Text-to-Image
Diffusers
template:sd-lora
flux
lora
rvorias commited on
Commit
5bfe7dd
·
verified ·
1 Parent(s): ee77966

Upload folder using huggingface_hub

Browse files
POV.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b8c421c9aca62193ead3a7ff3eb3a75795b10fca8e36b112967c966440533319
3
+ size 451202688
POV_000001500.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f52ce469b8622269349f23b768f20452033b40f908f24a77eb47e5529747660d
3
+ size 451202688
POV_000002000.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4cbdc485731997bc984808d8fd1c180559cf72c6c4d64ddb91f8c65e07ef83c0
3
+ size 451202688
POV_000002500.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e771aa9f6168322b28cb8c80c11e01e7ab0abf5594af4cac72c60b28bd7d14e7
3
+ size 451202688
README.md ADDED
@@ -0,0 +1,43 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ tags:
3
+ - diffusers
4
+ - text-to-image
5
+ - template:sd-lora
6
+ - base_model:black-forest-labs/FLUX.1-dev
7
+ - base_model:finetune:black-forest-labs/FLUX.1-dev
8
+ - license:other
9
+ - region:us
10
+ - flux
11
+ - lora
12
+ widget:
13
+ - output:
14
+ url: samples/1731033799883__000003000_0.jpg
15
+ text: wounded centaur, mythical creature pov
16
+ - output:
17
+ url: samples/1731033823464__000003000_1.jpg
18
+ text: ruins of athens, snake pov
19
+ - output:
20
+ url: samples/1731033847315__000003000_2.jpg
21
+ text: silver vampire sword pov
22
+ - output:
23
+ url: samples/1731033870885__000003000_3.jpg
24
+ text: man climbs the everast pov
25
+ base_model: black-forest-labs/FLUX.1-dev
26
+ trigger: pov
27
+ instance_prompt: pov
28
+ license: other
29
+ license_name: flux-1-dev-non-commercial-license
30
+ license_link: https://huggingface.co/black-forest-labs/FLUX.1-dev/blob/main/LICENSE.md
31
+ ---
32
+ # POV
33
+ Model trained with [AI Toolkit by Ostris](https://github.com/ostris/ai-toolkit) under the [Glif Loradex program](https://huggingface.co/glif-loradex-trainer) by [Glif](https://glif.app) user `bulbul`.
34
+ <Gallery />
35
+ ## Trigger words
36
+ You should use `pov` to trigger the image generation.
37
+ ## Download model
38
+ Weights for this model are available in Safetensors format.
39
+ [Download](/glif-loradex-trainer/bulbul_POV/tree/main) them in the Files & versions tab.
40
+
41
+ ## License
42
+ This model is licensed under the [flux-1-dev-non-commercial-license](https://huggingface.co/black-forest-labs/FLUX.1-dev/blob/main/LICENSE.md).
43
+
config.yaml ADDED
@@ -0,0 +1,69 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ config:
2
+ name: POV
3
+ process:
4
+ - datasets:
5
+ - cache_latents_to_disk: true
6
+ caption_dropout_rate: 0.2
7
+ caption_ext: txt
8
+ folder_path: /root/lorahub/POV/dataset
9
+ resolution:
10
+ - 512
11
+ - 768
12
+ - 1024
13
+ shuffle_tokens: true
14
+ token_dropout_rate: 0.01
15
+ device: cuda:0
16
+ model:
17
+ is_flux: true
18
+ name_or_path: black-forest-labs/FLUX.1-dev
19
+ quantize: true
20
+ text_encoder_bits: 8
21
+ network:
22
+ linear: 42
23
+ linear_alpha: 42
24
+ transformer_only: true
25
+ type: lora
26
+ performance_log_every: 500
27
+ sample:
28
+ height: 1024
29
+ neg: ''
30
+ prompts:
31
+ - wounded centaur, mythical creature [trigger]
32
+ - ruins of athens, snake [trigger]
33
+ - silver vampire sword [trigger]
34
+ - man climbs the everast [trigger]
35
+ sample_every: 500
36
+ sample_steps: 25
37
+ sampler: flowmatch
38
+ seed: 42
39
+ walk_seed: true
40
+ width: 1024
41
+ save:
42
+ dtype: float16
43
+ max_step_saves_to_keep: 3
44
+ save_every: 500
45
+ save_format: diffusers
46
+ train:
47
+ batch_size: 1
48
+ dtype: bf16
49
+ ema_config:
50
+ ema_decay: 0.99
51
+ use_ema: true
52
+ gradient_accumulation_steps: 1
53
+ gradient_checkpointing: true
54
+ linear_timesteps: true
55
+ loss_type: mse
56
+ lr: 0.0003
57
+ noise_scheduler: flowmatch
58
+ optimizer: adamw8bit
59
+ reg_weight: 1.0
60
+ steps: 3000
61
+ target_noise_multiplier: 1.0
62
+ train_text_encoder: false
63
+ train_unet: true
64
+ training_folder: /root/lorahub
65
+ trigger_word: pov
66
+ type: sd_trainer
67
+ job: extension
68
+ meta:
69
+ description: ''
samples/1731027628439__000000000_0.jpg ADDED
samples/1731027652209__000000000_1.jpg ADDED
samples/1731027675688__000000000_2.jpg ADDED
samples/1731027699669__000000000_3.jpg ADDED
samples/1731028639999__000000500_0.jpg ADDED
samples/1731028663572__000000500_1.jpg ADDED
samples/1731028687152__000000500_2.jpg ADDED
samples/1731028710745__000000500_3.jpg ADDED
samples/1731029652344__000001000_0.jpg ADDED
samples/1731029676097__000001000_1.jpg ADDED
samples/1731029699678__000001000_2.jpg ADDED
samples/1731029723361__000001000_3.jpg ADDED
samples/1731030704826__000001500_0.jpg ADDED
samples/1731030728402__000001500_1.jpg ADDED
samples/1731030752014__000001500_2.jpg ADDED
samples/1731030775625__000001500_3.jpg ADDED
samples/1731031726092__000002000_0.jpg ADDED
samples/1731031749659__000002000_1.jpg ADDED
samples/1731031773232__000002000_2.jpg ADDED
samples/1731031796814__000002000_3.jpg ADDED
samples/1731032763872__000002500_0.jpg ADDED
samples/1731032787441__000002500_1.jpg ADDED
samples/1731032811245__000002500_2.jpg ADDED
samples/1731032834841__000002500_3.jpg ADDED
samples/1731033799883__000003000_0.jpg ADDED
samples/1731033823464__000003000_1.jpg ADDED
samples/1731033847315__000003000_2.jpg ADDED
samples/1731033870885__000003000_3.jpg ADDED