tricky_questions_and_avg_calcs

.gitignore

@@ -0,0 +1 @@
+venv/

app.py

@@ -12,6 +12,8 @@ AVERAGE_COLUMN_NAME = "Average"
 SENTIMENT_COLUMN_NAME = "Sentiment"
 UNDERSTANDING_COLUMN_NAME = "Language understanding"
 PHRASEOLOGY_COLUMN_NAME = "Phraseology"
+TRICKY_QUESTIONS_COLUMN_NAME = "Tricky questions"
+IMPLICATURES_AVERAGE_COLUMN_NAME = "Implicatures average"
 
 # Function to load data from JSON file
 @st.cache_data
@@ -23,23 +25,63 @@ def load_data(file_path):
 # Function to style the DataFrame
 @st.cache_data
 def style_dataframe(df: pd.DataFrame):
-    df[RESULTS_COLUMN_NAME] = df.apply(lambda row: [row[SENTIMENT_COLUMN_NAME], row[UNDERSTANDING_COLUMN_NAME], row[PHRASEOLOGY_COLUMN_NAME]], axis=1)
+    # Calculate Implicatures average from the three columns
+    df[IMPLICATURES_AVERAGE_COLUMN_NAME] = df.apply(
+        lambda row: (row[SENTIMENT_COLUMN_NAME] + row[UNDERSTANDING_COLUMN_NAME] + row[PHRASEOLOGY_COLUMN_NAME]) / 3, 
+        axis=1
+    )
+    
+    # Calculate Average from all four columns
+    df[AVERAGE_COLUMN_NAME] = df.apply(
+        lambda row: (row[SENTIMENT_COLUMN_NAME] + row[UNDERSTANDING_COLUMN_NAME] + 
+                    row[PHRASEOLOGY_COLUMN_NAME] + row[TRICKY_QUESTIONS_COLUMN_NAME]) / 4, 
+        axis=1
+    )
+    
+    df[RESULTS_COLUMN_NAME] = df.apply(
+        lambda row: [row[SENTIMENT_COLUMN_NAME], row[UNDERSTANDING_COLUMN_NAME], 
+                    row[PHRASEOLOGY_COLUMN_NAME], row[TRICKY_QUESTIONS_COLUMN_NAME]], 
+        axis=1
+    )
+    
     # Insert the new column after the 'Average' column
     cols = list(df.columns)
-    cols.insert(cols.index(AVERAGE_COLUMN_NAME) + 1, cols.pop(cols.index(RESULTS_COLUMN_NAME)))
+    avg_index = cols.index(AVERAGE_COLUMN_NAME)
+    
+    # Remove columns from their current positions if they exist
+    if IMPLICATURES_AVERAGE_COLUMN_NAME in cols:
+        cols.pop(cols.index(IMPLICATURES_AVERAGE_COLUMN_NAME))
+    if RESULTS_COLUMN_NAME in cols:
+        cols.pop(cols.index(RESULTS_COLUMN_NAME))
+    
+    # Insert columns in the desired order
+    cols.insert(avg_index + 1, IMPLICATURES_AVERAGE_COLUMN_NAME)
+    cols.insert(avg_index + 2, RESULTS_COLUMN_NAME)
+    
     df = df[cols]
-    # Create a color ramp using Seaborn
     return df
 
