Nous Capybara 34B: Complete Benchmark Performance Guide

other nous-capybara-34b 2023-11-15 Apache 2.0

In-depth analysis of Nous Capybara 34B performance on MMLU, HumanEval, GSM8K, MATH, BBH benchmarks, including pricing and use cases.

Key Specifications

Vendorother
Versionnous-capybara-34b
Release Date2023-11-15
Context Window4096 tokens
Input Modalitiestext
Output Modalitiestext
LicenseApache 2.0
Documentationhttps://huggingface.co/models

Benchmark Performance

BenchmarkScoreUnitEvaluated AtNotesSource
MMLU53.2%2023-11-155-shotview
HUMANEVAL50.1pass@12023-11-15—view
GSM8K31.9%2023-11-150-shot CoTview
MATH23.1%2023-11-150-shot CoTview
BBH56.1%2023-11-153-shot CoTview
GPQA29.7%2023-11-150-shotview
IFEVAL52.9%2023-11-15prompt_strictview
ARC80.7%2023-11-15challengeview
MUSR31.7%2023-11-150-shotview
WINOGRANDE67.3%2023-11-150-shotview

Pricing

TierPriceCurrency
Input$0.4 / MtokUSD
Output$0.4 / MtokUSD
Cache Read$0 / MtokUSD
Cache Write$0 / MtokUSD

Source: https://huggingface.co/models · as of 2023-11-15

Compliance

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

Nous Capybara 34B

Modellöversikt

NousResearch Capybara 34B 对话模型, 4K 上下文, 基于 Yi 34B 微调, 改进长对话与推理能力。

Kärnspecifikationer

LeverantörVersionReleasedatumKontextfönsterInmatningsmodaliteterUtmatningsmodaliteterLicens
Othernous-capybara-34b2023-11-154KtexttextApache 2.0

Benchmarkprestanda

BenchmarkPoängEnhetAnteckningar
MMLU (Massive Multitask Language Understanding)53.2%5-shot
HumanEval50.1pass@1—
GSM8K (Grade School Math 8K)31.9%0-shot CoT
MATH23.1%0-shot CoT
BBH (BIG-Bench Hard)56.1%3-shot CoT
GPQA29.7%0-shot
IFEval52.9%prompt_strict
ARC80.7%challenge
MUSR31.7%0-shot
WinoGrande67.3%0-shot

Priser

InmatningUtmatningCacheläsningCacheskrivning
————

per miljon tokens

Styrkor

  • 可靠的通用模型。

Svagheter

  • MMLU 仅 53.2，知识推理偏弱。
  • 闭源专有模型，不支持自托管。
  • 上下文窗口 4K 偏小。

Användningsområden

  • 通用对话与问答

Referenser