Upload TRAINING_PLAN.md with huggingface_hub
Browse files- TRAINING_PLAN.md +19 -0
TRAINING_PLAN.md
ADDED
|
@@ -0,0 +1,19 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# Training plan: user_generated_content
|
| 2 |
+
|
| 3 |
+
| Setting | Value |
|
| 4 |
+
|---------|-------|
|
| 5 |
+
| **Steps** | 2500 |
|
| 6 |
+
| **Learning rate** | 0.0001 |
|
| 7 |
+
| **Trainer** | p-image-edit-trainer |
|
| 8 |
+
|
| 9 |
+
## Command (from repo root)
|
| 10 |
+
|
| 11 |
+
```bash
|
| 12 |
+
uv run python training/run_image_trainer.py run /Users/davidberenstein/Documents/programming/pruna/dataset-generator/training/runs/user_generated_content/input.zip \
|
| 13 |
+
--edit \
|
| 14 |
+
--steps 2500 \
|
| 15 |
+
--learning-rate 0.0001 \
|
| 16 |
+
-o training/user_generated_content_lora.zip
|
| 17 |
+
```
|
| 18 |
+
|
| 19 |
+
If the zip is **>100MB**, add `--serve`.
|