Flan-T5 XXL: Complete Benchmark Performance Guide

other flan-t5-xxl 2022-12-07 Apache 2.0

In-depth analysis of Flan-T5 XXL performance on MMLU, HumanEval, GSM8K, MATH, BBH benchmarks, including pricing and use cases.

Key Specifications

Vendorother
Versionflan-t5-xxl
Release Date2022-12-07
Context Window4096 tokens
Input Modalitiestext
Output Modalitiestext
LicenseApache 2.0
Documentationhttps://huggingface.co/models

Benchmark Performance

Benchmark Score Unit Evaluated At Notes Source
MMLU 52 % 2022-12-07 5-shot view
HUMANEVAL 35.4 pass@1 2022-12-07 — view
GSM8K 35.2 % 2022-12-07 0-shot CoT view
MATH 21.4 % 2022-12-07 0-shot CoT view
BBH 60.5 % 2022-12-07 3-shot CoT view
GPQA 28.6 % 2022-12-07 0-shot view
IFEVAL 51.3 % 2022-12-07 prompt_strict view
ARC 81.7 % 2022-12-07 challenge view
MUSR 42.2 % 2022-12-07 0-shot view
WINOGRANDE 74.2 % 2022-12-07 0-shot view

Pricing

Tier Price Currency
Input$0.5 / MtokUSD
Output$0.5 / MtokUSD
Cache Read$0 / MtokUSD
Cache Write$0 / MtokUSD

Source: https://huggingface.co/models · as of 2022-12-07

Compliance

  • Data Residency: self-host
  • SOC2: ✗
  • HIPAA: ✗
  • GDPR: ✗
  • ISO 27001: ✗

Flan-T5 XXL

Model Overview

Google Flan-T5 XXL 11B 指令微调模型, 4K 上下文, 多任务指令微调, 零样本能力突出。

Core Specifications

Vendor Version Release Date Context Window Input Modalities Output Modalities License
Other flan-t5-xxl 2022-12-07 4K text text Apache 2.0

Benchmark Performance

Benchmark Score Unit Notes
MMLU (Massive Multitask Language Understanding) 52.0 % 5-shot
HumanEval 35.4 pass@1 —
GSM8K (Grade School Math 8K) 35.2 % 0-shot CoT
MATH 21.4 % 0-shot CoT
BBH (BIG-Bench Hard) 60.5 % 3-shot CoT
GPQA 28.6 % 0-shot
IFEval 51.3 % prompt_strict
ARC 81.7 % challenge
MUSR 42.2 % 0-shot
WinoGrande 74.2 % 0-shot

Pricing

Input Output Cache Read Cache Write
— — — —

per million tokens

Strengths

  • Reliable general-purpose model.

Weaknesses

  • MMLU 52.0, weak knowledge reasoning.
  • HumanEval 35.4, coding weak.
  • Proprietary, not self-hostable.
  • Context window 4K is limited.

Use Cases

  • General chat and Q&A

References