Fingerprint Robustness Evaluation on Fingerprint Evaluation Set
100FSR BLEUP2F
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| P2FQuantization=32-bit (Default), Model Family=Llama, Model Variant=Base, Model Size=1B2026.05 | 100 | 100 | 100 | |
| P2FQuantization=16-bit, Model Family=Llama, Model Variant=Base, Model Size=1B2026.05 | 100 | 100 | 99.7 | |
| P2FQuantization=8-bit, Model Family=Llama, Model Variant=Base, Model Size=1B2026.05 | 99.7 | 100 | 100 | |
| P2FQuantization=32-bit (Default), Model Family=Qwen, Model Variant=Base, Model Size=7B2026.05 | 99.3 | 99 | 99 | |
| P2FQuantization=16-bit, Model Family=Llama, Model Variant=Base, Model Size=3B2026.05 | 99.3 | 99.3 | 99.3 | |
| P2FQuantization=32-bit (Default), Model Family=Qwen, Model Variant=Base, Model Size=0.5B2026.05 | 99 | 99 | 99 | |
| P2FQuantization=32-bit (Default), Model Family=Llama, Model Variant=Base, Model Size=3B2026.05 | 99 | 99.3 | 99.3 | |
| P2FQuantization=32-bit (Default), Model Family=Llama, Model Variant=Instruct, Model Size=1B2026.05 | 99 | 99 | 99 | |
| P2FQuantization=32-bit (Default), Model Family=Mistral, Model Variant=Base, Model Size=7B2026.05 | 99 | 99 | 99 | |
| P2FQuantization=16-bit, Model Family=Qwen, Model Variant=Base, Model Size=1.5B2026.05 | 99 | 99 | 99 | |
| P2FQuantization=16-bit, Model Family=Llama, Model Variant=Instruct, Model Size=1B2026.05 | 99 | 99 | 99 | |
| P2FQuantization=8-bit, Model Family=Qwen, Model Variant=Base, Model Size=3B2026.05 | 99 | 99 | 99.3 | |
| P2FQuantization=8-bit, Model Family=Llama, Model Variant=Instruct, Model Size=1B2026.05 | 99 | 98.7 | 98.7 | |
| P2FQuantization=32-bit (Default), Model Family=Qwen, Model Variant=Base, Model Size=1.5B2026.05 | 98.7 | 98.7 | 98.7 | |
| P2FQuantization=32-bit (Default), Model Family=Llama, Model Variant=Base, Model Size=7B2026.05 | 98.7 | 98.7 | 98.7 | |
| P2FQuantization=16-bit, Model Family=Qwen, Model Variant=Base, Model Size=0.5B2026.05 | 98.7 | 98.7 | 98.7 | |
| P2FQuantization=16-bit, Model Family=Qwen, Model Variant=Base, Model Size=7B2026.05 | 98.7 | 98.7 | 98.7 | |
| P2FQuantization=8-bit, Model Family=Qwen, Model Variant=Base, Model Size=7B2026.05 | 98.7 | 99 | 99 | |
| P2FQuantization=32-bit (Default), Model Family=Qwen, Model Variant=Base, Model Size=3B2026.05 | 98.3 | 99 | 99 | |
| P2FQuantization=32-bit (Default), Model Family=Qwen, Model Variant=Instruct, Model Size=1.5B2026.05 | 98.3 | 98.3 | 98.3 | |
| P2FQuantization=32-bit (Default), Model Family=Qwen, Model Variant=Instruct, Model Size=3B2026.05 | 98.3 | 98.3 | 98.3 | |
| P2FQuantization=32-bit (Default), Model Family=Llama, Model Variant=Instruct, Model Size=3B2026.05 | 98.3 | 98 | 98 | |
| P2FQuantization=16-bit, Model Family=Qwen, Model Variant=Base, Model Size=3B2026.05 | 98.3 | 99 | 99 | |
| P2FQuantization=16-bit, Model Family=Qwen, Model Variant=Instruct, Model Size=3B2026.05 | 98.3 | 98.3 | 98.3 | |
| P2FQuantization=16-bit, Model Family=Llama, Model Variant=Base, Model Size=7B2026.05 | 98.3 | 98.7 | 98.7 | |
| P2FQuantization=16-bit, Model Family=Llama, Model Variant=Instruct, Model Size=3B2026.05 | 98.3 | 98.3 | 98.3 | |
| P2FQuantization=16-bit, Model Family=Mistral, Model Variant=Base, Model Size=7B2026.05 | 98.3 | 98.3 | 98.3 | |
| P2FQuantization=8-bit, Model Family=Qwen, Model Variant=Instruct, Model Size=3B2026.05 | 98.3 | 98.3 | 98.3 | |
| P2FQuantization=8-bit, Model Family=Llama, Model Variant=Base, Model Size=3B2026.05 | 98.3 | 98.7 | 98.7 | |
| P2FQuantization=8-bit, Model Family=Llama, Model Variant=Base, Model Size=7B2026.05 | 98.3 | 98.7 | 98.7 | |
| P2FQuantization=8-bit, Model Family=Llama, Model Variant=Instruct, Model Size=3B2026.05 | 98.3 | 98 | 98 | |
| P2FQuantization=8-bit, Model Family=Mistral, Model Variant=Base, Model Size=7B2026.05 | 98.3 | 98.7 | 98.7 | |
| P2FQuantization=16-bit, Model Family=Qwen, Model Variant=Instruct, Model Size=1.5B2026.05 | 98 | 98 | 98 | |
| P2FQuantization=8-bit, Model Family=Qwen, Model Variant=Base, Model Size=1.5B2026.05 | 98 | 98.3 | 98.3 | |
| P2FQuantization=32-bit (Default), Model Family=Qwen, Model Variant=Instruct, Model Size=0.5B2026.05 | 97.7 | 99 | 99 | |
| P2FQuantization=32-bit (Default), Model Family=Qwen, Model Variant=Instruct, Model Size=7B2026.05 | 97.7 | 97.3 | 97.3 | |
| P2FQuantization=8-bit, Model Family=Qwen, Model Variant=Base, Model Size=0.5B2026.05 | 97.7 | 97.7 | 97.7 | |
| P2FQuantization=16-bit, Model Family=Qwen, Model Variant=Instruct, Model Size=7B2026.05 | 97.3 | 97.3 | 97.3 | |
| P2FQuantization=8-bit, Model Family=Qwen, Model Variant=Instruct, Model Size=0.5B2026.05 | 97.3 | 97.3 | 97.7 | |
| P2FQuantization=16-bit, Model Family=Qwen, Model Variant=Instruct, Model Size=0.5B2026.05 | 97 | 97.3 | 97.3 | |
| P2FQuantization=8-bit, Model Family=Qwen, Model Variant=Instruct, Model Size=1.5B2026.05 | 97 | 97.3 | 97.3 | |
| P2FQuantization=16-bit, Model Family=Llama, Model Variant=Instruct, Model Size=7B2026.05 | 96.3 | 97 | 97 | |
| P2FQuantization=8-bit, Model Family=Qwen, Model Variant=Instruct, Model Size=7B2026.05 | 96.3 | 97 | 97 | |
| P2FQuantization=32-bit (Default), Model Family=Llama, Model Variant=Instruct, Model Size=7B2026.05 | 96 | 96.3 | 96.3 | |
| P2FQuantization=8-bit, Model Family=Llama, Model Variant=Instruct, Model Size=7B2026.05 | 94.3 | 94.7 | 94.7 |