ELECTRA and GPT-4o: Cost-Effective Partners for Sentiment Analysis

Benchmark Model Rank Results
sentiment-analysis-on-dynasentGPT-4o Fine-Tuned (Minimal)#1Macro F1: 89
sentiment-analysis-on-dynasentGPT-4o-mini Fine-Tuned#2Macro F1: 86.9
sentiment-analysis-on-dynasentGPT-4o + ELECTRA Large FT (Prompt, Label, Examples)#3Macro F1: 81.53
sentiment-analysis-on-dynasentGPT-4o (Prompt)#4Macro F1: 80.22
sentiment-analysis-on-dynasentGPT-4o-mini + ELECTRA Large FT (Prompt, Label, Probabilities)#5Macro F1: 79.72
sentiment-analysis-on-dynasentGPT-4o-mini + ELECTRA Large FT (Prompt, Label)#6Macro F1: 77.94
sentiment-analysis-on-dynasentGPT-4o + ELECTRA Large FT#7Macro F1: 77.69
sentiment-analysis-on-dynasentGPT-4o-mini (Prompt)#8Macro F1: 77.35
sentiment-analysis-on-dynasentELECTRA Large Fine-Tuned#9Macro F1: 76.29
sentiment-analysis-on-dynasentGPT-4o-mini + ELECTRA Base FT#10Macro F1: 76.19
sentiment-analysis-on-dynasentELECTRA Base Fine-Tuned#11Macro F1: 71.83
sentiment-analysis-on-sentiment-mergedGPT-4o Fine-Tuned (Minimal)#1Macro F1: 86.99
sentiment-analysis-on-sentiment-mergedGPT-4o-mini Fine-Tuned#2Macro F1: 86.77
sentiment-analysis-on-sentiment-mergedGPT-4o-mini + ELECTRA Large FT (Prompt, Label)#3Macro F1: 83.49
sentiment-analysis-on-sentiment-mergedGPT-4o + ELECTRA Large FT (Prompt, Label, Examples)#4Macro F1: 83.09
sentiment-analysis-on-sentiment-mergedGPT-4o-mini + ELECTRA Base FT (Prompt, Label)#5Macro F1: 82.74
sentiment-analysis-on-sentiment-mergedELECTRA Large Fine-Tuned#6Macro F1: 82.36
sentiment-analysis-on-sentiment-mergedGPT-4o + ELECTRA Large FT (Prompt, Label)#7Macro F1: 81.57
sentiment-analysis-on-sentiment-mergedGPT-4o (Prompt)#8Macro F1: 80.14
sentiment-analysis-on-sentiment-mergedGPT-4o-mini (Prompt)#9Macro F1: 79.52
sentiment-analysis-on-sentiment-mergedELECTRA Base Fine-Tuned#10Macro F1: 79.29
sentiment-analysis-on-sst-3GPT-4o-mini Fine-Tuned#1Macro F1: 75.68
sentiment-analysis-on-sst-3GPT-4o Fine-Tuned (Minimal)#2Macro F1: 73.99
sentiment-analysis-on-sst-3GPT-4o + ELECTRA Large FT#3Macro F1: 72.94
sentiment-analysis-on-sst-3GPT-4o (Prompt)#4Macro F1: 72.2
sentiment-analysis-on-sst-3GPT-4o + ELECTRA Large FT (Prompt, Label, Examples)#5Macro F1: 72.06
sentiment-analysis-on-sst-3GPT-4o-mini + ELECTRA Large FT (Prompt, Label, Examples)#6Macro F1: 71.98
sentiment-analysis-on-sst-3GPT-4o-mini + ELECTRA Base FT#7Macro F1: 71.72
sentiment-analysis-on-sst-3GPT-4o-mini + ELECTRA Large FT (Prompt, Label)#8Macro F1: 70.99
sentiment-analysis-on-sst-3ELECTRA Large Fine-Tuned#9Macro F1: 70.90
sentiment-analysis-on-sst-3GPT-4o-mini (Prompt)#10Macro F1: 70.67
sentiment-analysis-on-sst-3ELECTRA Base Fine-Tuned#11Macro F1: 69.95