UX fix: README.md (download instructions / safetensors / pickle-free demo)
Browse files
README.md
CHANGED
|
@@ -33,6 +33,22 @@ DELETE zeroes it, CREATE allocates a fresh one, READ is just routing — no retr
|
|
| 33 |
**confidence margin** (top-1 − top-2) is used as an "I don't know which fact you mean" signal, so the
|
| 34 |
model refuses low-confidence queries.
|
| 35 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 36 |
## Load it (safetensors, no pickle)
|
| 37 |
|
| 38 |
```python
|
|
|
|
| 33 |
**confidence margin** (top-1 − top-2) is used as an "I don't know which fact you mean" signal, so the
|
| 34 |
model refuses low-confidence queries.
|
| 35 |
|
| 36 |
+
## Download
|
| 37 |
+
|
| 38 |
+
Use the Hugging Face Hub (handles the Git-LFS weights for you). **Don't `git clone` without
|
| 39 |
+
`git-lfs`** — you'd get 132-byte LFS pointer files instead of the real `model.safetensors` / `.pt`.
|
| 40 |
+
|
| 41 |
+
```bash
|
| 42 |
+
pip install huggingface_hub
|
| 43 |
+
# CPU-only deps (avoids the multi-GB CUDA stack):
|
| 44 |
+
pip install -r requirements.txt --extra-index-url https://download.pytorch.org/whl/cpu
|
| 45 |
+
```
|
| 46 |
+
```python
|
| 47 |
+
from huggingface_hub import snapshot_download
|
| 48 |
+
repo = snapshot_download("TilelliLab/Yaz") # -> local path with all files (real weights)
|
| 49 |
+
# then: cd into `repo` and run the snippets below, or `python demo.py --demo`.
|
| 50 |
+
```
|
| 51 |
+
|
| 52 |
## Load it (safetensors, no pickle)
|
| 53 |
|
| 54 |
```python
|