Sentiment Analysis on FOMC
71.08AccuracyDCS
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| DCSModel=Qwen3-4B2026.03 | 71.08 | 73.33 | — | — | — | |
| GPT-4Category=Original, Student=GPT-42025.12 | 71 | — | — | 84.32 | — | |
| BRIDGETeacher=GPT-4, TA=None, Student=Qwen2.5-7B-Instr.2025.12 | 70.8 | — | — | 74.22 | — | |
| DCSModel=DeepSeek-R1-14B2026.03 | 65.06 | 73.87 | — | — | — | |
| LLM JudgeModel=Qwen3-4B2026.03 | 63.86 | 69.39 | — | — | — | |
| Linear ProbeModel=Qwen3-4B2026.03 | 61.45 | 61.9 | — | — | — | |
| BRIDGETeacher=GPT-4, TA=None, Student=Mistral-7B-Instr.-v0.32025.12 | 60.5 | — | — | 57.94 | — | |
| DCSModel=Llama-3.1-8B2026.03 | 60.24 | 63.74 | — | — | — | |
| DCSModel=Llama-3.2-1B2026.03 | 59.04 | 71.19 | — | — | — | |
| Linear ProbeModel=Llama-3.1-8B2026.03 | 59.04 | 62.22 | — | — | — | |
| Logit-Based JudgeModel=Qwen3-4B2026.03 | 57.83 | 63.16 | — | — | — | |
| Linear ProbeModel=Llama-3.2-1B2026.03 | 55.42 | 66.67 | — | — | — | |
| Linear ProbeModel=DeepSeek-R1-14B2026.03 | 55.42 | 61.86 | — | — | — | |
| Qwen2.5-7B-Instr.Category=Original, Student=Qwen2.5-7B-Instr.2025.12 | 55.33 | — | — | 55.1 | — | |
| LLM JudgeModel=Llama-3.1-8B2026.03 | 54.22 | 64.15 | — | — | — | |
| Logit-Based JudgeModel=Llama-3.2-1B2026.03 | 53.01 | 69.29 | — | — | — | |
| Logit-Based JudgeModel=Llama-3.1-8B2026.03 | 53.01 | 69.29 | — | — | — | |
| LLM JudgeModel=DeepSeek-R1-14B2026.03 | 53.01 | 69.29 | — | — | — | |
| Logit-Based JudgeModel=DeepSeek-R1-14B2026.03 | 53.01 | 69.29 | — | — | — | |
| LLM JudgeModel=Llama-3.2-1B2026.03 | 50.6 | 53.93 | — | — | — | |
| Mistral-7B-Instr.-v0.3Category=Original, Student=Mistral-7B-Instr.-v0.32025.12 | 45.59 | — | — | 42.32 | — | |
| DictionaryModel=-2026.03 | 44.58 | 58.93 | — | — | — | |
| SupervisedModel=FOMC-RoBERTa2026.03 | 43.37 | 53.46 | — | — | — | |
| BRIDGETeacher=GPT-4, TA=Mistral-7B-Instr.-v0.3, Student=Pythia-410M2025.12 | 36.5 | — | — | 33.97 | — | |
| BRIDGETeacher=GPT-4, TA=Qwen2.5-7B-Instr., Student=Pythia-410M2025.12 | 35.6 | — | — | 33.1 | — | |
| BRIDGETeacher=GPT-4, TA=Qwen2.5-7B-Instr., Student=OPT-350M2025.12 | 34.5 | — | — | 33.82 | — | |
| BRIDGETeacher=GPT-4, TA=Qwen2.5-7B-Instr., Student=Bloomz-560M2025.12 | 33.19 | — | — | 32.66 | — | |
| BRIDGETeacher=GPT-4, TA=Mistral-7B-Instr.-v0.3, Student=Bloomz-560M2025.12 | 32.9 | — | — | 31.64 | — | |
| ULDTA=Qwen2.5-7B-Instr., Student=Pythia-410M2025.12 | 31.8 | — | — | 27.98 | — | |
| BRIDGETeacher=GPT-4, TA=Mistral-7B-Instr.-v0.3, Student=OPT-350M2025.12 | 31.4 | — | — | 31.14 | — | |
| ULDTA=Mistral-7B-Instr.-v0.3, Student=Pythia-410M2025.12 | 30.9 | — | — | 27.26 | — | |
| Pythia-410MCategory=Original, Student=Pythia-410M2025.12 | 30.13 | — | — | 26.89 | — | |
| BLKDTeacher=GPT-4, Student=Bloomz-560M2025.12 | 28.93 | — | — | 27.87 | — | |
| BLKDTeacher=GPT-4, Student=Pythia-410M2025.12 | 26.65 | — | — | 26.42 | — | |