 def styler(df: pd.DataFrame):
     palette = sns.color_palette("RdYlGn", as_cmap=True)
     # Apply reverse color gradient to the "Params" column
     params_palette = sns.color_palette("RdYlGn_r", as_cmap=True)  # Reversed RdYlGn palette
-    styled_df = df.style.background_gradient(cmap=palette, subset=[AVERAGE_COLUMN_NAME, SENTIMENT_COLUMN_NAME, PHRASEOLOGY_COLUMN_NAME, UNDERSTANDING_COLUMN_NAME]
-                                            ).background_gradient(cmap=params_palette, subset=["Params"]
-                                            ).set_properties(**{'text-align': 'center'}, subset=[AVERAGE_COLUMN_NAME, SENTIMENT_COLUMN_NAME, PHRASEOLOGY_COLUMN_NAME, UNDERSTANDING_COLUMN_NAME]
-                                            ).format("{:.2f}".center(10), subset=[AVERAGE_COLUMN_NAME, SENTIMENT_COLUMN_NAME, PHRASEOLOGY_COLUMN_NAME, UNDERSTANDING_COLUMN_NAME]
-                                            ).format("{:.1f}".center(10), subset=["Params"])
+    styled_df = df.style.background_gradient(
+        cmap=palette, 
+        subset=[AVERAGE_COLUMN_NAME, IMPLICATURES_AVERAGE_COLUMN_NAME, SENTIMENT_COLUMN_NAME, 
+                PHRASEOLOGY_COLUMN_NAME, UNDERSTANDING_COLUMN_NAME, TRICKY_QUESTIONS_COLUMN_NAME]
+    ).background_gradient(
+        cmap=params_palette, subset=["Params"]
+    ).set_properties(
+        **{'text-align': 'center'}, 
+        subset=[AVERAGE_COLUMN_NAME, IMPLICATURES_AVERAGE_COLUMN_NAME, SENTIMENT_COLUMN_NAME, 
+                PHRASEOLOGY_COLUMN_NAME, UNDERSTANDING_COLUMN_NAME, TRICKY_QUESTIONS_COLUMN_NAME]
+    ).format(
+        "{:.2f}".center(10), 
+        subset=[AVERAGE_COLUMN_NAME, IMPLICATURES_AVERAGE_COLUMN_NAME, SENTIMENT_COLUMN_NAME, 
+                PHRASEOLOGY_COLUMN_NAME, UNDERSTANDING_COLUMN_NAME, TRICKY_QUESTIONS_COLUMN_NAME]
+    ).format(
+        "{:.1f}".center(10), subset=["Params"]
+    )
     return styled_df
 
 
@@ -149,7 +191,7 @@ with tab1:
     # Closing filters in a expander
     with st.expander("Filtering benchmark data", icon='🔍'):
         # Filtering data, e.g. slider for params, average score, etc.
-        col_filter_params, col_filter_average, col_filter_sentiment, col_filter_understanding, col_filter_phraseology = st.columns(5, gap='medium')
+        col_filter_params, col_filter_average, col_filter_sentiment, col_filter_understanding, col_filter_phraseology, col_filter_tricky = st.columns(6, gap='medium')
     
         with col_filter_params:
             params_slider = st.slider("Models Size [B]", min_value=0.0, max_value=float(data['Params'].max()), value=(0.0, float(data['Params'].max())), step=0.1, format="%.1f")
@@ -170,6 +212,10 @@ with tab1:
         with col_filter_phraseology:
             phraseology_slider = st.slider("Phraseology score", step=0.1, min_value=0.0, max_value=5.0, value=(0.0, 5.0))
             data = data[(data[PHRASEOLOGY_COLUMN_NAME] >= phraseology_slider[0]) & (data[PHRASEOLOGY_COLUMN_NAME] <= phraseology_slider[1])]
+            
+        with col_filter_tricky:
+            tricky_slider = st.slider("Tricky questions score", step=0.1, min_value=0.0, max_value=5.0, value=(0.0, 5.0))
+            data = data[(data[TRICKY_QUESTIONS_COLUMN_NAME] >= tricky_slider[0]) & (data[TRICKY_QUESTIONS_COLUMN_NAME] <= tricky_slider[1])]
 
         # Extract unique provider names from the "Model" column
         providers = data["Model"].apply(lambda x: x.split('/')[0].lower()).unique()
@@ -191,6 +237,8 @@ with tab1:
         SENTIMENT_COLUMN_NAME: st.column_config.NumberColumn(SENTIMENT_COLUMN_NAME, help='Ability to analyze sentiment'),
         UNDERSTANDING_COLUMN_NAME: st.column_config.NumberColumn(UNDERSTANDING_COLUMN_NAME, help='Ability to understand language'),
         PHRASEOLOGY_COLUMN_NAME: st.column_config.NumberColumn(PHRASEOLOGY_COLUMN_NAME, help='Ability to understand phraseological compounds'),
+        TRICKY_QUESTIONS_COLUMN_NAME: st.column_config.NumberColumn(TRICKY_QUESTIONS_COLUMN_NAME, help='Ability to understand tricky questions'),
+        IMPLICATURES_AVERAGE_COLUMN_NAME: st.column_config.NumberColumn(IMPLICATURES_AVERAGE_COLUMN_NAME, help='Average of sentiment, understanding, and phraseology'),
         }, hide_index=True, disabled=True, height=500)
 
     # Add selection for models and create a bar chart for selected models using the AVERAGE_COLUMN_NAME, SENTIMENT_COLUMN_NAME, PHRASEOLOGY_COLUMN_NAME, UNDERSTANDING_COLUMN_NAME
@@ -201,7 +249,7 @@ with tab1:
         default_models.append(bielik_model)
     selected_models = st.multiselect("Select models to compare", data["Model"].unique(), default=default_models)
     selected_data = data[data["Model"].isin(selected_models)]
-    categories = [AVERAGE_COLUMN_NAME, SENTIMENT_COLUMN_NAME, PHRASEOLOGY_COLUMN_NAME, UNDERSTANDING_COLUMN_NAME]
+    categories = [AVERAGE_COLUMN_NAME, IMPLICATURES_AVERAGE_COLUMN_NAME, SENTIMENT_COLUMN_NAME, PHRASEOLOGY_COLUMN_NAME, UNDERSTANDING_COLUMN_NAME, TRICKY_QUESTIONS_COLUMN_NAME]
 
     if selected_models:
         # Kolorki do wyboru:

data.json

@@ -5,7 +5,8 @@
         "Average": 4.03025641025641,
         "Sentiment": 4.230769230769231,
         "Language understanding": 4.0,
-        "Phraseology": 3.86
+        "Phraseology": 3.86,
+        "Tricky questions": 4.0
     },
     {
         "Model": "alpindale/WizardLM-2-8x22B",
@@ -13,7 +14,8 @@
         "Average": 3.9133760683760683,
         "Sentiment": 3.7051282051282053,
         "Language understanding": 3.815,
-        "Phraseology": 4.22
+        "Phraseology": 4.22,
+        "Tricky questions": 4.0
     },
     {
         "Model": "meta-llama/Meta-Llama-3.1-70B-Instruct",
@@ -21,7 +23,8 @@
         "Average": 3.828974358974359,
         "Sentiment": 4.326923076923077,
         "Language understanding": 3.91,
-        "Phraseology": 3.25
+        "Phraseology": 3.25,
+        "Tricky questions": 4.0
     },
     {
         "Model": "meta-llama/Meta-Llama-3-70B-Instruct",
@@ -29,7 +32,8 @@
         "Average": 3.806538461538462,
         "Sentiment": 4.134615384615385,
         "Language understanding": 3.82,
-        "Phraseology": 3.465
+        "Phraseology": 3.465,
+        "Tricky questions": 4.0
     },
     {
         "Model": "speakleash/Bielik-11B-v2.3-Instruct",
@@ -37,7 +41,8 @@
         "Average": 3.7697863247863252,
         "Sentiment": 3.9743589743589745,
         "Language understanding": 3.785,
-        "Phraseology": 3.55
+        "Phraseology": 3.55,
+        "Tricky questions": 4.0
     },
     {
         "Model": "mistralai/Mixtral-8x22B-Instruct-v0.1",
@@ -45,7 +50,8 @@
         "Average": 3.6690170940170943,
         "Sentiment": 3.782051282051282,
         "Language understanding": 3.675,
-        "Phraseology": 3.55
+        "Phraseology": 3.55,
+        "Tricky questions": 4.0
     },
     {
         "Model": "speakleash/Bielik-11B-v2.1-Instruct",
@@ -53,7 +59,8 @@
         "Average": 3.6583760683760684,
         "Sentiment": 3.9551282051282053,
         "Language understanding": 3.915,
-        "Phraseology": 3.105
+        "Phraseology": 3.105,
+        "Tricky questions": 4.0
     },
     {
         "Model": "Qwen/Qwen2-72B-Instruct",
@@ -61,7 +68,8 @@
         "Average": 3.6442735042735044,
         "Sentiment": 3.7628205128205128,
         "Language understanding": 3.89,
-        "Phraseology": 3.28
+        "Phraseology": 3.28,
+        "Tricky questions": 4.0
     },
     {
         "Model": "speakleash/Bielik-11B-v2.0-Instruct",
@@ -69,7 +77,8 @@
         "Average": 3.614786324786325,
         "Sentiment": 3.9743589743589745,
         "Language understanding": 3.745,
-        "Phraseology": 3.125
+        "Phraseology": 3.125,
+        "Tricky questions": 4.0
     },
     {
         "Model": "speakleash/Bielik-11B-v2.2-Instruct",
@@ -77,7 +86,8 @@
         "Average": 3.565982905982906,
         "Sentiment": 3.717948717948718,
