +03:00
Umut	1cc5b576eb	fix: make artifacts truly optional to improve performance for large graphs	2022-06-22 17:05:12 +03:00
Umut	1e86c3b1e4	fix: don't allow signed clear inputs as tlu cannot be applied to them, add more signed input tests	2022-06-22 11:03:48 +03:00
Umut	a6b09ddf09	feat: support signed inputs	2022-06-22 11:03:48 +03:00
Umut	8a60a979cb	chore: remove manually set uint8 dtypes in tests as it's done automatically	2022-06-22 11:03:48 +03:00
Umut	ce1712c67c	fix: properly determine lca during fusing	2022-06-16 10:25:17 +03:00
Umut	11819fcf2f	feat: add support for np.ndarray.round() method	2022-06-16 10:25:17 +03:00
Umut	57aae5afdb	feat: add encrypted zeros and ones functions	2022-06-13 12:13:25 +03:00
Umut	59cacc35df	feat: raise proper error if function being compiled returns something unsupported	2022-06-13 11:04:03 +03:00
Umut	53e5dda732	feat: allow multiple graphs with the same name in debug artifacts	2022-06-10 13:48:10 +03:00
Umut	68e9ada9bf	feat: improve messages of parameter status and inputset related errors	2022-06-09 14:52:48 +03:00
Umut	dc501fb0ae	chore: convert an 8-bit test a 7-bit test to improve test time	2022-05-31 11:18:59 +03:00