davidberenstein1957 commited on
Commit
582bb79
·
verified ·
1 Parent(s): 50d85b1

Add files using upload-large-folder tool

Browse files
README.md CHANGED
@@ -2,8 +2,6 @@
2
  library_name: diffusers
3
  tags:
4
  - pruna-ai
5
- base_model:
6
- - black-forest-labs/FLUX.1-dev
7
  ---
8
 
9
  # Model Card for PrunaAI/FLUX.1-dev-smashed
@@ -41,21 +39,24 @@ The compression configuration of the model is stored in the `smash_config.json`
41
  "batcher": null,
42
  "cacher": "fora",
43
  "compiler": "torch_compile",
44
- "factorizer": "qkv_diffusers",
45
  "pruner": null,
46
- "quantizer": null,
47
- "fora_interval": 2,
48
  "fora_start_step": 2,
49
  "torch_compile_backend": "inductor",
50
  "torch_compile_dynamic": null,
51
  "torch_compile_fullgraph": true,
52
  "torch_compile_make_portable": false,
53
  "torch_compile_max_kv_cache_size": 400,
54
- "torch_compile_mode": "default",
55
  "torch_compile_seqlen_manual_cuda_graph": 100,
56
  "torch_compile_target": "model",
 
 
57
  "batch_size": 1,
58
- "device": "cuda",
 
59
  "save_fns": [
60
  "save_before_apply",
61
  "save_before_apply"
@@ -64,9 +65,9 @@ The compression configuration of the model is stored in the `smash_config.json`
64
  "diffusers"
65
  ],
66
  "reapply_after_load": {
67
- "factorizer": "qkv_diffusers",
68
  "pruner": null,
69
- "quantizer": null,
70
  "cacher": "fora",
71
  "compiler": "torch_compile",
72
  "batcher": null
 
2
  library_name: diffusers
3
  tags:
4
  - pruna-ai
 
 
5
  ---
6
 
7
  # Model Card for PrunaAI/FLUX.1-dev-smashed
 
39
  "batcher": null,
40
  "cacher": "fora",
41
  "compiler": "torch_compile",
42
+ "factorizer": null,
43
  "pruner": null,
44
+ "quantizer": "torchao",
45
+ "fora_interval": 3,
46
  "fora_start_step": 2,
47
  "torch_compile_backend": "inductor",
48
  "torch_compile_dynamic": null,
49
  "torch_compile_fullgraph": true,
50
  "torch_compile_make_portable": false,
51
  "torch_compile_max_kv_cache_size": 400,
52
+ "torch_compile_mode": "max-autotune-no-cudagraphs",
53
  "torch_compile_seqlen_manual_cuda_graph": 100,
54
  "torch_compile_target": "model",
55
+ "torchao_excluded_modules": "norm+embedding",
56
+ "torchao_quant_type": "int8dq",
57
  "batch_size": 1,
58
+ "device": "cpu",
59
+ "device_map": null,
60
  "save_fns": [
61
  "save_before_apply",
62
  "save_before_apply"
 
65
  "diffusers"
66
  ],
67
  "reapply_after_load": {
68
+ "factorizer": null,
69
  "pruner": null,
70
+ "quantizer": "torchao",
71
  "cacher": "fora",
72
  "compiler": "torch_compile",
73
  "batcher": null
model_index.json CHANGED
@@ -1,6 +1,6 @@
1
  {
2
  "_class_name": "FluxPipeline",
3
- "_diffusers_version": "0.33.1",
4
  "_name_or_path": "black-forest-labs/FLUX.1-dev",
5
  "feature_extractor": [
6
  null,
 
1
  {
2
  "_class_name": "FluxPipeline",
3
+ "_diffusers_version": "0.34.0",
4
  "_name_or_path": "black-forest-labs/FLUX.1-dev",
5
  "feature_extractor": [
6
  null,
scheduler/scheduler_config.json CHANGED
@@ -1,6 +1,6 @@
1
  {
2
  "_class_name": "FlowMatchEulerDiscreteScheduler",
3
- "_diffusers_version": "0.33.1",
4
  "base_image_seq_len": 256,
5
  "base_shift": 0.5,
6
  "invert_sigmas": false,
@@ -9,6 +9,7 @@
9
  "num_train_timesteps": 1000,
10
  "shift": 3.0,
11
  "shift_terminal": null,
 
12
  "time_shift_type": "exponential",
13
  "use_beta_sigmas": false,
14
  "use_dynamic_shifting": true,
 
1
  {
2
  "_class_name": "FlowMatchEulerDiscreteScheduler",
3
+ "_diffusers_version": "0.34.0",
4
  "base_image_seq_len": 256,
5
  "base_shift": 0.5,
6
  "invert_sigmas": false,
 
9
  "num_train_timesteps": 1000,
10
  "shift": 3.0,
11
  "shift_terminal": null,
12
+ "stochastic_sampling": false,
13
  "time_shift_type": "exponential",
14
  "use_beta_sigmas": false,
15
  "use_dynamic_shifting": true,
smash_config.json CHANGED
@@ -2,21 +2,24 @@
2
  "batcher": null,
3
  "cacher": "fora",
4
  "compiler": "torch_compile",
5
- "factorizer": "qkv_diffusers",
6
  "pruner": null,
7
- "quantizer": null,
8
- "fora_interval": 2,
9
  "fora_start_step": 2,
10
  "torch_compile_backend": "inductor",
11
  "torch_compile_dynamic": null,
12
  "torch_compile_fullgraph": true,
13
  "torch_compile_make_portable": false,
14
  "torch_compile_max_kv_cache_size": 400,
15
- "torch_compile_mode": "default",
16
  "torch_compile_seqlen_manual_cuda_graph": 100,
17
  "torch_compile_target": "model",
 
