Instructions to use phi-lab-rice/GRADE with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use phi-lab-rice/GRADE with Diffusers:
pip install -U diffusers transformers accelerate
import torch from diffusers import DiffusionPipeline # switch to "mps" for apple devices pipe = DiffusionPipeline.from_pretrained("phi-lab-rice/GRADE", 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
Record dataset README PR and remaining author actions
Browse files- REVIEW_STATUS.md +3 -2
REVIEW_STATUS.md
CHANGED
|
@@ -10,14 +10,15 @@ This file tracks the points in the artifact review against this model release.
|
|
| 10 |
| LPIPS first-run download | The README documents the AlexNet download and how to populate a persistent cache before offline use. |
|
| 11 |
| Anonymous Hugging Face rate limits | The README documents authenticated CLI downloads of the separate processed evaluation dataset. |
|
| 12 |
| Table 7 in local mode | Its output names the input source. It uses supplied reference results unless a fresh sampling-step CSV is present. |
|
| 13 |
-
| Stale dataset release language |
|
| 14 |
|
| 15 |
The 15 distinct model paths completed a one-frame inference check against the
|
| 16 |
available archived predictions. The full GRADE path also completed a one-frame
|
| 17 |
2D and 3D metric run. These checks establish basic load and execution, not a
|
| 18 |
fresh reproduction of the full 20-sequence paper results.
|
| 19 |
|
| 20 |
-
|
| 21 |
|
| 22 |
1. A permanent archive DOI for a versioned code, model, and evaluation release has not been created. The paper DOI in the model card identifies the paper, not this artifact package.
|
| 23 |
2. Author records confirming consent, anonymization, and redistribution clearance for the already public processed evaluation dataset have not been provided for this review. The model repository does not duplicate that dataset.
|
|
|
|
|
|
| 10 |
| LPIPS first-run download | The README documents the AlexNet download and how to populate a persistent cache before offline use. |
|
| 11 |
| Anonymous Hugging Face rate limits | The README documents authenticated CLI downloads of the separate processed evaluation dataset. |
|
| 12 |
| Table 7 in local mode | Its output names the input source. It uses supplied reference results unless a fresh sampling-step CSV is present. |
|
| 13 |
+
| Stale dataset release language | A corrected README is included here and has been submitted as [evaluation dataset PR #1](https://huggingface.co/datasets/mypersonalsharingspot11/evaluation_dataset/discussions/1). The dataset owner must merge it; the dataset's `main` README is still stale. |
|
| 14 |
|
| 15 |
The 15 distinct model paths completed a one-frame inference check against the
|
| 16 |
available archived predictions. The full GRADE path also completed a one-frame
|
| 17 |
2D and 3D metric run. These checks establish basic load and execution, not a
|
| 18 |
fresh reproduction of the full 20-sequence paper results.
|
| 19 |
|
| 20 |
+
Three review items require author or account action and are **not complete**:
|
| 21 |
|
| 22 |
1. A permanent archive DOI for a versioned code, model, and evaluation release has not been created. The paper DOI in the model card identifies the paper, not this artifact package.
|
| 23 |
2. Author records confirming consent, anonymization, and redistribution clearance for the already public processed evaluation dataset have not been provided for this review. The model repository does not duplicate that dataset.
|
| 24 |
+
3. The corrected README for the processed evaluation dataset is awaiting a merge by that repository's owner.
|