Key Specifications
| Vendor | other |
|---|
| Version | nous-capybara-34b |
|---|
| Release Date | 2023-11-15 |
|---|
| Context Window | 4096 tokens |
|---|
| Input Modalities | text |
|---|
| Output Modalities | text |
|---|
| License | Apache 2.0 |
|---|
| Documentation | https://huggingface.co/models |
|---|
Benchmark Performance
| Benchmark | Score | Unit | Evaluated At | Notes | Source |
|---|
| MMLU | 53.2 | % | 2023-11-15 | 5-shot | view |
| HUMANEVAL | 50.1 | pass@1 | 2023-11-15 | — | view |
| GSM8K | 31.9 | % | 2023-11-15 | 0-shot CoT | view |
| MATH | 23.1 | % | 2023-11-15 | 0-shot CoT | view |
| BBH | 56.1 | % | 2023-11-15 | 3-shot CoT | view |
| GPQA | 29.7 | % | 2023-11-15 | 0-shot | view |
| IFEVAL | 52.9 | % | 2023-11-15 | prompt_strict | view |
| ARC | 80.7 | % | 2023-11-15 | challenge | view |
| MUSR | 31.7 | % | 2023-11-15 | 0-shot | view |
| WINOGRANDE | 67.3 | % | 2023-11-15 | 0-shot | view |
Pricing
| Tier | Price | Currency |
|---|
| Input | $0.4 / Mtok | USD |
| Output | $0.4 / Mtok | USD |
| Cache Read | $0 / Mtok | USD |
| Cache Write | $0 / Mtok | USD |
Source:
https://huggingface.co/models
· as of 2023-11-15
Compliance
- Data Residency: self-host
- SOC2: ✗
- HIPAA: ✗
- GDPR: ✗
- ISO 27001: ✗
Nous Capybara 34B
Descripción del modelo
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。
Especificaciones principales
| Proveedor | Versión | Fecha de lanzamiento | Ventana de contexto | Modalidades de entrada | Modalidades de salida | Licencia |
|---|
| Other | nous-capybara-34b | 2023-11-15 | 4K | text | text | Apache 2.0 |
Rendimiento en benchmarks
| Benchmark | Puntuación | Unidad | Notas |
|---|
| MMLU (Massive Multitask Language Understanding) | 53.2 | % | 5-shot |
| HumanEval | 50.1 | pass@1 | — |
| GSM8K (Grade School Math 8K) | 31.9 | % | 0-shot CoT |
| MATH | 23.1 | % | 0-shot CoT |
| BBH (BIG-Bench Hard) | 56.1 | % | 3-shot CoT |
| GPQA | 29.7 | % | 0-shot |
| IFEval | 52.9 | % | prompt_strict |
| ARC | 80.7 | % | challenge |
| MUSR | 31.7 | % | 0-shot |
| WinoGrande | 67.3 | % | 0-shot |
Precios
| Entrada | Salida | Lectura caché | Escritura caché |
|---|
| — | — | — | — |
por millón de tokens
Fortalezas
Debilidades
- 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。
Casos de uso
Referencias