5350 Commits

Author	SHA1	Message	Date
Disty0	16d6c03d45	Optimum Quanto activations support	2024-08-21 17:30:45 +03:00
Disty0	6a58d5210b	Make eval use apply_compile_to_model	2024-08-19 20:06:51 +03:00
Disty0	b025d1dab1	Round memory size in settings	2024-08-19 18:56:28 +03:00
Disty0	cc89ed84f9	Rename cpu offload to model offload	2024-08-19 18:48:09 +03:00
Disty0	2586a18218	Don't add 0.1 to the GPU memory	2024-08-18 13:18:24 +03:00
Disty0	5e1da44bba	IPEX fix custom_fwd x2	2024-08-18 05:22:11 +03:00
Disty0	b862400ea2	IPEX fix AMP custom_fwd	2024-08-18 05:11:00 +03:00
Disty0	0734c75ef4	Add Heun FlowMatch	2024-08-18 02:17:55 +03:00
Disty0	a3f26c9df0	Convert Dynamic Attention SDP to a global SDP option	2024-08-18 01:44:27 +03:00
Disty0	5c857e8aaf	Add check for Flux attention processor	2024-08-18 00:35:10 +03:00
Disty0	bce3c7eabf	Fix --Xvram flags not activating offload	2024-08-18 00:23:15 +03:00
Disty0	5a75b120ae	Fix Lora with Balanced Offlaod	2024-08-16 21:22:41 +03:00
Disty0	3f5c3ba0d8	Add warning to Quanto with balanced and sequential offload	2024-08-16 02:58:43 +03:00
Disty0	7a6b45b410	Balanced offload move device map calcs	2024-08-16 02:49:42 +03:00
Disty0	80be079130	Flux load quant model to cpu	2024-08-15 22:17:41 +03:00
Disty0	f3f721e39a	Quanto disable gemm kernels	2024-08-14 20:26:46 +03:00
Disty0	237cab2aa1	Add offload check to cascade's vqgan	2024-08-14 17:42:49 +03:00
Disty0	8699e0cacb	Add gc to balanced offload	2024-08-14 17:37:07 +03:00
Disty0	119c37212d	More offload checks	2024-08-14 17:27:10 +03:00
Disty0	7edc864014	Add module name to disk offload path	2024-08-14 12:21:54 +03:00
Disty0	8619a7f35c	Better balanced offload	2024-08-14 11:53:35 +03:00
Disty0	a73716bf89	Add meta to device check	2024-08-13 23:35:56 +03:00
Disty0	97a5faed4c	Fix qint4	2024-08-13 23:24:42 +03:00
Disty0	ef1dedf5c1	Samplers prefer model defaults over diffusers defaults	2024-08-13 21:41:27 +03:00
Alexey Shmelev	7d805e90ee	there are no multiple models, so no need to check	2024-08-13 11:09:13 -04:00
Vladimir Mandic	94fce42ae1	Merge pull request #3368 from osanseviero/master ... Update to new Hugging Face stuff	2024-08-13 16:00:30 +02:00
osanseviero	04f4757b88	Update to new huggingface stuff	2024-08-13 10:48:11 +02:00
Disty0	26d1d42b06	IPEX update interpolate hijack	2024-08-12 10:47:00 +03:00
Disty0	dcedcbae9d	IPEX fixes	2024-08-12 03:01:46 +03:00
Disty0	0f7b7e83bc	Fix memory detection when no gpu is present	2024-08-12 01:41:54 +03:00
Disty0	70c2e84d26	Prompt cache support for Flux	2024-08-12 01:02:13 +03:00
Disty0	da9c46c628	Fix optimum-quanto not found	2024-08-12 00:49:31 +03:00
Disty0	6a1af56834	FLUX quant loading support	2024-08-11 23:54:05 +03:00
Disty0	fb89e26b54	Auto detect memory size ffor balaced offload	2024-08-11 19:01:49 +03:00
Disty0	4ba173280b	Check device index for balanced offload	2024-08-11 17:35:07 +03:00
Disty0	e3b087b6c0	Add balanced offload mode and make offload modes a single choice list	2024-08-11 17:27:30 +03:00
Seunghoon Lee	c2c4e17d51	enable FluxPipeline	2024-08-09 18:47:33 +09:00
Seunghoon Lee	a9f27991a3	enable diffusers_move_unet for Flux	2024-08-09 16:12:01 +09:00
Seunghoon Lee	9c4213e4a8	tcmalloc experiment	2024-08-09 16:11:12 +09:00
Seunghoon Lee	864263f570	accurate wsl check	2024-08-09 10:38:28 +09:00
Disty0	e729b57c07	Make ruff happy	2024-08-09 04:27:54 +03:00
Disty0	58d49f25c4	Custom sampler support for Cascade Decoder	2024-08-09 04:20:52 +03:00
Disty0	6431296ec1	Fix Cascade empty prompt encode	2024-08-09 00:32:19 +03:00
Seunghoon Lee	e1c40384ad	zluda hijack torch jit	2024-08-09 01:11:13 +09:00
Seunghoon Lee	c490b9cb51	fix first launch	2024-08-09 00:45:12 +09:00
Seunghoon Lee	61961f5d6d	hipblaslt check torch version	2024-08-09 00:35:46 +09:00
Seunghoon Lee	0d57fa3016	fix zluda torch cpp_extension	2024-08-08 14:44:53 +09:00
Disty0	db0f6c73dd	Cascade fix get_timestep_ratio_conditioning	2024-08-07 19:39:09 +03:00
Disty0	1cf87ef409	Change cascade load order	2024-08-07 05:03:26 +03:00
Disty0	9818950991	Cascade fixes	2024-08-07 04:54:41 +03:00