ozgurkara commited on
Commit
ca869e7
·
0 Parent(s):

BudgetPix-pMF: release

Browse files
.gitattributes ADDED
@@ -0,0 +1,35 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ *.7z filter=lfs diff=lfs merge=lfs -text
2
+ *.arrow filter=lfs diff=lfs merge=lfs -text
3
+ *.bin filter=lfs diff=lfs merge=lfs -text
4
+ *.bz2 filter=lfs diff=lfs merge=lfs -text
5
+ *.ckpt filter=lfs diff=lfs merge=lfs -text
6
+ *.ftz filter=lfs diff=lfs merge=lfs -text
7
+ *.gz filter=lfs diff=lfs merge=lfs -text
8
+ *.h5 filter=lfs diff=lfs merge=lfs -text
9
+ *.joblib filter=lfs diff=lfs merge=lfs -text
10
+ *.lfs.* filter=lfs diff=lfs merge=lfs -text
11
+ *.mlmodel filter=lfs diff=lfs merge=lfs -text
12
+ *.model filter=lfs diff=lfs merge=lfs -text
13
+ *.msgpack filter=lfs diff=lfs merge=lfs -text
14
+ *.npy filter=lfs diff=lfs merge=lfs -text
15
+ *.npz filter=lfs diff=lfs merge=lfs -text
16
+ *.onnx filter=lfs diff=lfs merge=lfs -text
17
+ *.ot filter=lfs diff=lfs merge=lfs -text
18
+ *.parquet filter=lfs diff=lfs merge=lfs -text
19
+ *.pb filter=lfs diff=lfs merge=lfs -text
20
+ *.pickle filter=lfs diff=lfs merge=lfs -text
21
+ *.pkl filter=lfs diff=lfs merge=lfs -text
22
+ *.pt filter=lfs diff=lfs merge=lfs -text
23
+ *.pth filter=lfs diff=lfs merge=lfs -text
24
+ *.rar filter=lfs diff=lfs merge=lfs -text
25
+ *.safetensors filter=lfs diff=lfs merge=lfs -text
26
+ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
27
+ *.tar.* filter=lfs diff=lfs merge=lfs -text
28
+ *.tar filter=lfs diff=lfs merge=lfs -text
29
+ *.tflite filter=lfs diff=lfs merge=lfs -text
30
+ *.tgz filter=lfs diff=lfs merge=lfs -text
31
+ *.wasm filter=lfs diff=lfs merge=lfs -text
32
+ *.xz filter=lfs diff=lfs merge=lfs -text
33
+ *.zip filter=lfs diff=lfs merge=lfs -text
34
+ *.zst filter=lfs diff=lfs merge=lfs -text
35
+ *tfevents* filter=lfs diff=lfs merge=lfs -text
README.md ADDED
@@ -0,0 +1,80 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: mit
3
+ base_model: Lyy0725/pMF
4
+ base_model_relation: finetune
5
+ pipeline_tag: unconditional-image-generation
6
+ tags:
7
+ - budgetany
8
+ - budgetpix
9
+ - adaptive-tokenization
10
+ - diffusion
11
+ - class-conditional
12
+ - imagenet
13
+ - one-step
14
+ - meanflow
15
+ ---
16
+
17
+ # BudgetPix-pMF
18
+
19
+ 🌐 [Project page](https://budgetany.github.io/budgetpix/)  |  💻 [Code](https://github.com/BudgetAny/BudgetPix/tree/main/pmf)  |  📄 [Paper](https://arxiv.org/abs/2610.12307)  |  🧩 [BudgetAny](https://huggingface.co/BudgetAny)
20
+
21
+ **One checkpoint, any token budget.** One-step class-conditional ImageNet generation in pixel space with [pixel MeanFlow (pMF)](https://github.com/Lyy-iiis/pMF), fine-tuned with BudgetPix compute-adaptive tokenization. At inference you choose how many tokens the transformer runs on: an entropy-based detail map merges the flattest regions into coarse tokens first and gives the fine tokens to the most detailed ones; the layout is chosen up front, by default from the model's own dense sample of the same class and noise. The same weights serve every budget.
22
+
23
+ > **Unofficial, open-source implementation**, taken from [budgetpix/BudgetPix-public](https://github.com/budgetpix/BudgetPix-public):
24
+ > a public reimplementation of BudgetPix: compute-adaptive tokenization for pixel-space image diffusion (JiT, MiniT2I,
25
+ > PixelDiT, pMF). This model is part of [BudgetAny](https://huggingface.co/BudgetAny), a community reimplementation of
26
+ > [BudgetPix](https://arxiv.org/abs/2610.12307) (Kara et al., 2026). It is not an official release.
27
+ > Thanks to the BudgetPix authors for the method.
