E-Anlia commited on
Commit
ae89acd
Β·
1 Parent(s): 811ab55

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +13 -3
README.md CHANGED
@@ -4,22 +4,32 @@ We have achieved some initial promising experimental results, but it is still no
4
 
5
  If you would like to test it, please contact the NewBieAi administrator. We may grant you access and testing permission after evaluating your situation.
6
 
7
- Basic Information:
8
- Anime Type
 
 
9
  >Text Encoder:
10
  Google/Gemma3-4b-it
11
  Jina Ai/Jina Clip v2
12
  Total: approximately 5b
 
13
  >Networks:
14
  Next-DIT (modified from 26 layers to 36 layers) 3.5b
 
15
  >VAE:
16
  Flux 1 Dev-VAE
 
17
  Pretrain Dataset: full dan + 1m e621
 
18
  Use 8*h200 train for four months (23000 h200 hour)
19
- β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”
20
  Restructure text data use XML format
 
 
21
  The model is currently in the train phase (60% complete).
 
22
  Overall progress is approximately 80%.
 
23
  The model is expected to be open sourced on Hugging Face on December 31, 2025.
24
 
25
  example:
 
4
 
5
  If you would like to test it, please contact the NewBieAi administrator. We may grant you access and testing permission after evaluating your situation.
6
 
7
+ >Basic Information:
8
+
9
+ >Anime Type
10
+
11
  >Text Encoder:
12
  Google/Gemma3-4b-it
13
  Jina Ai/Jina Clip v2
14
  Total: approximately 5b
15
+
16
  >Networks:
17
  Next-DIT (modified from 26 layers to 36 layers) 3.5b
18
+
19
  >VAE:
20
  Flux 1 Dev-VAE
21
+
22
  Pretrain Dataset: full dan + 1m e621
23
+
24
  Use 8*h200 train for four months (23000 h200 hour)
25
+
26
  Restructure text data use XML format
27
+
28
+
29
  The model is currently in the train phase (60% complete).
30
+
31
  Overall progress is approximately 80%.
32
+
33
  The model is expected to be open sourced on Hugging Face on December 31, 2025.
34
 
35
  example: