David Burnett
commited on
Commit
•
4c94ece
1
Parent(s):
6c915cf
Merge LCM SSD-1B unet wuth SSD-1B model
Browse files- .gitignore +5 -0
- README.md +27 -0
- model_index.json +33 -0
- scheduler/scheduler_config.json +18 -0
- text_encoder/config.json +24 -0
- text_encoder/model.fp16.safetensors +3 -0
- text_encoder/model.safetensors +3 -0
- text_encoder_2/config.json +24 -0
- text_encoder_2/model.fp16.safetensors +3 -0
- text_encoder_2/model.safetensors +3 -0
- tokenizer/merges.txt +0 -0
- tokenizer/special_tokens_map.json +24 -0
- tokenizer/tokenizer_config.json +33 -0
- tokenizer/vocab.json +0 -0
- tokenizer_2/merges.txt +0 -0
- tokenizer_2/special_tokens_map.json +24 -0
- tokenizer_2/tokenizer_config.json +33 -0
- tokenizer_2/vocab.json +0 -0
- unet/README.md +74 -0
- unet/config.json +93 -0
- unet/diffusion_pytorch_model.fp16.safetensors +3 -0
- unet/diffusion_pytorch_model.safetensors +3 -0
- unet/download.png +3 -0
- unet/image.png +3 -0
- vae/config.json +31 -0
- vae/diffusion_pytorch_model.fp16.safetensors +3 -0
- vae/diffusion_pytorch_model.safetensors +3 -0
.gitignore
ADDED
@@ -0,0 +1,5 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
.DS_Store
|
2 |
+
._.DS_Store
|
3 |
+
**/.DS_Store
|
4 |
+
**/._.DS_Store
|
5 |
+
|
README.md
CHANGED
@@ -1,3 +1,30 @@
|
|
1 |
---
|
2 |
license: openrail++
|
|
|
|
|
|
|
3 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
---
|
2 |
license: openrail++
|
3 |
+
tags:
|
4 |
+
- text-to-image
|
5 |
+
- stable-diffusion
|
6 |
---
|
7 |
+
|
8 |
+
This is a copy of the sdxSSD-1Bl model (https://huggingface.co/Vargol/lcm-ssd-1b-full-model) with the unet replaced
|
9 |
+
with the LCM distilled unet (https://huggingface.co/latent-consistency/lcm-ssd-1b) and scheduler config set to default to the LCM Scheduler.
|
10 |
+
|
11 |
+
This makes LCM SSD-B! run as a standard Diffusion Pipeline
|
12 |
+
|
13 |
+
```py
|
14 |
+
from diffusers import DiffusionPipeline
|
15 |
+
import torch
|
16 |
+
|
17 |
+
pipe = DiffusionPipeline.from_pretrained(
|
18 |
+
"Vargol/lcm-ssd-1b-full-model", variant='fp16', torch_dtype=torch.float16
|
19 |
+
).to("mps")
|
20 |
+
|
21 |
+
prompt = "Self-portrait oil painting, a beautiful cyborg with golden hair, 8k"
|
22 |
+
generator = torch.manual_seed(0)
|
23 |
+
|
24 |
+
image = pipe(
|
25 |
+
prompt=prompt, num_inference_steps=4, generator=generator, guidance_scale=8.0
|
26 |
+
).images[0]
|
27 |
+
|
28 |
+
image.save('distilled.png')
|
29 |
+
|
30 |
+
```
|
model_index.json
ADDED
@@ -0,0 +1,33 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_class_name": "StableDiffusionXLPipeline",
|
3 |
+
"_diffusers_version": "0.19.0",
|
4 |
+
"force_zeros_for_empty_prompt": true,
|
5 |
+
"scheduler": [
|
6 |
+
"diffusers",
|
7 |
+
"LCMScheduler"
|
8 |
+
],
|
9 |
+
"text_encoder": [
|
10 |
+
"transformers",
|
11 |
+
"CLIPTextModel"
|
12 |
+
],
|
13 |
+
"text_encoder_2": [
|
14 |
+
"transformers",
|
15 |
+
"CLIPTextModelWithProjection"
|
16 |
+
],
|
17 |
+
"tokenizer": [
|
18 |
+
"transformers",
|
19 |
+
"CLIPTokenizer"
|
20 |
+
],
|
21 |
+
"tokenizer_2": [
|
22 |
+
"transformers",
|
23 |
+
"CLIPTokenizer"
|
24 |
+
],
|
25 |
+
"unet": [
|
26 |
+
"diffusers",
|
27 |
+
"UNet2DConditionModel"
|
28 |
+
],
|
29 |
+
"vae": [
|
30 |
+
"diffusers",
|
31 |
+
"AutoencoderKL"
|
32 |
+
]
|
33 |
+
}
|
scheduler/scheduler_config.json
ADDED
@@ -0,0 +1,18 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_class_name": "LCMScheduler",
|
3 |
+
"_diffusers_version": "0.23.0",
|
4 |
+
"beta_end": 0.012,
|
5 |
+
"beta_schedule": "scaled_linear",
|
6 |
+
"beta_start": 0.00085,
|
7 |
+
"clip_sample": false,
|
8 |
+
"interpolation_type": "linear",
|
9 |
+
"num_train_timesteps": 1000,
|
10 |
+
"prediction_type": "epsilon",
|
11 |
+
"sample_max_value": 1.0,
|
12 |
+
"set_alpha_to_one": false,
|
13 |
+
"skip_prk_steps": true,
|
14 |
+
"steps_offset": 1,
|
15 |
+
"timestep_spacing": "leading",
|
16 |
+
"trained_betas": null,
|
17 |
+
"use_karras_sigmas": false
|
18 |
+
}
|
text_encoder/config.json
ADDED
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"architectures": [
|
3 |
+
"CLIPTextModel"
|
4 |
+
],
|
5 |
+
"attention_dropout": 0.0,
|
6 |
+
"bos_token_id": 0,
|
7 |
+
"dropout": 0.0,
|
8 |
+
"eos_token_id": 2,
|
9 |
+
"hidden_act": "quick_gelu",
|
10 |
+
"hidden_size": 768,
|
11 |
+
"initializer_factor": 1.0,
|
12 |
+
"initializer_range": 0.02,
|
13 |
+
"intermediate_size": 3072,
|
14 |
+
"layer_norm_eps": 1e-05,
|
15 |
+
"max_position_embeddings": 77,
|
16 |
+
"model_type": "clip_text_model",
|
17 |
+
"num_attention_heads": 12,
|
18 |
+
"num_hidden_layers": 12,
|
19 |
+
"pad_token_id": 1,
|
20 |
+
"projection_dim": 768,
|
21 |
+
"torch_dtype": "float32",
|
22 |
+
"transformers_version": "4.29.2",
|
23 |
+
"vocab_size": 49408
|
24 |
+
}
|
text_encoder/model.fp16.safetensors
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:5487ea0eee9c9a9bff8abd097908d4deff3ae1fa87b3b67397f8b9538139d447
|
3 |
+
size 246144864
|
text_encoder/model.safetensors
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:8377b1ca9d88fe06ec483dd7b3cfc62e5e8dbf8ddd252f455e79d659fa0553c5
|
3 |
+
size 492265880
|
text_encoder_2/config.json
ADDED
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"architectures": [
|
3 |
+
"CLIPTextModelWithProjection"
|
4 |
+
],
|
5 |
+
"attention_dropout": 0.0,
|
6 |
+
"bos_token_id": 0,
|
7 |
+
"dropout": 0.0,
|
8 |
+
"eos_token_id": 2,
|
9 |
+
"hidden_act": "gelu",
|
10 |
+
"hidden_size": 1280,
|
11 |
+
"initializer_factor": 1.0,
|
12 |
+
"initializer_range": 0.02,
|
13 |
+
"intermediate_size": 5120,
|
14 |
+
"layer_norm_eps": 1e-05,
|
15 |
+
"max_position_embeddings": 77,
|
16 |
+
"model_type": "clip_text_model",
|
17 |
+
"num_attention_heads": 20,
|
18 |
+
"num_hidden_layers": 32,
|
19 |
+
"pad_token_id": 1,
|
20 |
+
"projection_dim": 1280,
|
21 |
+
"torch_dtype": "float32",
|
22 |
+
"transformers_version": "4.29.2",
|
23 |
+
"vocab_size": 49408
|
24 |
+
}
|
text_encoder_2/model.fp16.safetensors
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:d3df577f6e3799c8e1bd9b40e30133710e02e8e25d0ce48cdcc790e7dfe12d6d
|
3 |
+
size 1389382880
|
text_encoder_2/model.safetensors
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:b84f413eebecbd049b72874c1df533a516510cb5a2489ae58c7e320209cf0ebe
|
3 |
+
size 2778702976
|
tokenizer/merges.txt
ADDED
The diff for this file is too large to render.
See raw diff
|
|
tokenizer/special_tokens_map.json
ADDED
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"bos_token": {
|
3 |
+
"content": "<|startoftext|>",
|
4 |
+
"lstrip": false,
|
5 |
+
"normalized": true,
|
6 |
+
"rstrip": false,
|
7 |
+
"single_word": false
|
8 |
+
},
|
9 |
+
"eos_token": {
|
10 |
+
"content": "<|endoftext|>",
|
11 |
+
"lstrip": false,
|
12 |
+
"normalized": true,
|
13 |
+
"rstrip": false,
|
14 |
+
"single_word": false
|
15 |
+
},
|
16 |
+
"pad_token": "<|endoftext|>",
|
17 |
+
"unk_token": {
|
18 |
+
"content": "<|endoftext|>",
|
19 |
+
"lstrip": false,
|
20 |
+
"normalized": true,
|
21 |
+
"rstrip": false,
|
22 |
+
"single_word": false
|
23 |
+
}
|
24 |
+
}
|
tokenizer/tokenizer_config.json
ADDED
@@ -0,0 +1,33 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"add_prefix_space": false,
|
3 |
+
"bos_token": {
|
4 |
+
"__type": "AddedToken",
|
5 |
+
"content": "<|startoftext|>",
|
6 |
+
"lstrip": false,
|
7 |
+
"normalized": true,
|
8 |
+
"rstrip": false,
|
9 |
+
"single_word": false
|
10 |
+
},
|
11 |
+
"clean_up_tokenization_spaces": true,
|
12 |
+
"do_lower_case": true,
|
13 |
+
"eos_token": {
|
14 |
+
"__type": "AddedToken",
|
15 |
+
"content": "<|endoftext|>",
|
16 |
+
"lstrip": false,
|
17 |
+
"normalized": true,
|
18 |
+
"rstrip": false,
|
19 |
+
"single_word": false
|
20 |
+
},
|
21 |
+
"errors": "replace",
|
22 |
+
"model_max_length": 77,
|
23 |
+
"pad_token": "<|endoftext|>",
|
24 |
+
"tokenizer_class": "CLIPTokenizer",
|
25 |
+
"unk_token": {
|
26 |
+
"__type": "AddedToken",
|
27 |
+
"content": "<|endoftext|>",
|
28 |
+
"lstrip": false,
|
29 |
+
"normalized": true,
|
30 |
+
"rstrip": false,
|
31 |
+
"single_word": false
|
32 |
+
}
|
33 |
+
}
|
tokenizer/vocab.json
ADDED
The diff for this file is too large to render.
See raw diff
|
|
tokenizer_2/merges.txt
ADDED
The diff for this file is too large to render.
See raw diff
|
|
tokenizer_2/special_tokens_map.json
ADDED
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"bos_token": {
|
3 |
+
"content": "<|startoftext|>",
|
4 |
+
"lstrip": false,
|
5 |
+
"normalized": true,
|
6 |
+
"rstrip": false,
|
7 |
+
"single_word": false
|
8 |
+
},
|
9 |
+
"eos_token": {
|
10 |
+
"content": "<|endoftext|>",
|
11 |
+
"lstrip": false,
|
12 |
+
"normalized": true,
|
13 |
+
"rstrip": false,
|
14 |
+
"single_word": false
|
15 |
+
},
|
16 |
+
"pad_token": "!",
|
17 |
+
"unk_token": {
|
18 |
+
"content": "<|endoftext|>",
|
19 |
+
"lstrip": false,
|
20 |
+
"normalized": true,
|
21 |
+
"rstrip": false,
|
22 |
+
"single_word": false
|
23 |
+
}
|
24 |
+
}
|
tokenizer_2/tokenizer_config.json
ADDED
@@ -0,0 +1,33 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"add_prefix_space": false,
|
3 |
+
"bos_token": {
|
4 |
+
"__type": "AddedToken",
|
5 |
+
"content": "<|startoftext|>",
|
6 |
+
"lstrip": false,
|
7 |
+
"normalized": true,
|
8 |
+
"rstrip": false,
|
9 |
+
"single_word": false
|
10 |
+
},
|
11 |
+
"clean_up_tokenization_spaces": true,
|
12 |
+
"do_lower_case": true,
|
13 |
+
"eos_token": {
|
14 |
+
"__type": "AddedToken",
|
15 |
+
"content": "<|endoftext|>",
|
16 |
+
"lstrip": false,
|
17 |
+
"normalized": true,
|
18 |
+
"rstrip": false,
|
19 |
+
"single_word": false
|
20 |
+
},
|
21 |
+
"errors": "replace",
|
22 |
+
"model_max_length": 77,
|
23 |
+
"pad_token": "!",
|
24 |
+
"tokenizer_class": "CLIPTokenizer",
|
25 |
+
"unk_token": {
|
26 |
+
"__type": "AddedToken",
|
27 |
+
"content": "<|endoftext|>",
|
28 |
+
"lstrip": false,
|
29 |
+
"normalized": true,
|
30 |
+
"rstrip": false,
|
31 |
+
"single_word": false
|
32 |
+
}
|
33 |
+
}
|
tokenizer_2/vocab.json
ADDED
The diff for this file is too large to render.
See raw diff
|
|
unet/README.md
ADDED
@@ -0,0 +1,74 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
library_name: diffusers
|
3 |
+
base_model: stabilityai/stable-diffusion-xl-base-1.0
|
4 |
+
tags:
|
5 |
+
- text-to-image
|
6 |
+
license: openrail++
|
7 |
+
inference: false
|
8 |
+
---
|
9 |
+
|
10 |
+
# Latent Consistency Model (LCM): SSD-1B
|
11 |
+
|
12 |
+
Latent Consistency Model (LCM) was proposed in [Latent Consistency Models: Synthesizing High-Resolution Images with Few-Step Inference](https://arxiv.org/abs/2310.04378)
|
13 |
+
by *Simian Luo, Yiqin Tan et al.* and [Simian Luo](https://huggingface.co/SimianLuo), [Suraj Patil](https://huggingface.co/valhalla), and [Daniel Gu](https://huggingface.co/dg845)
|
14 |
+
succesfully applied the same approach to create LCM for SDXL.
|
15 |
+
|
16 |
+
This checkpoint is a LCM distilled version of [`segmind/SSD-1B`](https://huggingface.co/segmind/SSD-1B) that allows
|
17 |
+
to reduce the number of inference steps to only between **2 - 8 steps**.
|
18 |
+
|
19 |
+
|
20 |
+
## Usage
|
21 |
+
|
22 |
+
LCM SDXL is supported in 🤗 Hugging Face Diffusers library from version v0.23.0 onwards. To run the model, first
|
23 |
+
install the latest version of the Diffusers library as well as `peft`, `accelerate` and `transformers`.
|
24 |
+
audio dataset from the Hugging Face Hub:
|
25 |
+
|
26 |
+
```bash
|
27 |
+
pip install --upgrade pip
|
28 |
+
pip install --upgrade diffusers transformers accelerate peft
|
29 |
+
```
|
30 |
+
|
31 |
+
### Text-to-Image
|
32 |
+
|
33 |
+
The model can be loaded with it's base pipeline `segmind/SSD-1B`. Next, the scheduler needs to be changed to [`LCMScheduler`](https://huggingface.co/docs/diffusers/v0.22.3/en/api/schedulers/lcm#diffusers.LCMScheduler) and we can reduce the number of inference steps to just 2 to 8 steps.
|
34 |
+
|
35 |
+
```python
|
36 |
+
from diffusers import UNet2DConditionModel, DiffusionPipeline, LCMScheduler
|
37 |
+
import torch
|
38 |
+
|
39 |
+
unet = UNet2DConditionModel.from_pretrained("latent-consistency/lcm-ssd-1b", torch_dtype=torch.float16, variant="fp16")
|
40 |
+
pipe = DiffusionPipeline.from_pretrained("segmind/SSD-1B", unet=unet, torch_dtype=torch.float16, variant="fp16")
|
41 |
+
|
42 |
+
pipe.scheduler = LCMScheduler.from_config(pipe.scheduler.config)
|
43 |
+
pipe.to("cuda")
|
44 |
+
|
45 |
+
prompt = "a close-up picture of an old man standing in the rain"
|
46 |
+
|
47 |
+
image = pipe(prompt, num_inference_steps=4, guidance_scale=1.0).images[0]
|
48 |
+
```
|
49 |
+
|
50 |
+
![](./image.png)
|
51 |
+
|
52 |
+
### Image-to-Image
|
53 |
+
|
54 |
+
Works as well! TODO docs
|
55 |
+
|
56 |
+
### Inpainting
|
57 |
+
|
58 |
+
Works as well! TODO docs
|
59 |
+
|
60 |
+
### ControlNet
|
61 |
+
|
62 |
+
Works as well! TODO docs
|
63 |
+
|
64 |
+
### T2I Adapter
|
65 |
+
|
66 |
+
Works as well! TODO docs
|
67 |
+
|
68 |
+
## Speed Benchmark
|
69 |
+
|
70 |
+
TODO
|
71 |
+
|
72 |
+
## Training
|
73 |
+
|
74 |
+
TODO
|
unet/config.json
ADDED
@@ -0,0 +1,93 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_class_name": "UNet2DConditionModel",
|
3 |
+
"_diffusers_version": "0.23.0.dev0",
|
4 |
+
"_name_or_path": "valhalla/lcm-ssd-distill-new-loss",
|
5 |
+
"act_fn": "silu",
|
6 |
+
"addition_embed_type": "text_time",
|
7 |
+
"addition_embed_type_num_heads": 64,
|
8 |
+
"addition_time_embed_dim": 256,
|
9 |
+
"attention_head_dim": [
|
10 |
+
5,
|
11 |
+
10,
|
12 |
+
20
|
13 |
+
],
|
14 |
+
"attention_type": "default",
|
15 |
+
"block_out_channels": [
|
16 |
+
320,
|
17 |
+
640,
|
18 |
+
1280
|
19 |
+
],
|
20 |
+
"center_input_sample": false,
|
21 |
+
"class_embed_type": null,
|
22 |
+
"class_embeddings_concat": false,
|
23 |
+
"conv_in_kernel": 3,
|
24 |
+
"conv_out_kernel": 3,
|
25 |
+
"cross_attention_dim": 2048,
|
26 |
+
"cross_attention_norm": null,
|
27 |
+
"down_block_types": [
|
28 |
+
"DownBlock2D",
|
29 |
+
"CrossAttnDownBlock2D",
|
30 |
+
"CrossAttnDownBlock2D"
|
31 |
+
],
|
32 |
+
"downsample_padding": 1,
|
33 |
+
"dropout": 0.0,
|
34 |
+
"dual_cross_attention": false,
|
35 |
+
"encoder_hid_dim": null,
|
36 |
+
"encoder_hid_dim_type": null,
|
37 |
+
"flip_sin_to_cos": true,
|
38 |
+
"freq_shift": 0,
|
39 |
+
"in_channels": 4,
|
40 |
+
"layers_per_block": 2,
|
41 |
+
"mid_block_only_cross_attention": null,
|
42 |
+
"mid_block_scale_factor": 1,
|
43 |
+
"mid_block_type": "UNetMidBlock2D",
|
44 |
+
"norm_eps": 1e-05,
|
45 |
+
"norm_num_groups": 32,
|
46 |
+
"num_attention_heads": null,
|
47 |
+
"num_class_embeds": null,
|
48 |
+
"only_cross_attention": false,
