Update README.md
Browse files
README.md
CHANGED
|
@@ -1,28 +1,3 @@
|
|
| 1 |
-
Ran find roberta files
|
| 2 |
-
Checked task Find roberta files
|
| 3 |
-
Used tool: manage_task
|
| 4 |
-
Analyzed HUGGINGFACE_ROBERTA_HOSTING.md:L1-61
|
| 5 |
-
Analyzed configuration_roberta_risk.py:L1-33
|
| 6 |
-
Analyzed modeling_roberta_risk.py:L1-100
|
| 7 |
-
Analyzed modeling_roberta_risk.py:L101-133
|
| 8 |
-
Analyzed host_roberta_on_hf.py:L1-100
|
| 9 |
-
Analyzed roberta_model.py:L1-100
|
| 10 |
-
Ran list files in src/risk_model
|
| 11 |
-
Analyzed train.py:L1-100
|
| 12 |
-
Analyzed test_roberta_model.py:L1-65
|
| 13 |
-
Analyzed roberta_model.py:L101-243
|
| 14 |
-
Analyzed calibrate.py:L1-100
|
| 15 |
-
Analyzed attribution.py:L1-50
|
| 16 |
-
Ran check artifacts/risk_model 路 Exited with code 1
|
| 17 |
-
Ran list artifacts directory
|
| 18 |
-
Analyzed taxonomy.json:L1-25
|
| 19 |
-
Ran find risk tests
|
| 20 |
-
Ran list all tests
|
| 21 |
-
Analyzed test_hf_loading.py:L1-48
|
| 22 |
-
Analyzed test_calibration.py:L1-100
|
| 23 |
-
Analyzed test_risk_bands.py:L1-100
|
| 24 |
-
|
| 25 |
-
````markdown
|
| 26 |
---
|
| 27 |
library_name: transformers
|
| 28 |
tags:
|
|
@@ -276,15 +251,3 @@ InferenceGuard Contributors. (2026). *InferenceGuard: Multi-Task Inferential Pri
|
|
| 276 |
- **Integrated Gradients:** An axiomatic attribution method that calculates the path integral of gradients along the straight line from a baseline input to the actual input embeddings.
|
| 277 |
- **Risk Bands:** Categorical classifications (LOW, MEDIUM, HIGH) used in user-facing applications to summarize numerical exposure scores into actionable thresholds.
|
| 278 |
|
| 279 |
-
## More Information
|
| 280 |
-
|
| 281 |
-
To deploy this model as part of the complete InferenceGuard pipeline or to set up live web monitoring, consult the documentation in README.md and the deployment recipes in docs/PRODUCTION_DEPLOYMENT.md.
|
| 282 |
-
|
| 283 |
-
## Model Card Authors
|
| 284 |
-
|
| 285 |
-
Engineered and evaluated by the InferenceGuard core development team.
|
| 286 |
-
|
| 287 |
-
## Model Card Contact
|
| 288 |
-
|
| 289 |
-
For questions, issues, or model weight contributions, open an issue in the project repository or contact the InferenceGuard development team.
|
| 290 |
-
````
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
library_name: transformers
|
| 3 |
tags:
|
|
|
|
| 251 |
- **Integrated Gradients:** An axiomatic attribution method that calculates the path integral of gradients along the straight line from a baseline input to the actual input embeddings.
|
| 252 |
- **Risk Bands:** Categorical classifications (LOW, MEDIUM, HIGH) used in user-facing applications to summarize numerical exposure scores into actionable thresholds.
|
| 253 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|