MPT 30B: Complete Benchmark Performance Guide

other mpt-30b 2023-06-22 Apache 2.0

In-depth analysis of MPT 30B performance on MMLU, HumanEval, GSM8K, MATH, BBH benchmarks, including pricing and use cases.

Key Specifications

Vendorother
Versionmpt-30b
Release Date2023-06-22
Context Window8192 tokens
Input Modalitiestext
Output Modalitiestext
LicenseApache 2.0
Documentationhttps://huggingface.co/models

Benchmark Performance

BenchmarkScoreUnitEvaluated AtNotesSource
MMLU55.2%2023-06-225-shotview
HUMANEVAL49.7pass@12023-06-22—view
GSM8K59.6%2023-06-220-shot CoTview
MATH21.2%2023-06-220-shot CoTview
BBH49.2%2023-06-223-shot CoTview
GPQA24.2%2023-06-220-shotview
IFEVAL51.2%2023-06-22prompt_strictview
ARC80.7%2023-06-22challengeview
MUSR28.9%2023-06-220-shotview
WINOGRANDE77.9%2023-06-220-shotview

Pricing

TierPriceCurrency
Input$0.4 / MtokUSD
Output$0.4 / MtokUSD
Cache Read$0 / MtokUSD
Cache Write$0 / MtokUSD

Source: https://huggingface.co/models · as of 2023-06-22

Compliance

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

MPT 30B

모델 개요

MosaicML MPT 30B 开源模型, 8K 上下文, 300 亿参数, 改进长上下文处理, Apache 2.0 可商用。

핵심 사양

공급업체버전출시일컨텍스트 창입력 모달리티출력 모달리티라이선스
Othermpt-30b2023-06-228KtexttextApache 2.0

벤치마크 성능

벤치마크점수단위비고
MMLU (Massive Multitask Language Understanding)55.2%5-shot
HumanEval49.7pass@1—
GSM8K (Grade School Math 8K)59.6%0-shot CoT
MATH21.2%0-shot CoT
BBH (BIG-Bench Hard)49.2%3-shot CoT
GPQA24.2%0-shot
IFEval51.2%prompt_strict
ARC80.7%challenge
MUSR28.9%0-shot
WinoGrande77.9%0-shot

가격

입력출력캐시 읽기캐시 쓰기
————

백만 토큰당

강점

  • 可靠的通用模型。

약점

  • MMLU 仅 55.2，知识推理偏弱。
  • HumanEval 49.7，代码能力较弱。
  • 闭源专有模型，不支持自托管。
  • 上下文窗口 8K 偏小。

사용 사례

  • 通用对话与问答

참고문헌