|
49 |
+
"out_channels": 4,
|
50 |
+
"projection_class_embeddings_input_dim": 2816,
|
51 |
+
"resnet_out_scale_factor": 1.0,
|
52 |
+
"resnet_skip_time_act": false,
|
53 |
+
"resnet_time_scale_shift": "default",
|
54 |
+
"reverse_transformer_layers_per_block": [
|
55 |
+
[
|
56 |
+
4,
|
57 |
+
4,
|
58 |
+
10
|
59 |
+
],
|
60 |
+
[
|
61 |
+
2,
|
62 |
+
1,
|
63 |
+
1
|
64 |
+
],
|
65 |
+
1
|
66 |
+
],
|
67 |
+
"sample_size": 128,
|
68 |
+
"time_cond_proj_dim": 256,
|
69 |
+
"time_embedding_act_fn": null,
|
70 |
+
"time_embedding_dim": null,
|
71 |
+
"time_embedding_type": "positional",
|
72 |
+
"timestep_post_act": null,
|
73 |
+
"transformer_layers_per_block": [
|
74 |
+
[
|
75 |
+
1
|
76 |
+
],
|
77 |
+
[
|
78 |
+
2,
|
79 |
+
2
|
80 |
+
],
|
81 |
+
[
|
82 |
+
4,
|
83 |
+
4
|
84 |
+
]
|
85 |
+
],
|
86 |
+
"up_block_types": [
|
87 |
+
"CrossAttnUpBlock2D",
|
88 |
+
"CrossAttnUpBlock2D",
|
89 |
+
"UpBlock2D"
|
90 |
+
],
|
91 |
+
"upcast_attention": null,
|
92 |
+
"use_linear_projection": true
|
93 |
+
}
|
unet/diffusion_pytorch_model.fp16.safetensors
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:81a14796546ca9150c6d3a502b862cba70f0bc94103ef446dea5a12851f761d0
|
3 |
+
size 2662954552
|
unet/diffusion_pytorch_model.safetensors
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:5333b678e8adb780e57a33fe52ca09b98d1a604fa99a1af39d63ee57364e6b96
|
3 |
+
size 5325787584
|
unet/download.png
ADDED
Git LFS Details
|
unet/image.png
ADDED
Git LFS Details
|
vae/config.json
ADDED
@@ -0,0 +1,31 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_class_name": "AutoencoderKL",
|
3 |
+
"_diffusers_version": "0.19.0",
|
4 |
+
"act_fn": "silu",
|
5 |
+
"block_out_channels": [
|
6 |
+
128,
|
7 |
+
256,
|
8 |
+
512,
|
9 |
+
512
|
10 |
+
],
|
11 |
+
"down_block_types": [
|
12 |
+
"DownEncoderBlock2D",
|
13 |
+
"DownEncoderBlock2D",
|
14 |
+
"DownEncoderBlock2D",
|
15 |
+
"DownEncoderBlock2D"
|
16 |
+
],
|
17 |
+
"force_upcast": true,
|
18 |
+
"in_channels": 3,
|
19 |
+
"latent_channels": 4,
|
20 |
+
"layers_per_block": 2,
|
21 |
+
"norm_num_groups": 32,
|
22 |
+
"out_channels": 3,
|
23 |
+
"sample_size": 1024,
|
24 |
+
"scaling_factor": 0.13025,
|
25 |
+
"up_block_types": [
|
26 |
+
"UpDecoderBlock2D",
|
27 |
+
"UpDecoderBlock2D",
|
28 |
+
"UpDecoderBlock2D",
|
29 |
+
"UpDecoderBlock2D"
|
30 |
+
]
|
31 |
+
}
|
vae/diffusion_pytorch_model.fp16.safetensors
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:6353737672c94b96174cb590f711eac6edf2fcce5b6e91aa9d73c5adc589ee48
|
3 |
+
size 167335342
|
vae/diffusion_pytorch_model.safetensors
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:78f6189c8492013e3cac81637a1f657f790a237387f8a9dfd6bfa5fee28eb646
|
3 |
+
size 334643268
|