28
+
29
+ ## Files
30
+
31
+ | file | base model | dense tokens | parameters | size | SHA-256 |
32
+ |---|---|---|---|---|---|
33
+ | `budgetpix_pmf_l16_256.pt` | pMF-L/16 (one-step), 256 px | 256 | 517 M | 2.07 GB | `dec78c17f667eefa…` |
34
+
35
+ `SHA256SUMS` lists the full checksums. Also in the repository: `pmf_released_L16_full_uv.pt` and `pmf_released_B16_full_uv.pt`, the released pMF weights converted to torch with the auxiliary v-branch.
36
+
37
+ ## Usage
38
+
39
+ ```bash
40
+ git clone https://github.com/BudgetAny/BudgetPix && cd BudgetPix/pmf # setup: see README.md
41
+ hf download BudgetAny/BudgetPix-pMF budgetpix_pmf_l16_256.pt --local-dir weights
42
+ python sample.py --ckpt weights/budgetpix_pmf_l16_256.pt --budget 128 --classes 207,88,417,933
43
+ ```
44
+
45
+ `--budget` is the number of tokens per image (256 is the dense grid).
46
+
47
+ ## Citation
48
+
49
+ <p align="center">
50
+ <a href="https://arxiv.org/abs/2610.12307"><b>BudgetPix: Compute-Adaptive Tokenization for Pixel-Space Image Diffusion</b></a>
51
+ </p>
52
+
53
+ <p align="center">
54
+ <a href="https://karaozgur.com/">Ozgur Kara</a><sup>1,†</sup>,
55
+ <a href="https://scholar.google.com/citations?user=Dv5gB7cAAAAJ&hl=en">Yujia Chen</a><sup>2</sup>,
56
+ <a href="https://scholar.google.com/citations?user=ve9A9rMAAAAJ&hl=en">Daniel Watson</a><sup>2</sup>,
57
+ <a href="https://scholar.google.com/citations?user=5H0arvkAAAAJ&hl=en">David Forsyth</a><sup>1</sup>,
58
+ <a href="https://scholar.google.com/citations?user=8kA3eDwAAAAJ&hl=en">James Matthew Rehg</a><sup>1</sup>,
59
+ <a href="https://scholar.google.com/citations?user=R-OrlSgAAAAJ&hl=en">Wen-Sheng Chu</a><sup>2,‡</sup>,
60
+ <a href="https://scholar.google.com/citations?user=3TVbh8QAAAAJ&hl=en">Du Tran</a><sup>2,‡</sup><br>
61
+ <sup>1</sup>University of Illinois Urbana-Champaign &nbsp;&nbsp; <sup>2</sup>Google<br>
62
+ <sub><sup>†</sup>Work done during an internship at Google. &nbsp; <sup>‡</sup>Joint last authors.</sub>
63
+ </p>
64
+
65
+ If you use this model or the code, please cite the BudgetPix paper:
66
+
67
+ ```bibtex
68
+ @article{kara2026budgetpix,
69
+ title={BudgetPix: Compute-Adaptive Tokenization for Pixel-Space Image Diffusion},
70
+ author={Ozgur Kara and Yujia Chen and Daniel Watson and David Forsyth and James Matthew Rehg and Wen-Sheng Chu and Du Tran},
71
+ journal={arXiv preprint arXiv:2610.12307},
72
+ year={2026}
73
+ }
74
+ ```
75
+
76
+ Built on [pixel MeanFlow](https://github.com/Lyy-iiis/pMF) and [APT](https://github.com/rccchoudhury/apt).
77
+
78
+ ## License
79
+
80
+ MIT, like the pMF weights it derives from.
SHA256SUMS ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ dec78c17f667eefa91324a6c728e456d54a90937fc7f8ae9b89a5ce6c941cc7a budgetpix_pmf_l16_256.pt
2
+ 948d7e88397a721f4744e89f2f6d2f6a37b5c9a2afa3ee064aeb0124a9ed09d4 pmf_released_B16_full_uv.pt
3
+ 0f75b6a4a795074c633da2ee03776d0e2baf5ea871d0c87be43f401b0f1e30f9 pmf_released_L16_full_uv.pt
budgetpix_pmf_l16_256.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:dec78c17f667eefa91324a6c728e456d54a90937fc7f8ae9b89a5ce6c941cc7a
3
+ size 2066526629
pmf_released_B16_full_uv.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:948d7e88397a721f4744e89f2f6d2f6a37b5c9a2afa3ee064aeb0124a9ed09d4
3
+ size 702015029
pmf_released_L16_full_uv.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0f75b6a4a795074c633da2ee03776d0e2baf5ea871d0c87be43f401b0f1e30f9
3
+ size 2047414187