Open-Platypus: Complete Benchmark Performance Guide

other open-platypus 2023-09-15 MIT

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

Key Specifications

Vendorother
Versionopen-platypus
Release Date2023-09-15
Context Window2048 tokens
Input Modalitiestext
Output Modalitiestext
LicenseMIT
Documentationhttps://huggingface.co/models

Benchmark Performance

BenchmarkScoreUnitEvaluated AtNotesSource
MMLU68.7%2023-09-155-shotview
HUMANEVAL48.2pass@12023-09-15—view
GSM8K43.9%2023-09-150-shot CoTview
MATH21%2023-09-150-shot CoTview
BBH63.7%2023-09-153-shot CoTview
GPQA34.9%2023-09-150-shotview
IFEVAL46%2023-09-15prompt_strictview
ARC80.1%2023-09-15challengeview
MUSR40%2023-09-150-shotview
WINOGRANDE66.7%2023-09-150-shotview

Pricing

TierPriceCurrency
Input$0.18 / MtokUSD
Output$0.18 / MtokUSD
Cache Read$0 / MtokUSD
Cache Write$0 / MtokUSD

Source: https://huggingface.co/models · as of 2023-09-15

Compliance

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

Open-Platypus

Visão geral do modelo

Garage bAInd Open-Platypus 数据集与模型, 2K 上下文, 基于 Llama 2 微调, 适合 STEM 推理任务。

Especificações principais

FornecedorVersãoData de lançamentoJanela de contextoModalidades de entradaModalidades de saídaLicença
Otheropen-platypus2023-09-152KtexttextMIT

Desempenho em benchmarks

BenchmarkPontuaçãoUnidadeNotas
MMLU (Massive Multitask Language Understanding)68.7%5-shot
HumanEval48.2pass@1—
GSM8K (Grade School Math 8K)43.9%0-shot CoT
MATH21.0%0-shot CoT
BBH (BIG-Bench Hard)63.7%3-shot CoT
GPQA34.9%0-shot
IFEval46.0%prompt_strict
ARC80.1%challenge
MUSR40.0%0-shot
WinoGrande66.7%0-shot

Preços

EntradaSaídaLeitura de cacheEscrita de cache
————

por milhão de tokens

Pontos fortes

  • 可靠的通用模型。

Pontos fracos

  • HumanEval 48.2，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。

Casos de uso

  • 通用对话与问答

Referências