Phi-3.5-mini-instruct
MicrosoftPhi开源权重MIT · 商用许可
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
60
agents
0
multimodal
缺少专门科学评测时，Science 由 LLM Stats 科学得分或推理能力估算。
排行榜排名
暂无排名数据
基准测试分数 (LLM Stats)
(LLM Stats (zeroeval))General
MMLU
69.0%自报
TruthfulQA
64.0%自报
Arena Hard
37.0%自报
Language
BoolQ
78.0%自报
MEGA TyDi QA
62.2%自报
MEGA MLQA
61.7%自报
MMMLU
55.4%自报
MMLU-Pro
47.4%自报
MEGA UDPOS
46.5%自报
Long Context
RULER
84.1%自报
RepoQA
77.0%自报
Math
GSM8k
86.2%自报
MATH
48.5%自报
MGSM
47.9%自报
Reasoning
ARC-C
84.6%自报
PIQA
81.0%自报
OpenBookQA
79.2%自报
Social IQa
74.7%自报
MEGA XStoryCloze
73.5%自报
MBPP
0.70 / 100自报
HellaSwagAI2 (2019)
69.4%自报
BIG-Bench Hard
69.0%自报
Winogrande
68.5%自报
MEGA XCOPA
63.1%自报
HumanEvalOpenAI (2021)
62.8%自报
Qasper
41.9%自报
GPQANYU + Cohere + Anthropic (2023)
30.4%自报
Summarization
GovReport
25.9%自报
SQuALITY
24.3%自报
QMSum
21.3%自报
SummScreenFD
16.0%自报
AA 评测指数
(Artificial Analysis)暂无 AA 评测数据
LLM Stats 分类评分
(LLM Stats (zeroeval))Psychology70
Reasoning70
Language60
Legal60
Math60
Physics60
Finance60
General60
Healthcare60
Code60
Creativity60
Long Context50
Chat40
Writing40
Biology30
Chemistry30
Summarization20
定价
暂无定价数据
速度
暂无速度数据
供应商价格排行
暂无提供商数据