Key Specifications
| Vendor | alibaba |
|---|
| Version | 2.5-0-5b |
|---|
| Release Date | 2024-09-19 |
|---|
| Context Window | 32768 tokens |
|---|
| Input Modalities | text |
|---|
| Output Modalities | text |
|---|
| License | Qwen License |
|---|
| Documentation | https://qwenlm.github.io/ |
|---|
Benchmark Performance
| Benchmark | Score | Unit | Evaluated At | Notes | Source |
|---|
| MMLU | 41.5 | % | 2024-09-19 | 5-shot | view |
| HUMANEVAL | 31 | pass@1 | 2024-09-19 | — | view |
| GSM8K | 36.2 | % | 2024-09-19 | 0-shot CoT | view |
| MATH | 15.4 | % | 2024-09-19 | 0-shot CoT | view |
| BBH | 51 | % | 2024-09-19 | 3-shot CoT | view |
| GPQA | 23.2 | % | 2024-09-19 | 0-shot | view |
| IFEVAL | 48.5 | % | 2024-09-19 | prompt_strict | view |
| ARC | 85.4 | % | 2024-09-19 | challenge | view |
| MUSR | 28.9 | % | 2024-09-19 | 0-shot | view |
| WINOGRANDE | 64.3 | % | 2024-09-19 | 0-shot | view |
Compliance
- Data Residency: CN
- SOC2: ✗
- HIPAA: ✗
- GDPR: ✗
- ISO 27001: ✗
Qwen2.5 0.5B
Visão geral do modelo
阿里巴巴 Qwen2.5 0.5B 超小型开源模型, 32K 上下文, 0.5B 参数, 适合极低资源环境与移动端。
Especificações principais
| Fornecedor | Versão | Data de lançamento | Janela de contexto | Modalidades de entrada | Modalidades de saída | Licença |
|---|
| Alibaba | 2.5-0-5b | 2024-09-19 | 32K | text | text | Qwen License |
Desempenho em benchmarks
| Benchmark | Pontuação | Unidade | Notas |
|---|
| MMLU (Massive Multitask Language Understanding) | 41.5 | % | 5-shot |
| HumanEval | 31.0 | pass@1 | — |
| GSM8K (Grade School Math 8K) | 36.2 | % | 0-shot CoT |
| MATH | 15.4 | % | 0-shot CoT |
| BBH (BIG-Bench Hard) | 51.0 | % | 3-shot CoT |
| GPQA | 23.2 | % | 0-shot |
| IFEval | 48.5 | % | prompt_strict |
| ARC | 85.4 | % | challenge |
| MUSR | 28.9 | % | 0-shot |
| WinoGrande | 64.3 | % | 0-shot |
Preços
| Entrada | Saída | Leitura de cache | Escrita de cache |
|---|
| — | — | — | — |
por milhão de tokens
Pontos fortes
Pontos fracos
- MMLU 仅 41.5，知识推理偏弱。
- HumanEval 31.0，代码能力较弱。
- 闭源专有模型，不支持自托管。
Casos de uso
Referências