dzft3w commited on
Commit
bf3b9c4
1 Parent(s): b6335be

Model card auto-generated by SimpleTuner

Browse files
Files changed (1) hide show
  1. README.md +117 -0
README.md ADDED
@@ -0,0 +1,117 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: creativeml-openrail-m
3
+ base_model: "black-forest-labs/FLUX.1-dev"
4
+ tags:
5
+ - stable-diffusion
6
+ - stable-diffusion-diffusers
7
+ - text-to-image
8
+ - diffusers
9
+ - simpletuner
10
+ - lora
11
+ - template:sd-lora
12
+ inference: true
13
+ widget:
14
+ - text: 'unconditional (blank prompt)'
15
+ parameters:
16
+ negative_prompt: 'blurry, cropped, ugly'
17
+ output:
18
+ url: ./assets/image_0_0.png
19
+ - text: 'breasts out photo of woman wearing blouse at beach'
20
+ parameters:
21
+ negative_prompt: 'blurry, cropped, ugly'
22
+ output:
23
+ url: ./assets/image_1_0.png
24
+ ---
25
+
26
+ # flux-training-2BoutOvalFlux1
27
+
28
+ This is a LoRA derived from [black-forest-labs/FLUX.1-dev](https://huggingface.co/black-forest-labs/FLUX.1-dev).
29
+
30
+
31
+
32
+ The main validation prompt used during training was:
33
+
34
+
35
+
36
+ ```
37
+ breasts out photo of woman wearing blouse at beach
38
+ ```
39
+
40
+ ## Validation settings
41
+ - CFG: `3.5`
42
+ - CFG Rescale: `0.0`
43
+ - Steps: `15`
44
+ - Sampler: `None`
45
+ - Seed: `1`
46
+ - Resolution: `1024`
47
+
48
+ Note: The validation settings are not necessarily the same as the [training settings](#training-settings).
49
+
50
+ You can find some example images in the following gallery:
51
+
52
+
53
+ <Gallery />
54
+
55
+ The text encoder **was not** trained.
56
+ You may reuse the base model text encoder for inference.
57
+
58
+
59
+ ## Training settings
60
+
61
+ - Training epochs: 9
62
+ - Training steps: 100
63
+ - Learning rate: 0.0001
64
+ - Effective batch size: 6
65
+ - Micro-batch size: 6
66
+ - Gradient accumulation steps: 1
67
+ - Number of GPUs: 1
68
+ - Prediction type: flow-matching
69
+ - Rescaled betas zero SNR: False
70
+ - Optimizer: AdamW, stochastic bf16
71
+ - Precision: Pure BF16
72
+ - Xformers: Enabled
73
+ - LoRA Rank: 64
74
+ - LoRA Alpha: None
75
+ - LoRA Dropout: 0.1
76
+ - LoRA initialisation style: default
77
+
78
+
79
+ ## Datasets
80
+
81
+ ### 2BoutOvalFlux1
82
+ - Repeats: 0
83
+ - Total number of images: 66
84
+ - Total number of aspect buckets: 1
85
+ - Resolution: 512 px
86
+ - Cropped: False
87
+ - Crop style: None
88
+ - Crop aspect: None
89
+
90
+
91
+ ## Inference
92
+
93
+
94
+ ```python
95
+ import torch
96
+ from diffusers import DiffusionPipeline
97
+
98
+ model_id = 'black-forest-labs/FLUX.1-dev'
99
+ adapter_id = 'dzft3w/flux-training-2BoutOvalFlux1'
100
+ pipeline = DiffusionPipeline.from_pretrained(model_id)
101
+ pipeline.load_lora_weights(adapter_id)
102
+
103
+ prompt = "breasts out photo of woman wearing blouse at beach"
104
+
105
+
106
+ pipeline.to('cuda' if torch.cuda.is_available() else 'mps' if torch.backends.mps.is_available() else 'cpu')
107
+ image = pipeline(
108
+ prompt=prompt,
109
+ num_inference_steps=15,
110
+ generator=torch.Generator(device='cuda' if torch.cuda.is_available() else 'mps' if torch.backends.mps.is_available() else 'cpu').manual_seed(1641421826),
111
+ width=1024,
112
+ height=1024,
113
+ guidance_scale=3.5,
114
+ ).images[0]
115
+ image.save("output.png", format="PNG")
116
+ ```
117
+