Text Classification
Transformers
Safetensors
smollm3
agent-safety
tool-calling
long-context
Eval Results (legacy)
Instructions to use ProCreations/auto-3b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use ProCreations/auto-3b with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-classification", model="ProCreations/auto-3b")# pip install -U transformers accelerate # Load model directly from transformers import AutoTokenizer, AutoModelForSequenceClassification tokenizer = AutoTokenizer.from_pretrained("ProCreations/auto-3b") model = AutoModelForSequenceClassification.from_pretrained("ProCreations/auto-3b", device_map="auto") - Notebooks
- Google Colab
- Kaggle
Release auto-3b (SmolLM3-3B-Base approve/deny classifier, 98.03% on Approve-or-Deny)
58323da verified Download calibration.json from ProCreations/auto-3b: direct link, hf CLI and curl.
- Browser
- Download file 711 Bytes
-
https://huggingface.co/ProCreations/auto-3b/resolve/main/calibration.json
- Command line
-
hf download hf://ProCreations/auto-3b/calibration.json
-
curl -L -o calibration.json https://huggingface.co/ProCreations/auto-3b/resolve/main/calibration.json
711 Bytes
| { | |
| "threshold": 0.295, | |
| "validation_metrics": { | |
| "n": 2581, | |
| "accuracy": 0.9833397907787679, | |
| "balanced_accuracy": 0.9833393179247119, | |
| "f1_deny": 0.9833655705996132, | |
| "auroc": 0.9978311360475756, | |
| "nll": 0.06344092496728404, | |
| "false_approve_rate": 0.016253869969040248, | |
| "false_deny_rate": 0.017067494181536073, | |
| "false_approve_count": 21, | |
| "false_deny_count": 22, | |
| "deny_count": 1292, | |
| "approve_count": 1289, | |
| "threshold": 0.295, | |
| "brier": 0.015282586776204823, | |
| "accuracy_ci95": [ | |
| 0.9776351922332552, | |
| 0.9876077597032694 | |
| ] | |
| }, | |
| "criterion": "balanced accuracy; ties prefer fewer false approvals then 0.5 proximity", | |
| "benchmark_used": false | |
| } | |