updates

app.py

@@ -48,6 +48,8 @@ imgs = glob.glob(os.path.join(data_path, '*.png'))
 # Extracting images that start with "hf_llm_diagram"
 hf_llm_diagrams = [img for img in imgs if 'hf_llm_diagram' in os.path.basename(img)]
 bigcode_diagrams = [img for img in imgs if 'bigcode' in os.path.basename(img)]
+mt_bench_diagrams = [img for img in imgs if 'mt_bench_leaderboard' in os.path.basename(img)]
+opencompass_diagrams = [img for img in imgs if 'opencompass_leaderboard' in os.path.basename(img)]
 
 # Getting the remaining images
 remaining_imgs = list(set(imgs) - set(hf_llm_diagrams) - set(bigcode_diagrams))
@@ -129,7 +131,16 @@ cols[0].image(bigcode_diagrams[0], use_column_width="auto")
 
 print_model_list(bigcode_diagrams[0],st,True)
 
+st.subheader("MT-Bench Models Leaderboard", divider=True)
+cols = st.columns(2)
+cols[0].image(mt_bench_diagrams[0], use_column_width="auto")
+
+print_model_list(mt_bench_diagrams[0],st,True)
 
+st.subheader("OpenCompass Models Leaderboard", divider=True)
+cols = st.columns(2)
+cols[0].image(opencompass_diagrams[0], use_column_width="auto")
+print_model_list(opencompass_diagrams[0],st,True)
 
 st.subheader("HuggingFace and Other Leaderboards: A Comparative Model Evaluation", divider=True)
 st.caption("Only models evaluated on both leaderboards are included.")

data/20230907_1306/mt_bench_leaderboard.json

@@ -0,0 +1 @@
+["WizardLM/WizardLM-70B-V1.0", "lmsys/vicuna-33b-v1.3", "WizardLM/WizardLM-13B-V1.2", "meta-llama/Llama-2-7b-chat-hf", "THUDM/chatglm2-6b", "lmsys/fastchat-t5-3b-v1.0", "BlinkDL/rwkv-4-raven"]