Instructions to use Anwaarma/try1 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- PEFT
How to use Anwaarma/try1 with PEFT:
from peft import PeftModel from transformers import AutoModelForSequenceClassification base_model = AutoModelForSequenceClassification.from_pretrained("Anwaarma/edos_taskB_llama3b_merged2_FINAL") model = PeftModel.from_pretrained(base_model, "Anwaarma/try1") - Transformers
How to use Anwaarma/try1 with Transformers:
# Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("Anwaarma/try1", dtype="auto") - Notebooks
- Google Colab
- Kaggle
Upload lora/thresholds_multiclass.json with huggingface_hub
Browse files
lora/thresholds_multiclass.json
ADDED
|
@@ -0,0 +1,11 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"type": "multiclass_adjusted_argmax",
|
| 3 |
+
"per_class_thresholds": {
|
| 4 |
+
"animosity": 0.165055513381958,
|
| 5 |
+
"derogation": 0.29203441739082336,
|
| 6 |
+
"prejudiced discussions": 0.0,
|
| 7 |
+
"threats, plans to harm and incitement": 0.12110953032970428
|
| 8 |
+
},
|
| 9 |
+
"dev_macro_f1": 0.6390521680841315,
|
| 10 |
+
"test_macro_f1": 0.5968829142666536
|
| 11 |
+
}
|