Phi-3.5-mini-instruct
MicrosoftPhiOpen WeightMIT · Commercial OK
描述
Phi-3.5-mini-instruct is a 3.8B-parameter model that supports up to 128K context tokens, with improved multilingual capabilities across over 20 languages. It underwent additional training and safety post-training to enhance instruction-following, reasoning, math, and code generation. Ideal for environments with memory or latency constraints, it uses an MIT license.
发布日期
2024-08-23
参数规模
3.8B
上下文长度
—
支持模态
text
能力雷达图
60
general
60
coding
60
reasoning
26
science估算
0
agents
0
multimodal
Science 在缺少专门科学评测时使用推理能力代理估算。
排行榜排名
| 领域 | #排名 | 分数 | 来源 |
|---|---|---|---|
| 推理能力 | 51 | 69.0 | LS |
基准测试分数 (LLM Stats)
Biology
GPQA
30.4%自报
Code
RepoQA
77.0%自报
HumanEval
62.8%自报
Creativity
Social IQa
74.7%自报
Arena Hard
37.0%自报
Finance
MMLU
69.0%自报
TruthfulQA
64.0%自报
MMLU-Pro
47.4%自报
General
ARC-C
84.6%自报
PIQA
81.0%自报
OpenBookQA
79.2%自报
MBPP
0.70 / 100自报
MMMLU
55.4%自报
Language
BoolQ
78.0%自报
MEGA XStoryCloze
73.5%自报
BIG-Bench Hard
69.0%自报
Winogrande
68.5%自报
MEGA XCOPA
63.1%自报
MEGA TyDi QA
62.2%自报
MEGA MLQA
61.7%自报
MEGA UDPOS
46.5%自报
SQuALITY
24.3%自报
Long Context
RULER
84.1%自报
Qasper
41.9%自报
GovReport
25.9%自报
QMSum
21.3%自报
SummScreenFD
16.0%自报
Math
GSM8k
86.2%自报
MATH
48.5%自报
MGSM
47.9%自报
Reasoning
HellaSwag
69.4%自报
AA 评测指数
暂无 AA 评测数据
LLM Stats 分类评分
Psychology70
Reasoning70
Code60
Creativity60
Finance60
General60
Healthcare60
Language60
Legal60
Math60
Physics60
Long Context50
Writing40
Biology30
Chemistry30
Summarization20
定价
暂无定价数据
速度
暂无速度数据
可用提供商
(LS 内部计价单位)暂无提供商数据