List your company
ResearchFormerly WizWand

Document-grounded Utterance Completion

135Model Wins CountLlama (ours)
70.5287.26104120.74Aug 8, 2025
Updated 4mo ago

Evaluation results

MethodLinks
Llama (ours)Evaluation Protocol=vs. Original, Model Variant=Post-trained2025.08135462919—
Gemma (ours)Evaluation Protocol=vs. Original, Model Variant=Post-trained2025.081214529916—
Gemma (ours)Evaluation Protocol=vs. Human, Model Variant=Post-trained2025.08109972591679.1
Llama (ours)Evaluation Protocol=vs. Human, Model Variant=Post-trained2025.08901142572075.3
Gemma (original)Evaluation Protocol=vs. Human, Model Variant=Original2025.08881302441971.9
Llama (original)Evaluation Protocol=vs. Human, Model Variant=Original2025.08831602271166
ClaudeEvaluation Protocol=vs. Human, Model Variant=Off-the-shelf2025.08761682201763.8
GPTEvaluation Protocol=vs. Human, Model Variant=Off-the-shelf2025.08731922031359