File size: 3,228 Bytes
c2a6fbc
 
 
 
 
 
 
 
 
 
 
581ce44
 
0b574f5
 
 
c2a6fbc
 
 
4565d23
c2a6fbc
5e6a69d
a07016a
 
1e8e3e4
44338a7
186c280
 
 
 
1e8e3e4
a07016a
f436e3f
 
186c280
44338a7
f436e3f
186c280
f436e3f
44338a7
c2a6fbc
44338a7
 
c2a6fbc
44338a7
 
 
 
 
 
0b574f5
 
44338a7
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
---
base_model: black-forest-labs/FLUX.1-dev
---

*Note that all these models are derivatives of black-forest-labs/FLUX.1-dev and therefore covered by the 
[FLUX.1 [dev] Non-Commercial License](https://huggingface.co/black-forest-labs/FLUX.1-dev/blob/main/LICENSE.md) license.*

*Some models are derivatives of finetunes, and are included with the permission of the finetuner*

# Optimised Flux GGUF models

A collection of GGUF models using mixed quantization (different layers quantized to different precision to optimise fidelity v. memory),
created using [mixed gguf converter](https://github.com/chrisgoringe/mixed-gguf-converter).

They can be loaded in ComfyUI using the [ComfyUI GGUF Nodes](https://github.com/city96/ComfyUI-GGUF). Just put the gguf files in your
models/unet directory.

## Naming convention (mx for 'mixed')

[original_model_name]_mxN_N.gguf

where N_N is the average number of bits per parameter.

## Good choices to start with
```
-  3_1 is the smallest yet - might work on 6 GB? 
-  3_8 might work on a 8 GB card
-  6_9 should be good for a 12 GB card
-  8_2 is a good choice for 16 GB cards if you want to add LoRAs etc
-  9_2 fits on a 16 GB card
```

## Speed?

On an A40 (plenty of VRAM), everything except the model identical, 
the time taken to generate an image (30 steps, deis sampler) was about 65% longer than for the full model (45s v 27s).

Quantised models will generally be slower because the weights have to be converted back into a native torch form when they are needed.

## How are these 'optimised'?

The optimization is based on a cost metric, representing the error introduced by quantizing a specified layer with a specified quant.
The data can be found [here](https://github.com/chrisgoringe/mixed-gguf-converter/tree/main/costs), and details of the process are below.

From this, any possible quantization can be given a cost and a benefit (bits saved). The possible quantizations are then sorted from
best (benefit/cost) to worst, and applied in order, until the required number of bits have been removed.

### Calculating costs

I created a database of the hidden states at the start and end of the transformer stack as follows:
- 240 prompts used for flux images popular at civit.ai were run through the full Flux.1-dev model with randomised resolution and step count.
- For a randomly selected step in the inference, the hidden states before and after the layer stack were captured.

To calculate the cost of quantizing a specific layer to a specific quant:
- A single layer in the transformer stack was quantized
- The 240 initial hidden states were run through the stack
- The cost is defined as the mean square difference between the outputs of the modified stack and the unmodified stack

The cost, therefore, is a measure of how much change is introduced into the output hidden states by the quantization.

## Not quantized

In all these models, the 'in' blocks, the final layer blocks, and all normalization scale parameters are not quantized. 
These represent of 0.54% of all parameters in the model.

In patch models (where the states were quantised using llama.cpp code), the biases are also not quantized. 
These represent 0.03% of all parameters in the model.