         "Language understanding": 3.73,
-        "Phraseology": 3.25
+        "Phraseology": 3.25,
+        "Tricky questions": 4.0
     },
     {
         "Model": "Qwen/Qwen1.5-72B-Chat",
@@ -85,7 +95,8 @@
         "Average": 3.3214529914529916,
         "Sentiment": 3.4743589743589745,
         "Language understanding": 3.515,
-        "Phraseology": 2.975
+        "Phraseology": 2.975,
+        "Tricky questions": 4.0
     },
     {
         "Model": "meta-llama/Meta-Llama-3.1-8B-Instruct",
@@ -93,7 +104,8 @@
         "Average": 3.3114529914529918,
         "Sentiment": 3.9743589743589745,
         "Language understanding": 3.38,
-        "Phraseology": 2.58
+        "Phraseology": 2.58,
+        "Tricky questions": 4.0
     },
     {
         "Model": "THUDM/glm-4-9b-chat",
@@ -101,7 +113,8 @@
         "Average": 3.2749145299145295,
         "Sentiment": 3.58974358974359,
         "Language understanding": 3.455,
-        "Phraseology": 2.78
+        "Phraseology": 2.78,
+        "Tricky questions": 4.0
     },
     {
         "Model": "mistralai/Mistral-Nemo-Instruct-2407",
@@ -109,7 +122,8 @@
         "Average": 3.223675213675214,
         "Sentiment": 3.641025641025641,
         "Language understanding": 3.29,
-        "Phraseology": 2.74
+        "Phraseology": 2.74,
+        "Tricky questions": 4.0
     },
     {
         "Model": "meta-llama/Meta-Llama-3-8B-Instruct",
@@ -117,7 +131,8 @@
         "Average": 3.172777777777778,
         "Sentiment": 3.3333333333333335,
         "Language understanding": 3.15,
-        "Phraseology": 3.035
+        "Phraseology": 3.035,
+        "Tricky questions": 4.0
     },
     {
         "Model": "upstage/SOLAR-10.7B-Instruct-v1.0",
@@ -125,7 +140,8 @@
         "Average": 3.1343162393162394,
         "Sentiment": 2.967948717948718,
         "Language understanding": 3.18,
-        "Phraseology": 3.255
+        "Phraseology": 3.255,
+        "Tricky questions": 4.0
     },
     {
         "Model": "speakleash/Bielik-7B-Instruct-v0.1",
@@ -133,7 +149,8 @@
         "Average": 3.126581196581197,
         "Sentiment": 3.58974358974359,
         "Language understanding": 3.475,
-        "Phraseology": 2.315
+        "Phraseology": 2.315,
+        "Tricky questions": 4.0
     },
     {
         "Model": "openchat/openchat-3.5-0106-gemma",
@@ -141,7 +158,8 @@
         "Average": 3.08525641025641,
         "Sentiment": 3.730769230769231,
         "Language understanding": 3.08,
-        "Phraseology": 2.445
+        "Phraseology": 2.445,
+        "Tricky questions": 4.0
     },
     {
         "Model": "mistralai/Mixtral-8x7B-Instruct-v0.1",
@@ -149,7 +167,8 @@
         "Average": 3.039230769230769,
         "Sentiment": 3.0576923076923075,
         "Language understanding": 3.175,
-        "Phraseology": 2.885
+        "Phraseology": 2.885,
+        "Tricky questions": 4.0
     },
     {
         "Model": "mistralai/Mistral-7B-Instruct-v0.3",
@@ -157,7 +176,8 @@
         "Average": 3.022307692307692,
         "Sentiment": 3.326923076923077,
         "Language understanding": 3.06,
-        "Phraseology": 2.68
+        "Phraseology": 2.68,
+        "Tricky questions": 4.0
     },
     {
         "Model": "berkeley-nest/Starling-LM-7B-alpha",
@@ -165,7 +185,8 @@
         "Average": 2.945897435897436,
         "Sentiment": 3.0576923076923075,
         "Language understanding": 2.925,
-        "Phraseology": 2.855
+        "Phraseology": 2.855,
+        "Tricky questions": 4.0
     },
     {
         "Model": "openchat/openchat-3.5-0106",
@@ -173,7 +194,8 @@
         "Average": 2.8500854700854696,
