Phi-1.5: Complete Benchmark Performance Guide

other phi-1-5 2023-09-14 MIT

In-depth analysis of Phi-1.5 performance on MMLU, HumanEval, GSM8K, MATH, BBH benchmarks, including pricing and use cases.

Key Specifications

Vendorother
Versionphi-1-5
Release Date2023-09-14
Context Window2048 tokens
Input Modalitiestext
Output Modalitiestext
LicenseMIT
Documentationhttps://huggingface.co/models

Benchmark Performance

BenchmarkScoreUnitEvaluated AtNotesSource
MMLU58.4%2023-09-145-shotview
HUMANEVAL39.2pass@12023-09-14—view
GSM8K34.2%2023-09-140-shot CoTview
MATH27.2%2023-09-140-shot CoTview
BBH49.5%2023-09-143-shot CoTview
GPQA16.9%2023-09-140-shotview
IFEVAL48%2023-09-14prompt_strictview
ARC75.9%2023-09-14challengeview
MUSR30.3%2023-09-140-shotview
WINOGRANDE74.1%2023-09-140-shotview

Pricing

TierPriceCurrency
Input$0.1 / MtokUSD
Output$0.1 / MtokUSD
Cache Read$0 / MtokUSD
Cache Write$0 / MtokUSD

Source: https://huggingface.co/models · as of 2023-09-14

Compliance

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

Phi-1.5

Panoramica del modello

Microsoft Phi-1.5 1.3B 模型, 2K 上下文, 改进推理能力, 在 1.3B 规模上接近 7B 模型水平。

Specifiche principali

FornitoreVersioneData di rilascioFinestra di contestoModalità di inputModalità di outputLicenza
Otherphi-1-52023-09-142KtexttextMIT

Prestazioni benchmark

BenchmarkPunteggioUnitàNote
MMLU (Massive Multitask Language Understanding)58.4%5-shot
HumanEval39.2pass@1—
GSM8K (Grade School Math 8K)34.2%0-shot CoT
MATH27.2%0-shot CoT
BBH (BIG-Bench Hard)49.5%3-shot CoT
GPQA16.9%0-shot
IFEval48.0%prompt_strict
ARC75.9%challenge
MUSR30.3%0-shot
WinoGrande74.1%0-shot

Prezzi

InputOutputLettura cacheScrittura cache
————

per milione di token

Punti di forza

  • 可靠的通用模型。

Punti deboli

  • MMLU 仅 58.4，知识推理偏弱。
  • HumanEval 39.2，代码能力较弱。
  • 闭源专有模型，不支持自托管。
  • 上下文窗口 2K 偏小。

Casi d’uso

  • 通用对话与问答

Riferimenti