List your company
ResearchFormerly WizWand

Question Answering on Musique (EM, F1, Time, Rate)

22.92EMLatentRAG△
1.16326.811612.4618.1084Mar 10, 2026Mar 23, 2026Apr 5, 2026Apr 19, 2026May 2, 2026May 15, 2026May 29, 2026
Updated 5mo ago

Evaluation results

MethodLinks
LatentRAG△Retrieval Model=jina-embeddings-v5-text-nano2026.0522.92—0.54—
LatentRAG♢Retrieval Model=jina-embeddings-v5-text-nano2026.0522.26—0.632—
AutoRefine△Retrieval Model=e5-base-v22026.0521.8—5.583—
LatentRAG△Retrieval Model=e5-base-v22026.0520.94—0.548—
All-pool PortfolioModel=Llama-3.1-70B-Instruct, k=4, l=32026.0520.89———
LatentRAG△Retrieval Model=Qwen3-Embedding-0.6B2026.0520.73—0.639—
Vendi-RAG Adaptive (T=20) (MPNet)Model=Llama-3.1-70B-Instruct, T=202026.0520.6———
LatentRAG△Retrieval Model=harrier-oss-v1-270m2026.0520.56—0.579—
AutoRefine△Retrieval Model=jina-embeddings-v5-text-nano2026.0520.52—5.614—
Search-R1♢Retrieval Model=jina-embeddings-v5-text-nano2026.0520.4—6.177—
SAPO-3B-InstructRetrieval Strategy=Agent, Backbone Model=Qwen2.5-3B2026.0320.3———
Search-R1♢Retrieval Model=e5-base-v22026.0520.19—6.416—
AutoRefine△Retrieval Model=harrier-oss-v1-270m2026.0520.15—5.511—
LatentRAG♢Retrieval Model=harrier-oss-v1-270m2026.0519.9—0.708—
Vendi-RAG Adaptive (T=10) (MPNet)Model=Llama-3.1-70B-Instruct, T=102026.0519.9———
LatentRAG♢Retrieval Model=Qwen3-Embedding-0.6B2026.0519.82—0.73—
LatentRAG△Retrieval Model=F2LLM-v2-330M2026.0519.82—0.566—
Search-R1♢Retrieval Model=Qwen3-Embedding-0.6B2026.0519.61—6.846—
SAPO-3B-BaseRetrieval Strategy=Agent, Backbone Model=Qwen2.5-3B2026.0319.6———
All-pool PortfolioModel=Gemma-3-27B-It, k=4, l=32026.0519.49———
AutoRefine△Retrieval Model=F2LLM-v2-330M2026.0519.03—5.412—
LatentRAG♢Retrieval Model=e5-base-v22026.0518.87—0.656—
AutoRefine△Retrieval Model=Qwen3-Embedding-0.6B2026.0518.66—5.553—
Search-R1♢Retrieval Model=harrier-oss-v1-270m2026.0518.62—6.613—
Search-R1♢Retrieval Model=F2LLM-v2-330M2026.0518.54—6.69—
LatentRAG♢Retrieval Model=F2LLM-v2-330M2026.0518.37—0.667—
SE-SearchRetrieval Strategy=Agent, Backbone Model=Qwen2.5-3B2026.0318.3———
All-pool PortfolioModel=Llama-3.1-70B-Instruct, k=4, l=22026.0518.2———
StepSearch-BaseRetrieval Strategy=Agent, Backbone Model=Qwen2.5-3B2026.0318.1———
CriticSearchRetrieval Strategy=Agent, Backbone Model=Qwen2.5-3B2026.0318———
R1-SearcherRetrieval Model=Qwen3-Embedding-0.6B2026.0517.83—10.063—
All-pool PortfolioModel=Gemma-3-27B-It, k=4, l=22026.0517.25———
InForageRetrieval Strategy=Agent, Backbone Model=Qwen2.5-3B2026.0317.2———
AutoRefine-InstructRetrieval Strategy=Agent, Backbone Model=Qwen2.5-3B2026.0316.9———
O2-SearcherRetrieval Strategy=Agent, Backbone Model=Qwen2.5-3B2026.0316———
AutoRefine-BaseRetrieval Strategy=Agent, Backbone Model=Qwen2.5-3B2026.0315.7———