         "Sentiment": 3.16025641025641,
         "Language understanding": 2.835,
-        "Phraseology": 2.555
+        "Phraseology": 2.555,
+        "Tricky questions": 4.0
     },
     {
         "Model": "internlm/internlm2-chat-20b",
@@ -181,7 +203,8 @@
         "Average": 2.8237606837606837,
         "Sentiment": 3.301282051282051,
         "Language understanding": 2.785,
-        "Phraseology": 2.385
+        "Phraseology": 2.385,
+        "Tricky questions": 4.0
     },
     {
         "Model": "01-ai/Yi-1.5-34B-Chat",
@@ -189,7 +212,8 @@
         "Average": 2.7756410256410255,
         "Sentiment": 3.076923076923077,
         "Language understanding": 2.87,
-        "Phraseology": 2.38
+        "Phraseology": 2.38,
+        "Tricky questions": 4.0
     },
     {
         "Model": "Voicelab/trurl-2-13b-academic",
@@ -197,7 +221,8 @@
         "Average": 2.74042735042735,
         "Sentiment": 3.301282051282051,
         "Language understanding": 2.755,
-        "Phraseology": 2.165
+        "Phraseology": 2.165,
+        "Tricky questions": 4.0
     },
     {
         "Model": "google/gemma-2-2b-it",
@@ -205,7 +230,8 @@
         "Average": 2.7974786324786325,
         "Sentiment": 3.3974358974359,
         "Language understanding": 2.9,
-        "Phraseology": 2.095
+        "Phraseology": 2.095,
+        "Tricky questions": 4.0
     },
     {
         "Model": "Qwen/Qwen2.5-3B-Instruct",
@@ -213,7 +239,8 @@
         "Average": 2.734572649572649,
         "Sentiment": 2.948717948717949,
         "Language understanding": 2.455,
-        "Phraseology": 2.8
+        "Phraseology": 2.8,
+        "Tricky questions": 4.0
     },
     {
         "Model": "NousResearch/Hermes-3-Llama-3.2-3B",
@@ -221,7 +248,8 @@
         "Average": 2.695128205128205,
         "Sentiment": 2.6153846153846154,
         "Language understanding": 2.705,
-        "Phraseology": 2.765
+        "Phraseology": 2.765,
+        "Tricky questions": 4.0
     },
     {
         "Model": "ibm-granite/granite-3.1-2b-instruct",
@@ -229,7 +257,8 @@
         "Average": 2.397307692307692,
         "Sentiment": 3.076923076923077,
         "Language understanding": 2.235,
-        "Phraseology": 1.88
+        "Phraseology": 1.88,
+        "Tricky questions": 4.0
     },
     {
         "Model": "meta-llama/Llama-3.2-1B-Instruct",
@@ -237,7 +266,8 @@
         "Average": 2.383974358974359,
         "Sentiment": 3.076923076923077,
         "Language understanding": 1.735,
-        "Phraseology": 2.34
+        "Phraseology": 2.34,
+        "Tricky questions": 4.0
     },
     {
         "Model": "microsoft/Phi-3.5-mini-instruct",
@@ -245,7 +275,8 @@
         "Average": 2.331965811965812,
         "Sentiment": 2.435897435897436,
         "Language understanding": 2.135,
-        "Phraseology": 2.425
+        "Phraseology": 2.425,
+        "Tricky questions": 4.0
     },
     {
         "Model": "meta-llama/Llama-3.2-3B-Instruct",
@@ -253,7 +284,8 @@
         "Average": 2.257136752136752,
         "Sentiment": 2.7564102564102564,
         "Language understanding": 2.295,
-        "Phraseology": 1.72
+        "Phraseology": 1.72,
+        "Tricky questions": 4.0
     },
     {
         "Model": "h2oai/h2o-danube2-1.8b-chat",
@@ -261,7 +293,8 @@
         "Average": 2.1455982905982904,
         "Sentiment": 2.371794871794872,
         "Language understanding": 1.595,
-        "Phraseology": 2.47
+        "Phraseology": 2.47,
+        "Tricky questions": 4.0
     },
     {
         "Model": "Qwen/Qwen2.5-1.5B-Instruct",
@@ -269,7 +302,8 @@
         "Average": 2.1232905982905983,
