PaperWeave

Graph-grounded benchmarking for scientific paper reasoning

Compare LLM-only, Basic RAG, and TigerGraph GraphRAG on the same question. Evaluation metrics compute live on each generated answer.

Pipelines
3
Domain
NLP / LLM papers
Core metric
Token reduction
Backend
TigerGraph GraphRAG

LLM-only Baseline

idle

Run a query to render this pipeline's answer, token usage, timing breakdown, and evidence.

Tokens
-
Latency
-
Cost
-
Sources
-
BERTSim
-
Judge
-
Hallucination
-
Retrieval quality
-

Basic RAG

idle

Run a query to render this pipeline's answer, token usage, timing breakdown, and evidence.

Tokens
-
Latency
-
Cost
-
Sources
-
BERTSim
-
Judge
-
Hallucination
-
Retrieval quality
-

TigerGraph GraphRAG

idle

Run a query to render this pipeline's answer, token usage, timing breakdown, and evidence.

Tokens
-
Latency
-
Cost
-
Sources
-
BERTSim
-
Judge
-
Hallucination
-
Retrieval quality
-

Live Comparison: Tokens, Latency, Accuracy, Hallucination

Evidence Graph Preview

Graph-connected evidence will appear here after a GraphRAG run.

Basic RAG Evidence

No sources returned yet.

GraphRAG Evidence

No sources returned yet.