Microsoft WizardMath 7B v1: Complete Benchmark Performance Guide

other wizardmath-7b-v1 2023-08-17 Llama 2 Community License

In-depth analysis of Microsoft WizardMath 7B v1 performance on MMLU, HumanEval, GSM8K, MATH, BBH benchmarks, including pricing and use cases.

Key Specifications

Vendorother
Versionwizardmath-7b-v1
Release Date2023-08-17
Context Window4096 tokens
Input Modalitiestext
Output Modalitiestext
LicenseLlama 2 Community License
Documentationhttps://huggingface.co/models

Benchmark Performance

Benchmark Score Unit Evaluated At Notes Source
MMLU 69.1 % 2023-08-17 5-shot view
HUMANEVAL 61.1 pass@1 2023-08-17 — view
GSM8K 83.7 % 2023-08-17 0-shot CoT view
MATH 65.8 % 2023-08-17 0-shot CoT view
BBH 74.4 % 2023-08-17 3-shot CoT view
GPQA 30.2 % 2023-08-17 0-shot view
IFEVAL 57.7 % 2023-08-17 prompt_strict view
ARC 91.4 % 2023-08-17 challenge view
MUSR 53.4 % 2023-08-17 0-shot view
WINOGRANDE 70.5 % 2023-08-17 0-shot view

Pricing

Tier Price Currency
Input$0.18 / MtokUSD
Output$0.18 / MtokUSD
Cache Read$0 / MtokUSD
Cache Write$0 / MtokUSD

Source: https://huggingface.co/models · as of 2023-08-17

Compliance

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

Microsoft WizardMath 7B v1

Model Overview

Microsoft WizardMath 7B v1 数学专用模型, 4K 上下文, 基于 Llama 2 7B, 在 GSM8K 上达到 54.9%。

Core Specifications

Vendor Version Release Date Context Window Input Modalities Output Modalities License
Other wizardmath-7b-v1 2023-08-17 4K text text Llama 2 Community License

Benchmark Performance

Benchmark Score Unit Notes
MMLU (Massive Multitask Language Understanding) 69.1 % 5-shot
HumanEval 61.1 pass@1 —
GSM8K (Grade School Math 8K) 83.7 % 0-shot CoT
MATH 65.8 % 0-shot CoT
BBH (BIG-Bench Hard) 74.4 % 3-shot CoT
GPQA 30.2 % 0-shot
IFEval 57.7 % prompt_strict
ARC 91.4 % challenge
MUSR 53.4 % 0-shot
WinoGrande 70.5 % 0-shot

Pricing

Input Output Cache Read Cache Write
— — — —

per million tokens

Strengths

  • Reliable general-purpose model.

Weaknesses

  • Proprietary, not self-hostable.
  • Context window 4K is limited.

Use Cases

  • General chat and Q&A

References