Prannaya-Laptop 2024-8-11:16:8:21

walledai · Aug 11, 2024 · 01c72f7 · 01c72f7
1 parent d57f4c4
commit 01c72f7
Show file tree

Hide file tree

Showing 10 changed files with 2,471 additions and 188 deletions.
diff --git a/docs/assets/short_paper.pdf b/docs/assets/short_paper.pdf
diff --git a/docs/components/dataset.md b/docs/components/dataset.md
@@ -134,4 +134,11 @@ dummy_dataset[0]
 # Prompt(prompt="what's up")
 ```
 
+You can use this to load data from Pandas by 
+
+## Conclusion
+
+We are able to 
+
+
 Hence, we are able to load several variety of datasets using our library. If there are any other ways you'd like us to support, do highlight in our [GitHub Issues](https://github.com/walledai/walledeval/issues).
diff --git a/docs/index.md b/docs/index.md
@@ -21,20 +21,18 @@
 
 > Grateful to [Tensorplex](https://www.tensorplex.ai/) for their support with computing resources!
 
-<!--
+
 ## 🔍 Quick Access
 
-<div class="grid cards">
-<a href="https://paperswithcode.com/paper/walledeval-a-comprehensive-safety-evaluation" class="card"><div markdown>
+<div class="grid cards" markdown>
 
-:simple-paperswithcode: Papers With Code
+- [:simple-github:{ .lg .middle } __<font size="4" style="margin-left:10px">GitHub</font>__](https://github.com/walledai/walledeval)
+- [:simple-arxiv:{ .lg .middle } __arXiv__](https://arxiv.org/abs/2408.03837)
+- [:simple-paperswithcode:{ .lg .middle } __Papers With Code__](https://paperswithcode.com/paper/walledeval-a-comprehensive-safety-evaluation)
+- [:simple-semanticscholar:{ .lg .middle } __Semantic Scholar__](https://www.semanticscholar.org/paper/WalledEval%3A-A-Comprehensive-Safety-Evaluation-for-Gupta-Yau/5c7da78b978e2ef6cc791cfbf98dafbcb59f758b)
 
-</div></a>
-<a href="https://www.semanticscholar.org/paper/WalledEval%3A-A-Comprehensive-Safety-Evaluation-for-Gupta-Yau/5c7da78b978e2ef6cc791cfbf98dafbcb59f758b" class="card">
-:simple-semanticscholar: Semantic Scholar
-</a>
 </div>
--->
+
 
 ## 📚 Resources
 

diff --git a/docs/quickstart/auto-red-teaming.md b/docs/quickstart/auto-red-teaming.md
@@ -8,7 +8,7 @@ Automated Red-Teaming allows users to automatically generate mutated malicious p
 - **Prompts**: a compilation of malicious prompts
 - **Mutators**: a way to create adverserial prompts from the malicious ones. This may or may not be generative.
 
- Here's how you can do this easily in WalledEval!
+Here's how you can do this easily in WalledEval!
 
 ```python title="auto_red_teaming_quickstart.py" linenums="1" hl_lines="23 33"
 import torch

diff --git a/docs/quickstart/prompts.md b/docs/quickstart/prompts.md
@@ -1,6 +1,3 @@
----
-hide: '["toc"]'
----
 # Prompt Benchmarking
 
 Most safety datasets aim to test LLMs on their creativity / restraint in generating responses to custom unsafe/safe queries. The general requirements for testing an LLM on Prompt Benchmarks is as follows:
@@ -63,3 +60,6 @@ logs[0]["score"] # True if safe, False if unsafe
 ```
 
 
+## Dataset Loading
+
+In this example, we load a list of prompts. Prompt Benchmarking is the most baseline benchmarking method as the only required parameter is **prompt**. Hence, we can input a list of strings 
diff --git a/docs/quickstart/refusal.md b/docs/quickstart/refusal.md