Total runs:
33.7K
Run Growth:
-1.1K
Growth Rate:
-3.30%
Sponsored by
PoYo.AI
ostris / accuracy_recovery_adapters
Model's Last Updated: January 02 2026
Introduction of accuracy_recovery_adapters
Model Details of accuracy_recovery_adapters
Accuracy Recovery Adapters
This repo contains various accuracy recovery adapters (ARAs) that I have trained, primarialy for use with AI Toolkit . An ARA is a LoRA that is trained via student teacher training with the student being quantized down to a low precision and the teacher having a high precision. The goal is to have a side chain LoRA, at bfloat16, that runs parallel to highly quantized layers in a network to compensate for the loss in precision that happens when weights are quantized. The training is done on a per layer basis in order to match the parent output as much as possible.
While this can be used on inference, my primary goal is to make large models finetunable on consumer grade hardware. With the 3bit Qwen Image adapter, it is now possible to train a LoRA on top of it, with 1 MP images, on a 24 GB GPU, such as a 3090/4090.
I have found the sweet spot, at least for Qwen-Image , is 3 bit quantization with a rank 16 adapter.
More info, examples, links, training scripts, AI Toolkit example configs, and adapters to some soon.
License
All adapters inherit the parent model license. Apache 2.0 for Apache 2.0, BFL License for BFL License, etc.
Qwen-Image 3 bit quantization
Runs of ostris accuracy_recovery_adapters on huggingface.co
0
Total runs
0
24-hour runs
0
3-day runs
0
7-day runs
0
30-day runs
More Information About accuracy_recovery_adapters huggingface.co Model
Url of accuracy_recovery_adapters
accuracy_recovery_adapters huggingface.co Url
Provider of accuracy_recovery_adapters huggingface.co
Other API from ostris
Updated:December 01 2025
Total runs:
17.6K
Run Growth:
-7.3K
Growth Rate:
-41.70%
Total runs:
13.9K
Run Growth:
132
Growth Rate:
0.95%
Total runs:
4.6K
Run Growth:
-14.5K
Growth Rate:
-279.23%
Total runs:
4.3K
Run Growth:
-2.2K
Growth Rate:
-50.66%
Total runs:
4.0K
Run Growth:
868
Growth Rate:
21.59%
Total runs:
3.1K
Run Growth:
-1.7K
Growth Rate:
-55.27%
Total runs:
2.6K
Run Growth:
-4.4K
Growth Rate:
-164.15%
Total runs:
2.4K
Run Growth:
-319
Growth Rate:
-13.47%
Total runs:
2.2K
Run Growth:
129
Growth Rate:
5.97%
Total runs:
1.4K
Run Growth:
572
Growth Rate:
41.15%
Total runs:
631
Run Growth:
631
Growth Rate:
100.00%
Total runs:
582
Run Growth:
-137
Growth Rate:
-23.91%
Total runs:
533
Run Growth:
5
Growth Rate:
0.94%
Total runs:
339
Run Growth:
122
Growth Rate:
35.16%
Total runs:
254
Run Growth:
114
Growth Rate:
44.88%
Total runs:
239
Run Growth:
20
Growth Rate:
8.37%
Total runs:
193
Run Growth:
86
Growth Rate:
43.88%
Total runs:
161
Run Growth:
-35
Growth Rate:
-21.34%
Total runs:
132
Run Growth:
40
Growth Rate:
31.25%
Total runs:
126
Run Growth:
-6
Growth Rate:
-4.76%
Total runs:
104
Run Growth:
19
Growth Rate:
18.27%
Total runs:
102
Run Growth:
21
Growth Rate:
21.00%
Total runs:
95
Run Growth:
-4
Growth Rate:
-4.21%
Total runs:
69
Run Growth:
37
Growth Rate:
55.22%
Total runs:
63
Run Growth:
29
Growth Rate:
46.77%
Total runs:
57
Run Growth:
14
Growth Rate:
25.00%
Total runs:
54
Run Growth:
40
Growth Rate:
74.07%
Total runs:
52
Run Growth:
19
Growth Rate:
37.25%
Total runs:
50
Run Growth:
20
Growth Rate:
37.74%
Total runs:
37
Run Growth:
23
Growth Rate:
62.16%
Total runs:
33
Run Growth:
-6
Growth Rate:
-18.75%
Total runs:
24
Run Growth:
-4
Growth Rate:
-16.67%
Total runs:
11
Run Growth:
-6
Growth Rate:
-54.55%
Total runs:
9
Run Growth:
-386
Growth Rate:
-4288.89%
Total runs:
0
Run Growth:
0
Growth Rate:
0.00%
Total runs:
0
Run Growth:
0
Growth Rate:
0.00%
Total runs:
0
Run Growth:
0
Growth Rate:
0.00%
Total runs:
0
Run Growth:
0
Growth Rate:
0.00%