 
18
  "batch_size": 1,
19
- "device": "cuda",
 
20
  "save_fns": [
21
  "save_before_apply",
22
  "save_before_apply"
@@ -25,9 +28,9 @@
25
  "diffusers"
26
  ],
27
  "reapply_after_load": {
28
- "factorizer": "qkv_diffusers",
29
  "pruner": null,
30
- "quantizer": null,
31
  "cacher": "fora",
32
  "compiler": "torch_compile",
33
  "batcher": null
 
2
  "batcher": null,
3
  "cacher": "fora",
4
  "compiler": "torch_compile",
5
+ "factorizer": null,
6
  "pruner": null,
7
+ "quantizer": "torchao",
8
+ "fora_interval": 3,
9
  "fora_start_step": 2,
10
  "torch_compile_backend": "inductor",
11
  "torch_compile_dynamic": null,
12
  "torch_compile_fullgraph": true,
13
  "torch_compile_make_portable": false,
14
  "torch_compile_max_kv_cache_size": 400,
15
+ "torch_compile_mode": "max-autotune-no-cudagraphs",
16
  "torch_compile_seqlen_manual_cuda_graph": 100,
17
  "torch_compile_target": "model",
18
+ "torchao_excluded_modules": "norm+embedding",
19
+ "torchao_quant_type": "int8dq",
20
  "batch_size": 1,
21
+ "device": "cpu",
22
+ "device_map": null,
23
  "save_fns": [
24
  "save_before_apply",
25
  "save_before_apply"
 
28
  "diffusers"
29
  ],
30
  "reapply_after_load": {
31
+ "factorizer": null,
32
  "pruner": null,
33
+ "quantizer": "torchao",
34
  "cacher": "fora",
35
  "compiler": "torch_compile",
36
  "batcher": null
text_encoder/config.json CHANGED
@@ -19,6 +19,6 @@
19
  "pad_token_id": 1,
20
  "projection_dim": 768,
21
  "torch_dtype": "bfloat16",
22
- "transformers_version": "4.52.4",
23
  "vocab_size": 49408
24
  }
 
19
  "pad_token_id": 1,
20
  "projection_dim": 768,
21
  "torch_dtype": "bfloat16",
22
+ "transformers_version": "4.53.1",
23
  "vocab_size": 49408
24
  }
text_encoder_2/config.json CHANGED
@@ -25,7 +25,7 @@
25
  "relative_attention_num_buckets": 32,
26
  "tie_word_embeddings": false,
27
  "torch_dtype": "bfloat16",
28
- "transformers_version": "4.52.4",
29
  "use_cache": true,
30
  "vocab_size": 32128
31
  }
 
25
  "relative_attention_num_buckets": 32,
26
  "tie_word_embeddings": false,
27
  "torch_dtype": "bfloat16",
28
+ "transformers_version": "4.53.1",
29
  "use_cache": true,
30
  "vocab_size": 32128
31
  }
text_encoder_2/model.safetensors.index.json CHANGED
@@ -1,5 +1,6 @@
1
  {
2
  "metadata": {
 
3
  "total_size": 9524621312
4
  },
5
  "weight_map": {
 
1
  {
2
  "metadata": {
3
+ "total_parameters": 4762310656,
4
  "total_size": 9524621312
5
  },
6
  "weight_map": {
transformer/config.json CHANGED
@@ -1,7 +1,7 @@
1
  {
2
  "_class_name": "FluxTransformer2DModel",
3
- "_diffusers_version": "0.33.1",
4
- "_name_or_path": "/root/.cache/huggingface/hub/models--black-forest-labs--FLUX.1-dev/snapshots/0ef5fff789c832c5c7f4e127f94c8b54bbcced44/transformer",
5
  "attention_head_dim": 128,
6
  "axes_dims_rope": [
7
  16,
 
1
  {
2
  "_class_name": "FluxTransformer2DModel",
3
+ "_diffusers_version": "0.34.0",
4
+ "_name_or_path": "/root/.cache/huggingface/hub/models--black-forest-labs--FLUX.1-dev/snapshots/3de623fc3c33e44ffbe2bad470d0f45bccf2eb21/transformer",
5
  "attention_head_dim": 128,
6
  "axes_dims_rope": [
7
  16,
vae/config.json CHANGED
@@ -1,7 +1,7 @@
1
  {
2
  "_class_name": "AutoencoderKL",
3
- "_diffusers_version": "0.33.1",
4
- "_name_or_path": "/root/.cache/huggingface/hub/models--black-forest-labs--FLUX.1-dev/snapshots/0ef5fff789c832c5c7f4e127f94c8b54bbcced44/vae",
5
  "act_fn": "silu",
6
  "block_out_channels": [
7
  128,
 
1
  {
2
  "_class_name": "AutoencoderKL",
3
+ "_diffusers_version": "0.34.0",
4
+ "_name_or_path": "/root/.cache/huggingface/hub/models--black-forest-labs--FLUX.1-dev/snapshots/3de623fc3c33e44ffbe2bad470d0f45bccf2eb21/vae",
5
  "act_fn": "silu",
6
  "block_out_channels": [
7
  128,