         "Sentiment": 2.7948717948717947,
         "Language understanding": 1.35,
-        "Phraseology": 2.225
+        "Phraseology": 2.225,
+        "Tricky questions": 4.0
     },
     {
         "Model": "utter-project/EuroLLM-1.7B-Instruct",
@@ -277,7 +311,8 @@
         "Average": 2.097863247863248,
         "Sentiment": 2.243589743589744,
         "Language understanding": 1.79,
-        "Phraseology": 2.26
+        "Phraseology": 2.26,
+        "Tricky questions": 4.0
     },
     {
         "Model": "LGAI-EXAONE/EXAONE-3.5-2.4B-Instruct",
@@ -285,7 +320,8 @@
         "Average": 2.062846282695529,
         "Sentiment": 1.9423076923076923,
         "Language understanding": 2.1155778894472363,
-        "Phraseology": 2.130653266331658
+        "Phraseology": 2.130653266331658,
+        "Tricky questions": 4.0
     },
     {
         "Model": "HuggingFaceTB/SmolLM2-1.7B-Instruct",
@@ -293,7 +329,8 @@
         "Average": 1.9102136752136751,
         "Sentiment": 2.275641025641025,
         "Language understanding": 1.1,
-        "Phraseology": 2.355
+        "Phraseology": 2.355,
+        "Tricky questions": 4.0
     },
     {
         "Model": "Qwen/Qwen2.5-0.5B-Instruct",
@@ -301,7 +338,8 @@
         "Average": 1.7950427350427354,
         "Sentiment": 1.955128205128205,
         "Language understanding": 0.835,
-        "Phraseology": 2.595
+        "Phraseology": 2.595,
+        "Tricky questions": 4.0
     },
     {
         "Model": "CYFRAGOVPL/Llama-PLLuM-70B-chat",
@@ -309,7 +347,8 @@
         "Average": 3.63,
         "Sentiment": 3.94,
         "Language understanding": 3.61,
-        "Phraseology": 3.35
+        "Phraseology": 3.35,
+        "Tricky questions": 4.0
     },
     {
         "Model": "CYFRAGOVPL/PLLuM-8x7B-nc-instruct",
@@ -317,7 +356,8 @@
         "Average": 3.56,
         "Sentiment": 3.88,
         "Language understanding": 3.59,
-        "Phraseology": 3.22
+        "Phraseology": 3.22,
+        "Tricky questions": 4.0
     },
     {
         "Model": "CYFRAGOVPL/Llama-PLLuM-70B-instruct",
@@ -325,15 +365,17 @@
         "Average": 3.56,
         "Sentiment": 3.78,
         "Language understanding": 3.63,
-        "Phraseology": 3.26
+        "Phraseology": 3.26,
+        "Tricky questions": 4.0
     },
     {
         "Model": "CYFRAGOVPL/PLLuM-8x7B-instruct",
         "Params": "46.7B",
-        "Average": 3.50,
+        "Average": 3.5,
         "Sentiment": 3.59,
         "Language understanding": 3.47,
-        "Phraseology": 3.46
+        "Phraseology": 3.46,
+        "Tricky questions": 4.0
     },
     {
         "Model": "CYFRAGOVPL/PLLuM-12B-instruct",
@@ -341,7 +383,8 @@
         "Average": 3.49,
         "Sentiment": 3.71,
         "Language understanding": 3.17,
-        "Phraseology": 3.59
+        "Phraseology": 3.59,
+        "Tricky questions": 4.0
     },
     {
         "Model": "CYFRAGOVPL/PLLuM-8x7B-nc-chat",
@@ -349,7 +392,8 @@
         "Average": 3.44,
         "Sentiment": 3.76,
         "Language understanding": 3.48,
-        "Phraseology": 3.08
+        "Phraseology": 3.08,
+        "Tricky questions": 4.0
     },
     {
         "Model": "CYFRAGOVPL/PLLuM-8x7B-chat",
@@ -357,7 +401,8 @@
         "Average": 3.41,
         "Sentiment": 3.44,
         "Language understanding": 3.45,
-        "Phraseology": 3.35
+        "Phraseology": 3.35,
+        "Tricky questions": 4.0
     },
     {
         "Model": "CYFRAGOVPL/PLLuM-12B-chat",
@@ -365,7 +410,8 @@
         "Average": 3.32,
         "Sentiment": 3.32,
         "Language understanding": 3.21,
-        "Phraseology": 3.43
+        "Phraseology": 3.43,