| BLKDTeacher=GPT-4, Student=OPT-350M2025.12 | 26.4 | — | — | 26.77 | — | |
| ULDTA=Mistral-7B-Instr.-v0.3, Student=OPT-350M2025.12 | 24.3 | — | — | 24.9 | — | |
| ULDTA=Qwen2.5-7B-Instr., Student=OPT-350M2025.12 | 23.8 | — | — | 24.75 | — | |
| ULDTA=Qwen2.5-7B-Instr., Student=Bloomz-560M2025.12 | 23.5 | — | — | 24.74 | — | |
| LionTeacher=GPT-4, Student=Bloomz-560M2025.12 | 23.4 | — | — | 24.2 | — | |
| LionTeacher=GPT-4, Student=OPT-350M2025.12 | 23.4 | — | — | 21.1 | — | |
| ULDTA=Mistral-7B-Instr.-v0.3, Student=Bloomz-560M2025.12 | 23.3 | — | — | 24.8 | — | |
| OPT-350MCategory=Original, Student=OPT-350M2025.12 | 22.85 | — | — | 23.88 | — | |
| Bloomz-560MCategory=Original, Student=Bloomz-560M2025.12 | 22.19 | — | — | 24.16 | — | |
| LionTeacher=GPT-4, Student=Pythia-410M2025.12 | 18.5 | — | — | 20.77 | — | |
| BERT2023.10 | — | 63.81 | — | — | — | |
| Dianjin-R1-7BDomain Focus=Financial2026.03 | — | — | — | 70.3 | 60.8 | |
| FiLM# Tokens (Financial Corpus)=2.4B2023.10 | — | 69.6 | — | — | — | |
| FiLM (5.5B)# Tokens (Financial Corpus)=5.5B2023.10 | — | 69.16 | — | — | — | |
| Fin-R1Domain Focus=Financial2026.03 | — | — | — | 61.4 | 50.2 | |
| FinBERT-A# Tokens (Financial Corpus)=237M2023.10 | — | 64.5 | — | — | — | |
| FinBERT-Y# Tokens (Financial Corpus)=4.9B2023.10 | — | 64.3 | — | — | — | |
| FinMA-7B-fullDomain Focus=Financial2026.03 | — | — | — | — | 45.9 | |
| FLANG-BERT# Tokens (Financial Corpus)=NA2023.10 | — | 64.93 | — | — | — | |
| FLANG-ROBERTa# Tokens (Financial Corpus)=NA2023.10 | — | 68.02 | — | — | — | |
| Gemini 2.5 Flash-LiteReasoning Capability=with Reasoning, Domain Focus=General2026.03 | — | — | — | 67.7 | 62.5 | |
| GPT-5 mini-highReasoning Capability=with Reasoning, Domain Focus=General2026.03 | — | — | — | 77.1 | 66.7 | |
| Llama-3.1-8B-InstructReasoning Capability=without Reasoning, Domain Focus=General2026.03 | — | — | — | 56.9 | 43.8 | |
| ODA-Fin-RL-8BDomain Focus=Financial2026.03 | — | — | — | 74.6 | 61 | |
| ODA-Fin-SFT-8BDomain Focus=Financial2026.03 | — | — | — | 72.1 | 63.9 | |
| Plutus-8B-InstructDomain Focus=Financial2026.03 | — | — | — | 36.2 | 43.5 | |
| Qwen2.5-7BReasoning Capability=without Reasoning, Domain Focus=General2026.03 | — | — | — | 47.8 | 26.5 | |
| Qwen2.5-7B-InstructReasoning Capability=without Reasoning, Domain Focus=General2026.03 | — | — | — | 68 | 56.4 | |
| Qwen3-32BReasoning Capability=without Reasoning, Domain Focus=General2026.03 | — | — | — | 74.7 | 61.7 | |
| Qwen3-4B-ThinkingReasoning Capability=with Reasoning, Domain Focus=General2026.03 | — | — | — | 72.5 | 63.8 | |
| Qwen3-8BReasoning Capability=without Reasoning, Domain Focus=General2026.03 | — | — | — | 71.5 | 57.5 | |
| ROBERTa2023.10 | — | 69.16 | — | — | — | |
| SEC-BERT# Tokens (Financial Corpus)=3.1B2023.10 | — | 65.07 | — | — | — | |
| Xuanyuan-6B-ChatDomain Focus=Financial2026.03 | — | — | — | 36.7 | 43.5 |