26 Commits

Author	SHA1	Message	Date
Yousef Rafat	3dd39efa03	some fixes	2025-12-02 23:40:31 +02:00
Yousef Rafat	a58133f188	updated the hunyuan moe forward ... splitted the forward statement between ready and pending experts	2025-12-01 22:58:53 +02:00
Yousef Rafat	76e14d69b2	operations + device + dtype | checkpoint skip	2025-12-01 18:53:43 +02:00
Yousef Rafat	88c350bfed	corrected img_ratio	2025-11-28 20:43:01 +02:00
Yousef Rafat	bd2c2f7375	removed additional token injection	2025-11-28 18:21:20 +02:00
Yousef Rafat	334041f6a6	efficiency improvements	2025-11-26 21:21:45 +02:00
Yousef Rafat	823870db53	fixed some mistakes/errors	2025-11-25 21:34:03 +02:00
Yousef Rafat	b4001bbd27	input correction and improvements	2025-11-25 00:25:27 +02:00
Yousef Rafat	86e9a7a669	updates to the input	2025-11-23 22:20:24 +02:00
Yousef Rafat	a3ac798d4e	decrease peak memory in moe forward	2025-11-22 23:02:42 +02:00
Yousef Rafat	ae8592ebf5	zero-copy and optimized moe loader	2025-11-22 12:37:10 +02:00
Yousef Rafat	4d982e83f6	added clip encoder	2025-11-21 17:22:37 +02:00
Yousef Rafat	b84af5b947	small attention fix	2025-11-17 23:03:52 +02:00
Yousef Rafat	3f71760913	resblock fix	2025-11-17 06:50:54 +02:00
Yousef Rafat	61b1efdaf0	vectrozied correct implementation of moe forward	2025-11-16 19:25:37 +02:00
Yousef Rafat	4a5509a4c5	.	2025-11-16 16:20:35 +02:00
Yousef Rafat	d731c58353	improving performance and fixing race condition	2025-11-16 16:19:39 +02:00
Yousef Rafat	12cc6924ac	meta init	2025-11-14 20:10:52 +02:00
Yousef Rafat	7b4c1e8031	async cache revamp ... Added an async loading and offloading of moe layers, having consistent memory with oom errors. Used to give oom error after the third layer with 24 giga bytes gpu, now goes to the end with consistent memory with minimal latency	2025-11-14 09:15:16 +02:00
Yousef Rafat	44346c4251	removed all errors	2025-11-08 19:49:02 +02:00
Yousef Rafat	5056a1f4d4	important fixes	2025-11-06 00:24:49 +02:00
Yousef Rafat	9e9c536c8e	fixes from testing	2025-11-04 23:55:16 +02:00
Yousef Rafat	ca119c44fb	returned kv cache for image generation	2025-11-01 23:06:11 +02:00
Yousef Rafat	10a17dc85d	a bunch of fixes	2025-11-01 16:40:49 +02:00
Yousef Rafat	a2fff60d4c	vectorized implementation of moe/fixes for issues	2025-10-31 23:53:13 +02:00
Yousef Rafat	de43880bdb	Hunyuan Image 3.0	2025-10-31 18:56:20 +02:00