+        "Tricky questions": 4.0
     },
     {
         "Model": "CYFRAGOVPL/PLLuM-12B-nc-instruct",
@@ -373,15 +419,17 @@
         "Average": 3.29,
         "Sentiment": 3.24,
         "Language understanding": 3.31,
-        "Phraseology": 3.32
+        "Phraseology": 3.32,
+        "Tricky questions": 4.0
     },
     {
         "Model": "CYFRAGOVPL/Llama-PLLuM-8B-instruct",
         "Params": "8.03B",
-        "Average": 3.20,
+        "Average": 3.2,
         "Sentiment": 3.24,
-        "Language understanding": 2.90,
-        "Phraseology": 3.46
+        "Language understanding": 2.9,
+        "Phraseology": 3.46,
+        "Tricky questions": 4.0
     },
     {
         "Model": "CYFRAGOVPL/Llama-PLLuM-8B-chat",
@@ -389,7 +437,8 @@
         "Average": 3.14,
         "Sentiment": 3.13,
         "Language understanding": 2.93,
-        "Phraseology": 3.36
+        "Phraseology": 3.36,
+        "Tricky questions": 4.0
     },
     {
         "Model": "CYFRAGOVPL/PLLuM-12B-nc-chat",
@@ -397,7 +446,8 @@
         "Average": 3.33,
         "Sentiment": 3.22,
         "Language understanding": 3.23,
-        "Phraseology": 3.54
+        "Phraseology": 3.54,
+        "Tricky questions": 4.0
     },
     {
         "Model": "Qwen/Qwen2.5-72B-Instruct",
@@ -405,7 +455,8 @@
         "Average": 3.9923076923076923,
         "Sentiment": 4.076923076923077,
         "Language understanding": 3.97,
-        "Phraseology": 3.93
+        "Phraseology": 3.93,
+        "Tricky questions": 4.0
     },
     {
         "Model": "Qwen/Qwen2.5-32B-Instruct",
@@ -413,7 +464,8 @@
         "Average": 3.8047008547008545,
         "Sentiment": 3.8141025641025643,
         "Language understanding": 3.565,
-        "Phraseology": 4.035
+        "Phraseology": 4.035,
+        "Tricky questions": 4.0
     },
     {
         "Model": "mistralai/Mistral-Small-24B-Instruct-2501",
@@ -421,7 +473,8 @@
         "Average": 3.79508547008547,
         "Sentiment": 3.91025641025641,
         "Language understanding": 3.6,
-        "Phraseology": 3.875
+        "Phraseology": 3.875,
+        "Tricky questions": 4.0
     },
     {
         "Model": "meta-llama/Llama-3.3-70B-Instruct",
@@ -429,7 +482,8 @@
         "Average": 3.7332905982905977,
         "Sentiment": 4.294871794871795,
         "Language understanding": 3.865,
-        "Phraseology": 3.04
+        "Phraseology": 3.04,
+        "Tricky questions": 4.0
     },
     {
         "Model": "Qwen/Qwen2.5-14B-Instruct",
@@ -437,7 +491,8 @@
         "Average": 3.61508547008547,
         "Sentiment": 3.91025641025641,
         "Language understanding": 3.565,
-        "Phraseology": 3.37
+        "Phraseology": 3.37,
+        "Tricky questions": 4.0
     },
     {
         "Model": "microsoft/phi-4",
@@ -445,7 +500,8 @@
         "Average": 3.4976495726495727,
         "Sentiment": 3.717948717948718,
         "Language understanding": 3.54,
-        "Phraseology": 3.235
+        "Phraseology": 3.235,
+        "Tricky questions": 4.0
     },
     {
         "Model": "Qwen/Qwen2.5-7B-Instruct",
@@ -453,7 +509,8 @@
         "Average": 3.2258974358974357,
         "Sentiment": 3.5576923076923075,
         "Language understanding": 3.025,
-        "Phraseology": 3.095
+        "Phraseology": 3.095,
+        "Tricky questions": 4.0
     },
     {
         "Model": "microsoft/Phi-4-mini-instruct",
@@ -461,6 +518,7 @@
         "Average": 2.455769230769231,
         "Sentiment": 2.6923076923076925,
         "Language understanding": 2.43,
-        "Phraseology": 2.245
+        "Phraseology": 2.245,
+        "Tricky questions": 4.0
     }
 ]