List your company
ResearchFormerly WizWand

Sentiment Analysis on SST-2 (Accuracy)

97.48AccuracyUD+-XXL
84.916888.178491.4494.7016Nov 15, 2022Jun 5, 2023Dec 24, 2023Jul 13, 2024Jan 31, 2025Aug 21, 2025Mar 11, 2026
Updated 5mo ago

Evaluation results

MethodLinks
UD+-XXLMode=Fully-supervised, Backbone=T5-XXL encoder, Model Parameters=11B2022.1197.48
Tay et al. (2022)Mode=Fully-supervised, Model Parameters=4x of UD (~44B)2022.1197.3
PRewrite-SBackbone=PaLM2-S2026.0396.6
PRewrite-IBackbone=PaLM2-S2026.0396.5
PRewrite (Initial Prompt)Backbone=PaLM2-S2026.0396.3
FADS-ICLBackbone=Llama-1 13B, Shots=128-shots2024.0595.7
ColD-Fusion2022.1295.16
GENICLBackbone=LLaMA-7B, Evaluation Setting=Few-shot (ICL)2025.0595
ICLBackbone=Llama-1 13B, Shots=128-shots2024.0594.9
ICLBackbone=Llama-2 13B, Shots=128-shots2024.0594.9
kNN-promptingBackbone=Llama-2 13B, Shots=128-shots2024.0594.6
kNN-promptBackbone=Llama-2 7B, Shots=128-shots2024.0594.5
Multitask2022.1294.27
ICLBackbone=Llama-1 30B, Shots=128-shots2024.0594
kNN-promptBackbone=Llama-1 13B, Shots=128-shots2024.0593.9
kNN-promptBackbone=Llama-1 30B, Shots=128-shots2024.0593.9
Finetune2022.1293.85
ICRModel=Llama3.1-70B2025.0993.8
ICLBackbone=Llama-2 7B, Shots=128-shots2024.0593.6
kNN-promptingBackbone=Llama-1 13B, Shots=128-shots2024.0593.5
LLM-RBackbone=LLaMA-7B, Evaluation Setting=Few-shot (ICL)2025.0593.4
ICLBackbone=Llama-1 7B, Shots=128-shots2024.0593.3
kNN-promptBackbone=Llama-1 7B, Shots=128-shots2024.0593.3
ICLBackbone=Llama-2 70B, Shots=128-shots2024.0593.2
Zero-shotModel=Llama3.1-70B2025.0993.2
kNN-promptBackbone=Llama-2 13B, Shots=128-shots2024.0593.1
FADS-ICLBackbone=Llama-2 13B, Shots=128-shots2024.0593.1
TIACBMModel=GPT-2, Fine-Tuning Strategy=TIACBM (ours)2025.0292.96
CAMEmode=fine-tuning, batch size=32k2023.0792.9
Baselinemode=fine-tuning2023.0792.8
CAMEmode=fine-tuning, batch size=8k2023.0792.8
kNN-promptingBackbone=Llama-1 7B, Shots=128-shots2024.0592.8
kNN-promptingBackbone=Llama-2 7B, Shots=128-shots2024.0592.8
kNN-promptingBackbone=Llama-1 30B, Shots=128-shots2024.0592.8
Cyclic DecayingModel=GPT-2, Fine-Tuning Strategy=Cyclic Decaying2025.0292.74
DecTn=256, Update model parameters=false2022.1292.7
FADS-ICLBackbone=Llama-1 30B, Shots=128-shots2024.0592.7
FADS-ICLBackbone=Llama-2 70B, Shots=128-shots2024.0592.6
Ankner et al. (2024)Model=GPT-2, Fine-Tuning Strategy=Ankner et al. (2024)2025.0292.6
ConstantModel=GPT-2, Fine-Tuning Strategy=Constant2025.0292.54
Fine-tuningn=64, Update model parameters=true2022.1292.5
DecTn=64, Update model parameters=false2022.1292.4
FADS-ICLBackbone=Llama-1 7B, Shots=128-shots2024.0592.4
E5baseBackbone=LLaMA-7B, Evaluation Setting=Few-shot (ICL)2025.0592.4
ConventionalModel=GPT-2, Fine-Tuning Strategy=Conventional2025.0292.35
Poesina et al. (2024)Model=GPT-2, Fine-Tuning Strategy=Poesina et al. (2024)2025.0292.27
