Skip to content

conflicted merge - #4

Open
Geramy wants to merge 377 commits into
lemonade-sdk:lemonadefrom
leejet:master
Open

conflicted merge#4
Geramy wants to merge 377 commits into
lemonade-sdk:lemonadefrom
leejet:master

Conversation

@Geramy

@Geramy Geramy commented May 6, 2026

Copy link
Copy Markdown
Member

No description provided.

kenvandine pushed a commit that referenced this pull request Jun 1, 2026
@Phqen1x
Merge pull request #4 from Phqen1x/revert-cuda-artifact-names
2791d9d 
ci: revert CUDA artifact names to original sd-*-bin-* format
wbruna and others added 29 commits June 7, 2026 22:30
@wbruna
chore: rename and reformat gits_noise.inl (#1617)
81abfb2
@leejet
refactor: move photomaker into generation extension (#1618)
2a07540
@leejet
refactor: split model loader from model definitions (#1619)
b3d56d0
@fszontagh
perf: cap planner budget when model dwarfs the streaming budget (#1612)
17a2b4a
@stduhpf
apg: normalize diff_norm calculation by tensor size (#1620)
138da14
@RapidMark
feat: set tensor names on block params (#1622)
19bdfe2
@RapidMark
fix: correct mask shape for masked flash attention (#1625)
1b702a5
@stduhpf
feat: add circular RoPE support for ideogram4 (#1627)
c20769b
@Cyberhan123
feat: add free_sd_images function to manage memory for C API (#1633)
1fb6b22
@Cyberhan123
chore: add script for automatic code formatting (#1636)
1365008
@Detensable
fix: SD3 conditioning crash when clip_l text encoder is missing (#1638)
3a54597
@leejet
refactor: centralize runner weight staging and cleanup (#1644)
563137a
@leejet
refactor: manage upscaler params through model manager (#1645)
9b0fceb
@leejet
refactor: route all runner params through model manager (#1649)
8d4c7af
@leejet
fix: mark LoKR w2_a tensor as applied (#1650)
276025e
@leejet
feat: support disk params backend (#1651)
bdb431a
@leejet
refactor: remove vae_decode_only context flag (#1653)
749186c
@leejet
refactor: route cpu placement through backend specs (#1654)
5db680c
@leejet
docs: replace example option lists with help commands
17d70b9
@leejet
refactor: simplify ControlNet output caching (#1655)
9838264
@stduhpf
feat: add RPC support (#1629)
c2df4e1
@leejet
docs: refresh README guide links
6f00939
@leejet
sync: update ggml (#1656)
517abc7
@leejet
feat: support backend-specific max-vram budgets
bb90bfa
@leejet
fix: allow oversized Vulkan parameter tensors (#1662)
6e66a1a
@RapidMark
feat: add PuLID-Flux identity-injection support (#1595)
93527fd
@leejet
fix: simplify PuLID ID extraction setup (#1664)
146b6cc
@wbruna @leejet
feat: support for cancelling generations (#1124)
5a34bc7 
* feat: support for canceling the ongoing generation

* return partial image batches on cancel

---------

Co-authored-by: leejet <leejet714@gmail.com>
@wbruna
fix: correct conversion from sd_type_t to ggml_type (#1519)
710bc91
leejet and others added 30 commits September 23, 2026 02:23
@leejet
fix: preserve alpha when upscaling RGBA images with ESRGAN (#2029)
c92d73c
@stduhpf
feat: add latent2rgba preview for Qwen-Image 2.1 (#2032)
241518b
@leejet
feat: add configurable conditioning cache for all models (#2034)
e6281b6
@leejet
feat: add Qwen Image 2.1 prefix KV cache (#2035)
2dc7f54
@assouan
fix: add graph cuts for MiniMax-H3 text conditioning (#1900)
3674693
@nanguoyu
perf: run one-frame Wan VAE convolutions as 2D convolutions (#2038)
70c1dbc
@leejet
ci: automatically close PRs from organization-owned forks
2a4ebba
@nanguoyu
fix: map mmapped weights through Metal buffers instead of CPU buffers (…
500ef5f 
…#2037)
@leejet
refactor: centralize circular RoPE and extend image model support (#2039
88411ef 
)
@stduhpf
feat: optimize cfg special cases with guidance schdeule (#2033)
caa111a
@nbeerbower
feat: add a stand-alone upscale endpoint to the server (#2026)
4dfe8f5
@leejet
feat: add configurable Qwen cache types and early cache scheduling (#…
740c7ae 
…2045)
@leejet
fix: reserve 128 MiB headroom when selecting monolithic execution (#2046
1a2330d 
)
@leejet
fix: align Qwen Image 2.1 flow schedule with official defaults (#2048)
b167b94
@leejet
fix: scale Qwen Image 2.1 VAE convolutions (#2054)
0a9340c
@Yi-111-a
fix: keep the server frontend install out of a parent pnpm workspace (#…
5e7b291 
…2056)
@leejet
fix: map Qwen Image 2.1 LoRAs to fused MLP weights (#2057)
510bccf
@leejet
fix: load safetensors index shards without recursion (#2058)
4c3cf75
@losewayy
feat: add PixArt model family support (#2047)
39ada08
@leejet
refactor: define VAE tile dimensions in image pixels (#2059)
19bbbca
@leejet
refactor: align PixArt weights with upstream layout (#2061)
2f88688
@losewayy @leejet
fix: improve GGUF metadata parsing and reader selection (#2062)
168f7b8 
Co-authored-by: leejet <leejet714@gmail.com>
@Yi-111-a
fix: handle filesystem errors during lora and upscaler cache scan (#2064
9947eeb 
)
@akleine
docs: update hugging face links for MageFlow diffusion and VAE models (…
61a83b8 
…#2065)
@leejet
feat: add Ming-Image Design support (#2063)
f8890b9
@leejet
sync: update ggml
47e83d7
@leejet
fix: load INT8 convrot LLM embeddings correctly (#2067)
ede32a6
@leejet
fix: update ggml to fix INT8 convrot backend fallback (#2069)
27f7c43
@leejet
fix: keep scaled INT8 convrot matmuls on Vulkan (#2070)
42ab1c1
@leejet
feat: enable hip INT8 tensorwise matmul and convrot (#2071)
3f8527a
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

No reviews

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.