List your company
ResearchFormerly WizWand

Long-context Question Answering on NarrativeQA

53.56F1 ScoreMiA-RAG
-2.142412.318826.7841.2412Sep 11, 2024Nov 27, 2024Feb 12, 2025May 1, 2025Jul 17, 2025Oct 2, 2025Dec 19, 2025
Updated 8mo ago

Evaluation results

MethodLinks
MiA-RAGRetriever=MiA-Emb-8B, +S (Retriever)=true, Generator=MiA-Gen-14B, +S (Generator)=true, top-k=102025.1253.56
MiA-RAGRetriever=MiA-Emb-8B, +S (Retriever)=true, Generator=MiA-Gen-14B, +S (Generator)=true, top-k=52025.1253.52
MiARetriever=MiA-Emb-8B, +S (Retriever)=true, Generator=Qwen2.5-72B, +S (Generator)=true, top-k=102025.1253.15
MiA-RAGRetriever=MiA-Emb-8B, +S (Retriever)=true, Generator=MiA-Gen-14B, +S (Generator)=true, top-k=32025.1252.48
MiA (Gen-Only)Retriever=Qwen3-8B, +S (Retriever)=false, Generator=MiA-Gen-14B, +S (Generator)=true, top-k=102025.1251.99
MiA (Gen-Only)Retriever=Qwen3-8B, +S (Retriever)=false, Generator=Qwen2.5-72B, +S (Generator)=true, top-k=102025.1251.81
MiARetriever=MiA-Emb-8B, +S (Retriever)=true, Generator=Qwen2.5-72B, +S (Generator)=true, top-k=52025.1251.04
MiA (Gen-Only)Retriever=Qwen3-8B, +S (Retriever)=false, Generator=MiA-Gen-14B, +S (Generator)=true, top-k=32025.1250.55
MiA (Gen-Only)Retriever=Qwen3-8B, +S (Retriever)=false, Generator=MiA-Gen-14B, +S (Generator)=true, top-k=52025.1250.08
MiARetriever=MiA-Emb-8B, +S (Retriever)=true, Generator=Qwen2.5-72B, +S (Generator)=true, top-k=32025.1250.05
MiA (Emb-Only)Retriever=MiA-Emb-8B, +S (Retriever)=true, Generator=Qwen2.5-72B, +S (Generator)=false, top-k=102025.1249.88
VanillaRetriever=Qwen3-8B, +S (Retriever)=false, Generator=Qwen2.5-72B, +S (Generator)=false, top-k=102025.1249.06
MiA (Gen-Only)Retriever=Qwen3-8B, +S (Retriever)=false, Generator=Qwen2.5-72B, +S (Generator)=true, top-k=52025.1248.46
MiA (Emb-Only)Retriever=MiA-Emb-8B, +S (Retriever)=true, Generator=Qwen2.5-72B, +S (Generator)=false, top-k=52025.1248.06
MiARetriever=MiA-Emb-8B, +S (Retriever)=true, Generator=Qwen2.5-14B, +S (Generator)=true, top-k=102025.1247.87
MiA (Gen-Only)Retriever=Qwen3-8B, +S (Retriever)=false, Generator=Qwen2.5-72B, +S (Generator)=true, top-k=32025.1247.67
MiA (Emb-Only)Retriever=MiA-Emb-8B, +S (Retriever)=true, Generator=Qwen2.5-14B, +S (Generator)=false, top-k=102025.1247.13
MiA (Emb-Only)Retriever=MiA-Emb-8B, +S (Retriever)=true, Generator=Qwen2.5-14B, +S (Generator)=false, top-k=52025.1246.77
MiARetriever=MiA-Emb-8B, +S (Retriever)=true, Generator=Qwen2.5-14B, +S (Generator)=true, top-k=52025.1246.66
MiA (Emb-Only)Retriever=MiA-Emb-8B, +S (Retriever)=true, Generator=Qwen2.5-72B, +S (Generator)=false, top-k=32025.1246.38
MiA (Emb-Only)Retriever=MiA-Emb-8B, +S (Retriever)=true, Generator=Qwen2.5-14B, +S (Generator)=false, top-k=32025.1245.89
MiA (Gen-Only)Retriever=Qwen3-8B, +S (Retriever)=false, Generator=Qwen2.5-14B, +S (Generator)=true, top-k=102025.1245.83
VanillaRetriever=Qwen3-8B, +S (Retriever)=false, Generator=Qwen2.5-72B, +S (Generator)=false, top-k=52025.1245.51
VanillaRetriever=Qwen3-8B, +S (Retriever)=false, Generator=Qwen2.5-14B, +S (Generator)=false, top-k=102025.1244.38
MiARetriever=MiA-Emb-8B, +S (Retriever)=true, Generator=Qwen2.5-14B, +S (Generator)=true, top-k=32025.1244.38
MiA (Gen-Only)Retriever=Qwen3-8B, +S (Retriever)=false, Generator=Qwen2.5-14B, +S (Generator)=true, top-k=52025.1243.32
MiA (Gen-Only)Retriever=Qwen3-8B, +S (Retriever)=false, Generator=Qwen2.5-14B, +S (Generator)=true, top-k=32025.1243.04
VanillaRetriever=Qwen3-8B, +S (Retriever)=false, Generator=Qwen2.5-72B, +S (Generator)=false, top-k=32025.1241.13
VanillaRetriever=Qwen3-8B, +S (Retriever)=false, Generator=Qwen2.5-14B, +S (Generator)=false, top-k=52025.1240.99
VanillaRetriever=Qwen3-8B, +S (Retriever)=false, Generator=Qwen2.5-14B, +S (Generator)=false, top-k=32025.1239.32
Summary-OnlyGenerator=Qwen2.5-72B, +S (Generator)=true2025.1239.24
Summary-OnlyGenerator=Qwen2.5-14B, +S (Generator)=true2025.1238.03
MambaTraining Protocol=Trained from scratch, Training Tokens=>1T, Input Context Length=16K2024.0927.9
MistralTraining Protocol=Trained from scratch, Training Tokens=>1T, Input Context Length=16K2024.0925.1
GSATraining Protocol=Finetuned from Mistral 7B, Training Tokens=20B, Input Context Length=16K, Base Model=Mistral 7B2024.0919.2
GLATraining Protocol=Finetuned from Mistral 7B, Training Tokens=20B, Input Context Length=16K, Base Model=Mistral 7B2024.0917.2
RWKV6Training Protocol=Trained from scratch, Training Tokens=>1T, Input Context Length=16K2024.0914.4
RetNetTraining Protocol=Finetuned from Mistral 7B, Training Tokens=20B, Input Context Length=16K, Base Model=Mistral 7B2024.090