Skip to content

conflicted merge - #4

Open
Geramy wants to merge 289 commits into
lemonade-sdk:lemonadefrom
leejet:master
Open

conflicted merge#4
Geramy wants to merge 289 commits into
lemonade-sdk:lemonadefrom
leejet:master

Conversation

@Geramy

@Geramy Geramy commented May 6, 2026

Copy link
Copy Markdown
Member

No description provided.

wbruna and others added 30 commits April 29, 2026 23:26
@wbruna @stduhpf
@Cyberhan123 @leejet
feat: transition from compile-time to runtime backend discovery (#1448)
b8079e2 
Co-authored-by: Stéphane du Hamel <stephduh@live.fr>
Co-authored-by: Cyberhan123 <255542417@qq.com>
Co-authored-by: leejet <leejet714@gmail.com>
@leejet
perf: speed up tensor_to_sd_image conversion (#1466)
3d6064b
@fszontagh
fix: skip empty MultiLoraAdapter when no LoRAs target a model (#1469)
9097ce5
@wbruna
feat: adapt res samplers for flow models for eta > 0 (#1436)
586b6f1
@leejet
feat: add max-vram based segmented param offload (#1476)
90e87bc
@cphlipot
feat: display server url with "http://" prefix. (#1486)
6ee0684
@l8bloom
docs: add new go bindings for stable-diffusion.cpp (#1480)
60477fd
@daniandtheweb
feat: add Euler CFG++ and Euler-A CFG++ samplers (#1354)
9d68341
@wbruna @pwilkin
@junmo-kim @leejet
feat: support for memory-mapping model weights (#1414)
57ff2eb 
Co-authored-by: Piotr Wilkin <piotr.wilkin@syndatis.com>
Co-authored-by: Junmo Kim <me@junmo.kim>
Co-authored-by: leejet <leejet714@gmail.com>
@candrews
fix: Use PkgConfig for WebP and WebM (#1400)
eeac950
@leejet
feat: add hidream o1 image support (#1485)
0665a7f
@leejet
docs: add CONTRIBUTING.md
381e0df
@leejet
docs: add .github/pull_request_template.md
0b82969
@wbruna
chore: do not report the fake VAE "allocation" as an error (#1494)
686856e
@cphlipot
fix: Fix broken GCC 16 build (enforce C11/C++17 compile ) (#1478)
db08b84
@wbruna
refactor: unify Euler, Euler Ancestral and DDIM implementations (#1474)
fd1a279
@leejet
feat: auto-detect max VRAM budget with --max-vram -1 (#1498)
38b14ad
@superm1
ci: switch over ROCm builds to artifacts both for stable and preview …
839f6a9 
…releases (#1281)
@superm1
ci: update ROCm Windows builds (#1282)
0c1ca17
@leejet @stduhpf
feat: add module backend assignment support (#1500)
3633072 
Co-authored-by: Stéphane du Hamel <stephduh@live.fr>
@leejet
fix: avoid repeated T5 EOS tokens in Anima prompt weights (#1501)
d7ecbe1
@kwsp
docs: update zit example to 8 steps (#1294)
bd17f53
@superm1
ci: update ROCm builds for Windows and Linux to use ROCm 7.13 (#1504)
3b4d26f
@leejet
feat: add ltx2.3 support (#1463)
67dda3f 
* add GemmaTokenizer

* add basic ltx2.3 support

* change vocab file encoding

* fix ci

* fix ubuntu build

* add temporal tiling support

* add ltx audio support

* update ggml submodule url

* fix generate_video

* add i2v support

* minify bundled Gemma tokenizer vocab sources

* pass video fps into temporal rope embeddings

* fix av_ca_timestep_scale_multiplier

* add LTX2Scheduler support

* update docs

* fix ci
@stduhpf @leejet
fix: handle stereo format in sd_audio (#1489)
cde20d5 
Co-authored-by: leejet <leejet714@gmail.com>
@stduhpf
feat: add ltxav latent2rgb projection matrix (#1502)
06accf2
@leejet
feat: add ltx2.3 flf2v support (#1505)
e43b24c
@leejet
refactor: split guidance composition (#1506)
50134e5
@wbruna
feat: add Gradient Estimation sampler (#1484)
e7eb92f
@leejet
refactor: minify vocab files (#1509)
baf7eda
leejet and others added 30 commits August 4, 2026 23:14
@leejet
feat: add minimax-h3 support (#1854)
ea7f0c8
@stduhpf
feat: trained Minimax VAE Latent2rgb proj (#1856)
bfbef5b
@leejet
fix: map Qwen3-VL DeepStack GGUF tensor names (#1858)
c6beeef
@stduhpf
fix: preserve "token_refiner" token for MiniMax H3 LoRAs (#1864)
b4f1fd6
@danielhanchen
fix: fail with a message when MiniMax-H3 is run in img_gen mode (#1863)
487de75
@leejet
feat: support INT8 ConvRot safetensors (#1857)
bcc7e29
@LostRuins
fix: replace free_compute_buffer with runner_done in vae (#1872)
06c359f
@leejet
sync: update ggml (#1873)
fabe481
@leejet
fix(ci): trigger builds for ggml updates
de298c2
@stduhpf
feat: add taeh3 support (#1874)
6100d83
@fszontagh
fix: prevent gallocr hash overflow in tiny graph-cut segments (#1880)
58b6cb6
@fszontagh
fix: re-clamp streaming VRAM budget to currently free memory (#1878)
1706b32
@wbruna
fix: mark graph cuts with both a prefix and a suffix (#1883)
88b044b
@vmobilis
fix: make max_order of lms sampler configurable (#1885)
760717a
@vmobilis
fix: guard against missing sampler/scheduler names (#1887)
16304cc
@leejet
chore: format code
97d2990
@vmobilis
fix: use sd_get_preview_interval() (#1907)
12ee60d
@vmobilis
feat: configurable image / video compression (#1909)
0a565f2
@leejet
feat: support standard Qwen3-VL weights for MiniMax-H3 (#1910)
50d6405
@leejet
feat: load scaled FP8 weights without upfront conversion (#1913)
be0e344
@leejet
fix: match exact weights in LLM config detection (#1923)
2c92949
@pwilkin @leejet
feat: add LTX-2.5 support (#1893)
afd5306 
Co-authored-by: leejet <leejet714@gmail.com>
@jk212h20
fix: correct MiniMax H3 reference audio encoding (#1886)
c797899
@jk212h20
fix: correct MiniMax H3 audio Euler steps (#1908)
dc4000d
@leejet
feat: use backend-native FP8 matmul when supported (#1916)
2540a4f
@leejet
sync: update ggml
134c821
@vmobilis @leejet
feat: additional --preview-interval values (#1915)
d9b6e27 
Co-authored-by: leejet <leejet714@gmail.com>
@vmobilis
feat: support numbering for preview images (#1895)
9029655
@leejet
fix: use carrier sampling for MiniMax H3 audio (#1924)
40e605f
@leejet
feat: generalize temporal tiling across video VAEs (#1926)
6b3edaa
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.