Spaces:
Running
Running
presence: sync card
Browse files
README.md
CHANGED
|
@@ -81,7 +81,7 @@ Instead of retraining or repeating fine-tuning, Authentrics works directly with
|
|
| 81 |
|
| 82 |
**Demonstrates:** catastrophic forgetting, regression recovery, model optimization
|
| 83 |
|
| 84 |
-
- [Source code (GitHub)](https://github.com/Authentrics-ai/authentrics-
|
| 85 |
- [Model on Hugging Face](https://huggingface.co/authentrics/ztom-gemma3-1b-recovery)
|
| 86 |
|
| 87 |
---
|
|
@@ -96,7 +96,7 @@ Authentrics identifies where the unwanted learning was introduced and removes it
|
|
| 96 |
|
| 97 |
**Demonstrates:** memorization analysis, training attribution, machine unlearning
|
| 98 |
|
| 99 |
-
- [Source code (GitHub)](https://github.com/Authentrics-ai/authentrics-analysis-examples)
|
| 100 |
- [Model on Hugging Face](https://huggingface.co/authentrics/pii-removal-medical-llama32)
|
| 101 |
|
| 102 |
---
|
|
@@ -113,7 +113,7 @@ Authentrics analyzes the model's internal structure and behavior to expose routi
|
|
| 113 |
|
| 114 |
**Demonstrates:** model interpretability, MoE routing, internal model analysis
|
| 115 |
|
| 116 |
-
- [Source code (GitHub)](https://github.com/Authentrics-ai/authentrics-model-analysis-experiments)
|
| 117 |
- [Model on Hugging Face](https://huggingface.co/authentrics/mixtral-expert-routing)
|
| 118 |
|
| 119 |
#### Find Where a Model Changed Across Its Training Run
|
|
|
|
| 81 |
|
| 82 |
**Demonstrates:** catastrophic forgetting, regression recovery, model optimization
|
| 83 |
|
| 84 |
+
- [Source code (GitHub)](https://github.com/Authentrics-ai/authentrics-analysis-examples/blob/main/hf_legal_understanding.ipynb)
|
| 85 |
- [Model on Hugging Face](https://huggingface.co/authentrics/ztom-gemma3-1b-recovery)
|
| 86 |
|
| 87 |
---
|
|
|
|
| 96 |
|
| 97 |
**Demonstrates:** memorization analysis, training attribution, machine unlearning
|
| 98 |
|
| 99 |
+
- [Source code (GitHub)](https://github.com/Authentrics-ai/authentrics-analysis-examples/blob/main/hf_medical_chatbot.ipynb)
|
| 100 |
- [Model on Hugging Face](https://huggingface.co/authentrics/pii-removal-medical-llama32)
|
| 101 |
|
| 102 |
---
|
|
|
|
| 113 |
|
| 114 |
**Demonstrates:** model interpretability, MoE routing, internal model analysis
|
| 115 |
|
| 116 |
+
- [Source code (GitHub)](https://github.com/Authentrics-ai/authentrics-model-analysis-experiments#mistralmixtral-moe-expert-activation)
|
| 117 |
- [Model on Hugging Face](https://huggingface.co/authentrics/mixtral-expert-routing)
|
| 118 |
|
| 119 |
#### Find Where a Model Changed Across Its Training Run
|