Update README.md
Browse files
README.md
CHANGED
|
@@ -6,6 +6,8 @@ license: llama3
|
|
| 6 |
This model is fine-tuned to enhance resistance to indirect prompt injection attacks, particularly in tasks such as email and document summarization.
|
| 7 |
It leverages specific data delimiters (*\<\<\<data\>\>\>* and *\<\<\</data\>\>\>*) to safely handle untrusted input by ignoring any instructions within those markers.
|
| 8 |
|
|
|
|
|
|
|
| 9 |
## Usage
|
| 10 |
To use the model, it's crucial to set the appropriate system message that was used during fine-tuning. The system message ensures the model treats text within <<<data>>> and <<</data>>> as data and disregards any embedded instructions.
|
| 11 |
|
|
|
|
| 6 |
This model is fine-tuned to enhance resistance to indirect prompt injection attacks, particularly in tasks such as email and document summarization.
|
| 7 |
It leverages specific data delimiters (*\<\<\<data\>\>\>* and *\<\<\</data\>\>\>*) to safely handle untrusted input by ignoring any instructions within those markers.
|
| 8 |
|
| 9 |
+
This was the output of research described in this WithSecure Labs article: https://labs.withsecure.com/publications/llama3-prompt-injection-hardening.
|
| 10 |
+
|
| 11 |
## Usage
|
| 12 |
To use the model, it's crucial to set the appropriate system message that was used during fine-tuning. The system message ensures the model treats text within <<<data>>> and <<</data>>> as data and disregards any embedded instructions.
|
| 13 |
|