Update README.md
Browse files
README.md
CHANGED
|
@@ -1,22 +1,51 @@
|
|
| 1 |
---
|
|
|
|
| 2 |
base_model: unsloth/llama-3-8b-bnb-4bit
|
| 3 |
tags:
|
| 4 |
-
- text-generation-inference
|
| 5 |
-
- transformers
|
| 6 |
- unsloth
|
| 7 |
-
- llama
|
| 8 |
- trl
|
| 9 |
-
|
| 10 |
-
|
| 11 |
-
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 12 |
---
|
| 13 |
|
| 14 |
-
#
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 15 |
|
| 16 |
-
|
| 17 |
-
|
| 18 |
-
|
|
|
|
|
|
|
| 19 |
|
| 20 |
This llama model was trained 2x faster with [Unsloth](https://github.com/unslothai/unsloth)
|
| 21 |
|
| 22 |
-
[<img src="https://raw.githubusercontent.com/unslothai/unsloth/main/images/unsloth%20made%20with%20love.png" width="200"/>](https://github.com/unslothai/unsloth)
|
|
|
|
| 1 |
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
base_model: unsloth/llama-3-8b-bnb-4bit
|
| 4 |
tags:
|
|
|
|
|
|
|
| 5 |
- unsloth
|
|
|
|
| 6 |
- trl
|
| 7 |
+
- lora
|
| 8 |
+
- ethics
|
| 9 |
+
- compliance
|
| 10 |
+
- gdpr
|
| 11 |
+
- transformers
|
| 12 |
+
- gradio
|
| 13 |
+
- safetensors
|
| 14 |
+
model-index:
|
| 15 |
+
- name: PrismAI-Llama-Risk-Monitor
|
| 16 |
+
results: []
|
| 17 |
+
datasets:
|
| 18 |
+
- coastalcph/lex_glue
|
| 19 |
+
- ai4privacy/pii-masking-65k
|
| 20 |
+
- Anthropic/hh-rlhf
|
| 21 |
---
|
| 22 |
|
| 23 |
+
# PrismAI: Llama-Risk-Complaint
|
| 24 |
+
|
| 25 |
+
This model is a fine-tuned version of Llama-3-8B optimized for **Corporate Compliance and Ethical Monitoring**.
|
| 26 |
+
|
| 27 |
+
## Model Description
|
| 28 |
+
PrismAI is trained to identify and mitigate risks associated with:
|
| 29 |
+
- **GDPR Violations:** Accidental sharing of PII (Names, Addresses, ID numbers).
|
| 30 |
+
- **Workplace Ethics:** Unconscious bias in hiring or management communication.
|
| 31 |
+
- **Legal Risks:** Improper handling of "Opt-out" lists or internal secrets.
|
| 32 |
+
|
| 33 |
+
## Training Procedure
|
| 34 |
+
- **Library:** Unsloth & PEFT
|
| 35 |
+
- **Hardware:** NVIDIA T4 GPU
|
| 36 |
+
- **Quantization:** 4-bit (bitsandbytes)
|
| 37 |
+
- **Method:** Supervised Fine-Tuning (SFT)
|
| 38 |
+
|
| 39 |
+
## Use Case Examples
|
| 40 |
+
**Input:** "I'm sending Sarah's home address (123 Maple St) to the external marketing vendor."
|
| 41 |
+
**Output:** "🛡️ Risk Detected: PII exposure. Ensure Sarah's consent is documented and use a secure encrypted channel for home address sharing."
|
| 42 |
|
| 43 |
+
## How to use
|
| 44 |
+
```python
|
| 45 |
+
from unsloth import FastLanguageModel
|
| 46 |
+
model, tokenizer = FastLanguageModel.from_pretrained("nirmanpatel/llama-risk-compliant")
|
| 47 |
+
```
|
| 48 |
|
| 49 |
This llama model was trained 2x faster with [Unsloth](https://github.com/unslothai/unsloth)
|
| 50 |
|
| 51 |
+
[<img src="https://raw.githubusercontent.com/unslothai/unsloth/main/images/unsloth%20made%20with%20love.png" width="200"/>](https://github.com/unslothai/unsloth)
|