Default Branch

main 868e06eb8b · tests: fix test_model_install.py · Updated 2025-01-04 00:21:23 +08:00

ZIP  TAR.GZ

Branches

ryan/model-offload-5-dynamic-cache-limits b800fffcbb · Fix ModelCache execution device selection in unit tests. · Updated 2025-01-04 07:11:34 +08:00	5 17		ZIP  TAR.GZ
maryhipp/limit-board-name-length f00b8fc8e3 · fix(api): limit board_name length to 300 characters · Updated 2025-01-04 05:28:17 +08:00	0 1		ZIP  TAR.GZ
gh-pages cbc5624c99 · Deployed 868e06e with MkDocs version: 1.6.1 · Updated 2025-01-04 00:24:28 +08:00	15168 1		ZIP  TAR.GZ
ryan/model-offload-4-model-cache-partial-load bbc078a364 · Add get_effective_device(...) utility to aid in determining the effective device of models that are partially loaded. · Updated 2025-01-01 02:55:27 +08:00	5 9		ZIP  TAR.GZ
ryan/model-offload-3-smart-lora-patcher 6d7314ac0a · Consolidate the LayerPatching patching modes into a single implementation. · Updated 2024-12-24 23:57:54 +08:00	1 7		ZIP  TAR.GZ
ryan/model-offload-prototype-lora-bnb-autocast-tensor 510ed6ed1f · Make CachedModelWithPartialLoad work with models that have non-persistent buffers. · Updated 2024-12-23 23:46:37 +08:00	59 65		ZIP  TAR.GZ
psyche/experiment/typegen-ci 582d67b907 · experiment: fix types · Updated 2024-12-20 06:59:17 +08:00	37 3		ZIP  TAR.GZ
ryan/model-offload-prototype-lora-bnb f01e41ceaf · First pass at dynamically calculating the working memory requirements for the VAE decoding operation. Still need to tune SD3 and FLUX. · Updated 2024-12-20 04:26:16 +08:00	59 57		ZIP  TAR.GZ
ryan/lora-refactor 3ed6e65a6e · Enable LoRAPatcher.apply_smart_lora_patches(...) throughout the stack. · Updated 2024-12-13 06:41:50 +08:00	121 15		ZIP  TAR.GZ
ryan/lora-refactor-quantized-flux-tools 5422bb74c6 · ruff · Updated 2024-12-13 06:28:44 +08:00	121 5		ZIP  TAR.GZ
ryan/model-offload-prototype-lora-alternative-autocast-impl f109914eb3 · WIP - messing around with some alternative autocast implementations · Updated 2024-12-11 10:58:56 +08:00	145 38		ZIP  TAR.GZ
psyche/feat/app/custom-build-inv-ctx a1a3e60431 · feat(app): process accepts custom invocation context builder · Updated 2024-12-11 06:54:02 +08:00	121 1		ZIP  TAR.GZ
ryan/model-offload-prototype-lora f6045682c0 · Fix bug with partial offload of model buffers. · Updated 2024-12-11 06:19:17 +08:00	145 37		ZIP  TAR.GZ
ryan/model-offload-prototype 2144d21f80 · Maintain a read-only CPU state dict copy in CachedModelWithPartialLoad. · Updated 2024-12-07 05:49:24 +08:00	145 23		ZIP  TAR.GZ
ryan/model-offload-prototype-checkpoint1 987393853c · Create CachedModelOnlyFullLoad class. · Updated 2024-12-06 02:43:50 +08:00	145 16		ZIP  TAR.GZ
lstein/feat/diffusers-v0.30 a81acacb2a · fix ruff error · Updated 2024-12-03 08:37:51 +08:00	138 19		ZIP  TAR.GZ
ryan/flux-dynamic-seq-len 8d04ec3f95 · Improve docs related to dynamic T5 sequence length selection. · Updated 2024-11-30 00:11:51 +08:00	171 2		ZIP  TAR.GZ
ryan/flux-min-t5-seq-len 437d1087a2 · Dynamically select smaller t5 seq len to save inference time. · Updated 2024-11-29 08:15:32 +08:00	201 4		ZIP  TAR.GZ
ryan/fix-depth-anything-postprocessing e22f0f2203 · Update DepthAnything post-processing logic to avoid artifacts caused by numerical overflow. · Updated 2024-11-27 22:54:30 +08:00	203 1		ZIP  TAR.GZ
psyche/feat/ui/more-batch-types 64e5c6add7 · feat: more batch types (wip) · Updated 2024-11-27 11:01:07 +08:00	211 1		ZIP  TAR.GZ