Chat Bison: Complete Benchmark Performance Guide

google chat-bison 2023-05-10 Proprietary

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

Key Specifications

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

Benchmark Performance

BenchmarkScoreUnitEvaluated AtNotesSource
MMLU63.1%2023-05-105-shotview
HUMANEVAL36pass@12023-05-10—view
GSM8K52.4%2023-05-100-shot CoTview
MATH19.8%2023-05-100-shot CoTview
BBH54.6%2023-05-103-shot CoTview
GPQA25%2023-05-100-shotview
IFEVAL48.2%2023-05-10prompt_strictview
ARC82.6%2023-05-10challengeview
MUSR32.8%2023-05-100-shotview
WINOGRANDE66.2%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: ✓

Chat Bison

Modellöversikt

Google Vertex AI Chat Bison 对话模型, 基于 PaLM 2, 8K 上下文, 适合企业级对话应用。

Kärnspecifikationer

LeverantörVersionReleasedatumKontextfönsterInmatningsmodaliteterUtmatningsmodaliteterLicens
Googlechat-bison2023-05-108KtexttextProprietary

Benchmarkprestanda

BenchmarkPoängEnhetAnteckningar
MMLU (Massive Multitask Language Understanding)63.1%5-shot
HumanEval36.0pass@1—
GSM8K (Grade School Math 8K)52.4%0-shot CoT
MATH19.8%0-shot CoT
BBH (BIG-Bench Hard)54.6%3-shot CoT
GPQA25.0%0-shot
IFEval48.2%prompt_strict
ARC82.6%challenge
MUSR32.8%0-shot
WinoGrande66.2%0-shot

Priser

InmatningUtmatningCacheläsningCacheskrivning
————

per miljon tokens

Styrkor

  • 可靠的通用模型。

Svagheter

  • HumanEval 36.0，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。

Användningsområden

  • 通用对话与问答

Referenser