Best Vendi RetrieverModel=Gemma-3-27B-It2026.0514.27———
Vendi-Portfolio (MPNet)Model=Llama-3.1-70B-Instruct, k=5, l=22026.0514.27———
Best DS RetrieverModel=Gemma-3-27B-It2026.0513.94———
Nearest-neighbor retrieval (MPNet)Model=Llama-3.1-70B-Instruct2026.0513.94———
EXSEARCH-BaseRetrieval Strategy=Agent, Backbone Model=Qwen2.5-3B2026.0313.8———
DeepRAGRetrieval Model=Qwen3-Embedding-0.6B2026.0513.2—4.149—
Vendi-RAG Adaptive (T=20) (MPNet)Model=Gemma-3-27B-It, T=202026.0513.07———
ZeroSearch-BaseRetrieval Strategy=Agent, Backbone Model=Qwen2.5-3B2026.0313———
Nearest-neighbor retrieval (MPNet)Model=Gemma-3-27B-It2026.0512.91———
Vendi-Portfolio (MPNet)Model=Gemma-3-27B-It, k=5, l=22026.0512.62———
Vendi-RAG Adaptive (T=10) (MPNet)Model=Gemma-3-27B-It, T=102026.0512.53———
Search-R1-InstructRetrieval Strategy=Agent, Backbone Model=Qwen2.5-3B2026.0312.4———
ZeroSearchRetrieval Model=Qwen3-Embedding-0.6B2026.0511.75—5.027—
Best GraphDense RetrieverModel=Gemma-3-27B-It2026.0511.34———
Best Vendi RetrieverModel=Llama-3.1-70B-Instruct2026.0511.21———
Best DS RetrieverModel=Llama-3.1-70B-Instruct2026.0510.92———
Search-o1Retrieval Model=Qwen3-Embedding-0.6B2026.0510.8—6.928—
LooCompReader=Llama-3.3-70B Instruct, Retrieved chunks=Top-202026.0310.720.10.1647
ReSearch-InstructRetrieval Strategy=Agent, Backbone Model=Qwen2.5-3B2026.039.5———
Best GraphDense RetrieverModel=Llama-3.1-70B-Instruct2026.058.73———
ReSearch-BaseRetrieval Strategy=Agent, Backbone Model=Qwen2.5-3B2026.037.4———
Iter-RetGenRetrieval Model=Qwen3-Embedding-0.6B2026.057.16—2.513—
IRCoTRetrieval Strategy=IRCoT, Backbone Model=Qwen2.5-3B2026.036.7———
Search-R1-BaseRetrieval Strategy=Agent, Backbone Model=Qwen2.5-3B2026.036.6———
No retrievalModel=Gemma-3-27B-It2026.056.08———
R1-InstructRetrieval Strategy=None, Backbone Model=Qwen2.5-3B2026.036———
No retrievalModel=Llama-3.1-70B-Instruct2026.055.88———
LooCompReader=Llama-3.1-8B Instruct, Retrieved chunks=Top-52026.035.712.70.03810.8
R1-BaseRetrieval Strategy=None, Backbone Model=Qwen2.5-3B2026.035.5———
Search-o1Retrieval Strategy=Agent, Backbone Model=Qwen2.5-3B2026.035.4———
Naive RAGRetrieval Model=Qwen3-Embedding-0.6B2026.054.96—0.422—
Naive RAGRetrieval Strategy=Naive RAG, Backbone Model=Qwen2.5-3B2026.034.7———
SFTRetrieval Strategy=None, Backbone Model=Qwen2.5-3B2026.034.4———
Direct InferRetrieval Model=N/A2026.053.19—0.239—
Direct GenerationRetrieval Strategy=None, Backbone Model=Qwen2.5-3B2026.032———
GRPOBackbone=Qwen2.5-7B-Instruct2025.10—25——
GSPOBackbone=Qwen2.5-7B-Instruct2025.10—25.4——
IGPOBackbone=Qwen2.5-7B-Instruct2025.10—32.7——
PPOBackbone=Qwen2.5-7B-Instruct2025.10—26.2——
Reinforce++Backbone=Qwen2.5-7B-Instruct2025.10—23.7——
RLOOBackbone=Qwen2.5-7B-Instruct2025.10—24.8——