Hugging Face's logo

glif
/
loradex-flux-dev

Model card Files Community
1
loradex-flux-dev
File size: 2,413 Bytes 
97a4459
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
 
config:
  name: flux_dev_mnemos
  process:
  - datasets:
    - cache_latents_to_disk: true
      caption_dropout_rate: 0.2
      caption_ext: txt
      folder_path: /root/lorahub/flux_dev_mnemos/dataset
      resolution:
      - 512
      - 768
      - 1024
      shuffle_tokens: false
      token_dropout_rate: 0.01
    device: cuda:0
    model:
      is_flux: true
      name_or_path: black-forest-labs/FLUX.1-dev
      quantize: true
      text_encoder_bits: 8
    network:
      linear: 42
      linear_alpha: 42
      transformer_only: true
      type: lora
    performance_log_every: 500
    sample:
      height: 1024
      neg: ''
      prompts:
      - wounded centaur, mythical creature [trigger]
      - ruins of athens, snake [trigger]
      - silver vampire sword [trigger]
      - ethereal forest spirit, glowing eyes [trigger]
      - ancient cosmic map, constellation [trigger]
      - shadow phoenix rising, ash [trigger]
      sample_every: 500
      sample_steps: 25
      sampler: flowmatch
      seed: 42
      walk_seed: true
      width: 1024
    save:
      dtype: float16
      max_step_saves_to_keep: 3
      save_every: 500
      save_format: diffusers
    train:
      batch_size: 1
      dtype: bf16
      ema_config:
        ema_decay: 0.99
        use_ema: true
      gradient_accumulation_steps: 1
      gradient_checkpointing: true
      linear_timesteps: true
      loss_type: mse
      lr: 0.0002
      noise_scheduler: flowmatch
      optimizer: adamw8bit
      reg_weight: 1.0
      steps: 2000
      target_noise_multiplier: 1.0
      train_text_encoder: false
      train_unet: true
    training_folder: /root/lorahub
    trigger_word: perspective of mnemos
    type: sd_trainer
job: extension
meta:
  description: This dataset comprises 49 unique abstract shadow shapes, each tagged
    with distinctive visual descriptors and conceptual associations. The images, derived
    from a creative process involving forest-found objects and human silhouettes,
    represent a fusion of natural forms and imagined landscapes. This LoRA model is
    designed to generate enigmatic, high-contrast compositions that evoke a sense
    of mystery and intrigue, blending elements of nature, mythology, and cosmic imagery.
    Created through a collaborative effort between human creativity and AI analysis,
    it aims to inspire novel visual interpretations and artistic explorations.