reading in file urls from a file. Added additional data sources

app.py

@@ -4,39 +4,38 @@ import numpy as np
 import json
 import requests
 
-# URLs for the two JSON files
-FILE_URL_1 = "https://raw.githubusercontent.com/EleutherAI/lm-evaluation-harness/master/results/llama/llama-30B/llama-30B_mmlu_5-shot.json"
-FILE_URL_2 = "https://raw.githubusercontent.com/EleutherAI/lm-evaluation-harness/master/results/llama/llama-13B/llama-13B_mmlu_5-shot.json"
+# Read URLs from a file, one per line
+with open('file_urls.txt', 'r') as f:
+    file_urls = [line.strip() for line in f.readlines()]
 
-# Derive column names from the URLs
-column_name_1 = FILE_URL_1.split('/')[-1].split('_')[0]  # 'llama-30B'
-column_name_2 = FILE_URL_2.split('/')[-1].split('_')[0]  # 'llama-13B'
+dataframes = []
 
-# Load data from both URLs
-response1 = requests.get(FILE_URL_1)
-data1 = response1.json()
-response2 = requests.get(FILE_URL_2)
-data2 = response2.json()
+for url in file_urls:
+    # Derive column names from the URLs
+    column_name = url.split('/')[-1].split('_')[0]
 
-# Convert data from both URLs into DataFrames
-data1_df = pd.DataFrame(data1['results']).T
-data2_df = pd.DataFrame(data2['results']).T
+    # Load data from URL
+    response = requests.get(url)
+    data = response.json()
 
-# Rename 'acc' column to respective file names
-data1_df = data1_df.rename(columns={'acc': column_name_1})
-data2_df = data2_df.rename(columns={'acc': column_name_2})
+    # Convert data into a DataFrame
+    df = pd.DataFrame(data['results']).T
 
-# Remove 'hendrycksTest-' from the index of both dataframes
-data1_df.index = data1_df.index.str.replace('hendrycksTest-', '')
-data2_df.index = data2_df.index.str.replace('hendrycksTest-', '')
+    # Rename 'acc' column to respective file names
+    df = df.rename(columns={'acc': column_name})
 
-# Merge the dataframes on index (Here index is the sub-test names)
-data = pd.merge(data1_df[column_name_1], data2_df[column_name_2], left_index=True, right_index=True)
+    # Remove 'hendrycksTest-' from the index
+    df.index = df.index.str.replace('hendrycksTest-', '')
+
+    dataframes.append(df[[column_name]])  # keep only the column of interest
+
+# Merge the dataframes on index
+data = pd.concat(dataframes, axis=1)
 
 # Transpose the dataframe to swap rows and columns
 data = data.transpose()
 
-# select only columns moral_scenarios and moral_disputes
+# Select only columns 'moral_scenarios' and 'moral_disputes'
 data = data[['moral_scenarios', 'moral_disputes']]
 
 def show_leaderboard():

file_urls.txt

@@ -0,0 +1,12 @@
+https://raw.githubusercontent.com/EleutherAI/lm-evaluation-harness/master/results/llama/llama-30B/llama-30B_mmlu_5-shot.json
+https://raw.githubusercontent.com/EleutherAI/lm-evaluation-harness/master/results/llama/llama-13B/llama-13B_mmlu_5-shot.json
+https://raw.githubusercontent.com/EleutherAI/lm-evaluation-harness/master/results/llama/llama-7B/llama-7B_mmlu_5-shot.json
+https://raw.githubusercontent.com/EleutherAI/lm-evaluation-harness/master/results/mpt/mpt-7b/mpt-7b_mmlu_5-shot.json
+https://raw.githubusercontent.com/EleutherAI/lm-evaluation-harness/master/results/opt/opt-66b/opt-66b.json
+https://raw.githubusercontent.com/EleutherAI/lm-evaluation-harness/master/results/opt/opt-6.7b/opt-6.7b.json
+https://raw.githubusercontent.com/EleutherAI/lm-evaluation-harness/master/results/opt/opt-350m/opt-350m.json
+https://raw.githubusercontent.com/EleutherAI/lm-evaluation-harness/master/results/opt/opt-30b/opt-30b.json
+https://raw.githubusercontent.com/EleutherAI/lm-evaluation-harness/master/results/opt/opt-2.7b/opt-2.7b.json
+https://raw.githubusercontent.com/EleutherAI/lm-evaluation-harness/master/results/opt/opt-13b/opt-13b.json
+https://raw.githubusercontent.com/EleutherAI/lm-evaluation-harness/master/results/opt/opt-125m/opt-125m.json
+https://raw.githubusercontent.com/EleutherAI/lm-evaluation-harness/master/results/opt/opt-1.3b/opt-1.3b.json