added gemma

app.py

@@ -7,15 +7,33 @@ import plotly.express as px
 from config import other_info_dict
 from utils import *
 # %%
-st.title("Microsoft Phi-2 LLM assessment")
+st.title("LLM assessments: Microsoft's Phi-2 and Google's Gemma-7b")
 # st.image('model_card.png', caption='Hugging face description', use_column_width=True)
-st.write("""
-    Microsoft Phi-2 (https://huggingface.co/microsoft/phi-2) is a Transformer model with 2.7 billion parameters. Performance on benchmarks for common sense, language understanding, and logical reasoning is nearly state-of-the-art among models with less than 13 billion parameters. Unlike typical Large Language Models (LLM), Phi-2 has not been fine-tuned through reinforcement learning from human feedback.""")
 
 import urllib.request 
 import os
-prefix_post_processing = os.environ["POST_PROCESSING_JSON"]
 
+model_select = st.selectbox(
+    'Select the model:',
+    [
+        "Microsoft's Phi-2",
+        "Google's Gemma"
+    ])
+
+if model_select == "Microsoft's Phi-2":
+    prefix_post_processing = os.environ["POST_PROCESSING_JSON"]
+    st.write("""
+    Microsoft's Phi-2 (https://huggingface.co/microsoft/phi-2) is a Transformer model with 2.7 billion parameters. Performance on benchmarks for common sense, language understanding, and logical reasoning is nearly state-of-the-art among models with less than 13 billion parameters. Unlike typical Large Language Models (LLM), Phi-2 has not been fine-tuned through reinforcement learning from human feedback.""")
+
+else:
+    prefix_post_processing = os.environ["POST_PROCESSING_JSON_GEMMA"]
+    st.write("""
+    Google Gemma-7b (https://huggingface.co/google/gemma-7b) is a Large Language Models (LLM) with 8.54 billion parameters. As per the https://storage.googleapis.com/deepmind-media/gemma/gemma-report.pdf, Gemma-7b performs well in language understanding, reasoning, and safety tasks. This model is one of state of the art open models built based on similar technologies that were used to create Google's Gemini models.""")
+
+
+
+
+# prefix_post_processing = os.environ["POST_PROCESSING_JSON"]
 
 st.header('Evaluation dataset')
 st.write(other_info_dict['data_description'])
@@ -317,9 +335,6 @@ merged_t_fair_df = pd.concat(t_fair_dfs_list, axis=0)
 
 
 fig_fair =  px.scatter(merged_t_fair_df, x='Category', y='Estimate', error_y='Diff upper', error_y_minus='Diff lower', color='Prompt', symbol='Prompt')
-
-# fig_fair = None
-
 fig_fair.update_layout(yaxis_title="Performance in %")
 
 st.plotly_chart(fig_fair, theme="streamlit", use_container_width=True)