Spaces:
Runtime error
Runtime error
Update app.py
Browse files
app.py
CHANGED
@@ -177,7 +177,6 @@ with gr.Blocks(theme=gr.themes.Soft()) as demo:
|
|
177 |
<div>
|
178 |
<div style = "font-size: 13px;">
|
179 |
<p><In Live Auditing Mode, you gain the ability to probe the LLM directly./p>
|
180 |
-
|
181 |
<p>First, select the LLM you wish to audit. Then, enter your question. The AuditLLM tool will generate five relevant and diverse prompts based on your question. You can now select these prompts for auditing the LLMs. Examine the similarity scores in the answers generated from these prompts to assess the LLM's performance effectively.</p>
|
182 |
|
183 |
</div>
|
@@ -220,9 +219,7 @@ with gr.Blocks(theme=gr.themes.Soft()) as demo:
|
|
220 |
<div>
|
221 |
<div style = "font-size: 13px;">
|
222 |
<p>In batch auditing mode, you have the capability to probe the LLM. To begin, you must first select the LLM you wish to audit and then input the questions you intend to explore. For each question submitted, the model will generate five prompts, each accompanied by its respective answers.</p>
|
223 |
-
|
224 |
<p>To tailor the generation of these five prompts from your original question, you can adjust the relevance and diversity scores. The relevance score determines how closely the generated prompts should align with the original question, while the diversity score dictates the variance among the prompts themselves.</p>
|
225 |
-
|
226 |
<p>Upon completion, please provide your email address. We will compile and send the answers to you promptly.</p>
|
227 |
</div>
|
228 |
""")
|
@@ -243,4 +240,4 @@ with gr.Blocks(theme=gr.themes.Soft()) as demo:
|
|
243 |
|
244 |
|
245 |
# Launch the Gradio app
|
246 |
-
demo.launch()
|
|
|
177 |
<div>
|
178 |
<div style = "font-size: 13px;">
|
179 |
<p><In Live Auditing Mode, you gain the ability to probe the LLM directly./p>
|
|
|
180 |
<p>First, select the LLM you wish to audit. Then, enter your question. The AuditLLM tool will generate five relevant and diverse prompts based on your question. You can now select these prompts for auditing the LLMs. Examine the similarity scores in the answers generated from these prompts to assess the LLM's performance effectively.</p>
|
181 |
|
182 |
</div>
|
|
|
219 |
<div>
|
220 |
<div style = "font-size: 13px;">
|
221 |
<p>In batch auditing mode, you have the capability to probe the LLM. To begin, you must first select the LLM you wish to audit and then input the questions you intend to explore. For each question submitted, the model will generate five prompts, each accompanied by its respective answers.</p>
|
|
|
222 |
<p>To tailor the generation of these five prompts from your original question, you can adjust the relevance and diversity scores. The relevance score determines how closely the generated prompts should align with the original question, while the diversity score dictates the variance among the prompts themselves.</p>
|
|
|
223 |
<p>Upon completion, please provide your email address. We will compile and send the answers to you promptly.</p>
|
224 |
</div>
|
225 |
""")
|
|
|
240 |
|
241 |
|
242 |
# Launch the Gradio app
|
243 |
+
demo.launch()
|