Instructions to use hf-internal-testing/tiny-cosmos3-modular-pipe with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use hf-internal-testing/tiny-cosmos3-modular-pipe with Diffusers:
pip install -U diffusers transformers accelerate
import torch from diffusers import DiffusionPipeline # switch to "mps" for apple devices pipe = DiffusionPipeline.from_pretrained("hf-internal-testing/tiny-cosmos3-modular-pipe", dtype=torch.bfloat16, device_map="cuda") prompt = "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k" image = pipe(prompt).images[0] - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- Draw Things
- DiffusionBee
Shrink tiny Cosmos3 transformer for CPU tests
#2
by atharvajoshi10 - opened
transformer/config.json
CHANGED
|
@@ -1,7 +1,6 @@
|
|
| 1 |
{
|
| 2 |
"_class_name": "Cosmos3OmniTransformer",
|
| 3 |
"_diffusers_version": "0.40.0.dev0",
|
| 4 |
-
"_name_or_path": "/tmp/cosmos3-tiny-modular-prototype-20260715",
|
| 5 |
"action_dim": 10,
|
| 6 |
"action_gen": true,
|
| 7 |
"attention_bias": false,
|
|
@@ -9,27 +8,27 @@
|
|
| 9 |
"backbone_type": "cosmos3_qwen3vl",
|
| 10 |
"base_fps": 24,
|
| 11 |
"enable_fps_modulation": true,
|
| 12 |
-
"head_dim":
|
| 13 |
"hidden_act": "silu",
|
| 14 |
-
"hidden_size":
|
| 15 |
-
"intermediate_size":
|
| 16 |
"latent_channel": 4,
|
| 17 |
"latent_patch_size": 2,
|
| 18 |
-
"num_attention_heads":
|
| 19 |
"num_embodiment_domains": 32,
|
| 20 |
-
"num_hidden_layers":
|
| 21 |
-
"num_key_value_heads":
|
| 22 |
"patch_latent_dim": 16,
|
| 23 |
"qk_norm_for_text": true,
|
| 24 |
"rms_norm_eps": 1e-06,
|
| 25 |
"rope_axes_dim": [
|
| 26 |
-
|
| 27 |
1,
|
| 28 |
1
|
| 29 |
],
|
| 30 |
"rope_scaling": {
|
| 31 |
"mrope_section": [
|
| 32 |
-
|
| 33 |
1,
|
| 34 |
1
|
| 35 |
]
|
|
|
|
| 1 |
{
|
| 2 |
"_class_name": "Cosmos3OmniTransformer",
|
| 3 |
"_diffusers_version": "0.40.0.dev0",
|
|
|
|
| 4 |
"action_dim": 10,
|
| 5 |
"action_gen": true,
|
| 6 |
"attention_bias": false,
|
|
|
|
| 8 |
"backbone_type": "cosmos3_qwen3vl",
|
| 9 |
"base_fps": 24,
|
| 10 |
"enable_fps_modulation": true,
|
| 11 |
+
"head_dim": 6,
|
| 12 |
"hidden_act": "silu",
|
| 13 |
+
"hidden_size": 6,
|
| 14 |
+
"intermediate_size": 12,
|
| 15 |
"latent_channel": 4,
|
| 16 |
"latent_patch_size": 2,
|
| 17 |
+
"num_attention_heads": 1,
|
| 18 |
"num_embodiment_domains": 32,
|
| 19 |
+
"num_hidden_layers": 1,
|
| 20 |
+
"num_key_value_heads": 1,
|
| 21 |
"patch_latent_dim": 16,
|
| 22 |
"qk_norm_for_text": true,
|
| 23 |
"rms_norm_eps": 1e-06,
|
| 24 |
"rope_axes_dim": [
|
| 25 |
+
1,
|
| 26 |
1,
|
| 27 |
1
|
| 28 |
],
|
| 29 |
"rope_scaling": {
|
| 30 |
"mrope_section": [
|
| 31 |
+
1,
|
| 32 |
1,
|
| 33 |
1
|
| 34 |
]
|
transformer/diffusion_pytorch_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:382866740d429a75e2ca32857f115ab8d5ce3bc430ff0e4d0191c9afb6a7f1ce
|
| 3 |
+
size 7311624
|