Spaces:
Running
on
Zero
Running
on
Zero
Update utils.py
Browse files
utils.py
CHANGED
|
@@ -191,7 +191,7 @@ class MolecularPropertyPredictionModel():
|
|
| 191 |
self.base_model = AutoModelForSequenceClassification.from_pretrained(
|
| 192 |
"ChemFM/ChemFM-3B",
|
| 193 |
config=config,
|
| 194 |
-
device_map="
|
| 195 |
trust_remote_code=True,
|
| 196 |
token = os.environ.get("TOKEN")
|
| 197 |
)
|
|
@@ -284,7 +284,10 @@ class MolecularPropertyPredictionModel():
|
|
| 284 |
for i, batch in tqdm(enumerate(test_loader), total=len(test_loader), desc="Evaluating"):
|
| 285 |
with torch.no_grad():
|
| 286 |
batch = {k: v.to(self.base_model.device) for k, v in batch.items()}
|
|
|
|
|
|
|
| 287 |
outputs = self.base_model(**batch)
|
|
|
|
| 288 |
if task_type == "regression": # TODO: check if the model is regression or classification
|
| 289 |
y_pred.append(outputs.logits.cpu().detach().numpy())
|
| 290 |
else:
|
|
|
|
| 191 |
self.base_model = AutoModelForSequenceClassification.from_pretrained(
|
| 192 |
"ChemFM/ChemFM-3B",
|
| 193 |
config=config,
|
| 194 |
+
device_map="cuda",
|
| 195 |
trust_remote_code=True,
|
| 196 |
token = os.environ.get("TOKEN")
|
| 197 |
)
|
|
|
|
| 284 |
for i, batch in tqdm(enumerate(test_loader), total=len(test_loader), desc="Evaluating"):
|
| 285 |
with torch.no_grad():
|
| 286 |
batch = {k: v.to(self.base_model.device) for k, v in batch.items()}
|
| 287 |
+
print(self.base_model.device)
|
| 288 |
+
print(batch)
|
| 289 |
outputs = self.base_model(**batch)
|
| 290 |
+
print(output)
|
| 291 |
if task_type == "regression": # TODO: check if the model is regression or classification
|
| 292 |
y_pred.append(outputs.logits.cpu().detach().numpy())
|
| 293 |
else:
|