Code Bison: Complete Benchmark Performance Guide

google code-bison 2023-05-10 Proprietary

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

Key Specifications

Vendorgoogle
Versioncode-bison
Release Date2023-05-10
Context Window8192 tokens
Input Modalitiestext
Output Modalitiestext
LicenseProprietary
Documentationhttps://ai.google.dev/gemini-api/docs

Benchmark Performance

BenchmarkScoreUnitEvaluated AtNotesSource
MMLU65%2023-05-105-shotview
HUMANEVAL76.2pass@12023-05-10—view
GSM8K55%2023-05-100-shot CoTview
MATH34.4%2023-05-100-shot CoTview
BBH71.4%2023-05-103-shot CoTview
GPQA29.1%2023-05-100-shotview
IFEVAL59.2%2023-05-10prompt_strictview
ARC86.1%2023-05-10challengeview
MUSR49.8%2023-05-100-shotview
WINOGRANDE77.9%2023-05-100-shotview

Pricing

TierPriceCurrency
Input$0.5 / MtokUSD
Output$1 / MtokUSD
Cache Read$0 / MtokUSD
Cache Write$0 / MtokUSD

Source: https://ai.google.dev/pricing · as of 2023-05-10

Compliance

  • Data Residency: US
  • SOC2: ✓
  • HIPAA: ✗
  • GDPR: ✓
  • ISO 27001: ✓

Code Bison

Visão geral do modelo

Google Vertex AI Code Bison 代码生成模型, 基于 PaLM 2, 8K 上下文, 支持多语言代码生成与补全。

Especificações principais

FornecedorVersãoData de lançamentoJanela de contextoModalidades de entradaModalidades de saídaLicença
Googlecode-bison2023-05-108KtexttextProprietary

Desempenho em benchmarks

BenchmarkPontuaçãoUnidadeNotas
MMLU (Massive Multitask Language Understanding)65.0%5-shot
HumanEval76.2pass@1—
GSM8K (Grade School Math 8K)55.0%0-shot CoT
MATH34.4%0-shot CoT
BBH (BIG-Bench Hard)71.4%3-shot CoT
GPQA29.1%0-shot
IFEval59.2%prompt_strict
ARC86.1%challenge
MUSR49.8%0-shot
WinoGrande77.9%0-shot

Preços

EntradaSaídaLeitura de cacheEscrita de cache
————

por milhão de tokens

Pontos fortes

  • 可靠的通用模型。

Pontos fracos

  • 闭源专有模型，不支持自托管。
  • 上下文窗口 8K 偏小。

Casos de uso

  • 代码生成与调试

Referências