FADS-ICLBackbone=Llama-2 7B, Shots=128-shots2024.0592.2
Rewrite (Pc)Backbone=Gemma-2-9B-IT, Strategy=Score + context rewrite2026.0392.2
Fine-tuningn=256, Update model parameters=true2022.1292
kNN-promptBackbone=Llama-2 70B, Shots=128-shots2024.0592
TEMPERA2026.0392
kNN-promptingBackbone=Llama-2 70B, Shots=128-shots2024.0591.9
DecTn (Shots)=16, Tr. Time (s)=3, # Query=1, # Param. (K)=1302022.1291
Few-shot*Model=Llama3.1-70B2025.0991
DecTn (Shots)=12022.1290.8
GPT Large FTBackbone=GPT Large, Shots=N/A (Fine-tuned)2024.0590.7
FADS-ICLcandidate pool size (m)=2562024.0590.5
BBTv2n (Shots)=16, Tr. Time (s)=9856, # Query=8000, # Param. (K)=122022.1290.3
RLPrompt2026.0390.1
Few-shot*Model=Qwen3-32B2025.0989.8
BBTn (Shots)=16, Tr. Time (s)=10512, # Query=8000, # Param. (K)=0.52022.1289.6
PromptBoostingn (Shots)=42022.1288.9
FADS-ICLNumber of training shots (m)=128, LLM scale=1.5B2024.0588.9
FADS-ICLBackbone=GPT-2 1.5B, Shots=128-shots2024.0588.9
FADS-ICLcandidate pool size (m)=1282024.0588.9
Latent Concept LearningLLM=GPT3-c (6.7B)2023.0188.8
I2CLModel=Llama3.1-70B2025.0988.8
EPRBackbone=LLaMA-7B, Evaluation Setting=Few-shot (ICL)2025.0588.7
SimilarLLM=GPT3-d (175B)2023.0188.5
BERT Large FTBackbone=BERT Large, Shots=N/A (Fine-tuned)2024.0588.3
FADS-ICLNumber of training shots (m)=64, LLM scale=1.5B2024.0588.2
FADS-ICLcandidate pool size (m)=642024.0588.2
FADS-ICLNumber of training shots (m)=32, LLM scale=1.5B2024.0588
kNN-promptBackbone=GPT-2 0.8B, Shots=128-shots2024.0588
FADS-ICLcandidate pool size (m)=322024.0588
Latent Concept LearningLLM=GPT3-d (175B)2023.0187.8
SBERTBackbone=LLaMA-7B, Evaluation Setting=Few-shot (ICL)2025.0587.8
FADS-ICLBackbone=GPT-2 0.8B, Shots=128-shots2024.0587.7
Rewrite (Ps)Backbone=Gemma-2-9B-IT, Strategy=Score-only rewrite2026.0387.7
DecTn (Shots)=42022.1287.6
PromptBoostingn (Shots)=16, Tr. Time (s)=644, # Query=10, # Param. (K)=0.42022.1287.6
Latent Concept LearningLLM=GPT3-b (1.3B)2023.0187.3
ProGenClassifier=DistillBERT2023.0587.2
kNN-promptingNumber of training shots (m)=64, LLM scale=1.5B2024.0587.2
RLPromptn (Shots)=16, Tr. Time (s)=65579, # Query=12000, # Param. (K)=31002022.1287
SuperGenClassifier=DistillBERT2023.0586.7
PromptBoostingn (Shots)=12022.1286.7
BBTv2n (Shots)=42022.1286.6
FADS-ICLNumber of training shots (m)=16, LLM scale=1.5B2024.0586.6
UniformLLM=GPT3-d (175B)2023.0186.5
ICRModel=Qwen3-32B2025.0986.4
FVModel=Llama3.1-70B2025.0986.4
Latent Concept LearningLLM=GPT2-l (774M)2023.0186.2
kNN-promptNumber of training shots (m)=128, LLM scale=1.5B2024.0586.1
kNN-promptBackbone=GPT-2 1.5B, Shots=128-shots2024.0586.1
kNN-promptingNumber of training shots (m)=128, LLM scale=1.5B2024.0586
kNN-promptingBackbone=GPT-2 1.5B, Shots=128-shots2024.0586
SimilarLLM=GPT3-c (6.7B)2023.0185.7
I2CLModel=Qwen3-32B2025.0985.6
Latent Concept LearningLLM=GPT3-a (350M)2023.0185.4