Phi-3.5-mini-instruct
MicrosoftPhi開源權重MIT · 商用許可
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.
發布日期
2024-08-23
參數規模
3.8B
上下文長度
—
支援模態
text
能力雷達圖
60
general
60
coding
60
reasoning
26
science估算
60
agents
0
multimodal
缺少專門科學評測時，Science 由 LLM Stats 科學得分或推理能力估算。
排行榜排名
暫無排名資料
基準測試分數 (LLM Stats)
(LLM Stats (zeroeval))General
MMLU
69.0%自報
TruthfulQA
64.0%自報
Arena Hard
37.0%自報
Language
BoolQ
78.0%自報
MEGA TyDi QA
62.2%自報
MEGA MLQA
61.7%自報
MMMLU
55.4%自報
MMLU-Pro
47.4%自報
MEGA UDPOS
46.5%自報
Long Context
RULER
84.1%自報
RepoQA
77.0%自報
Math
GSM8k
86.2%自報
MATH
48.5%自報
MGSM
47.9%自報
Reasoning
ARC-C
84.6%自報
PIQA
81.0%自報
OpenBookQA
79.2%自報
Social IQa
74.7%自報
MEGA XStoryCloze
73.5%自報
MBPP
0.70 / 100自報
HellaSwagAI2 (2019)
69.4%自報
BIG-Bench Hard
69.0%自報
Winogrande
68.5%自報
MEGA XCOPA
63.1%自報
HumanEvalOpenAI (2021)
62.8%自報
Qasper
41.9%自報
GPQANYU + Cohere + Anthropic (2023)
30.4%自報
Summarization
GovReport
25.9%自報
SQuALITY
24.3%自報
QMSum
21.3%自報
SummScreenFD
16.0%自報
AA 評測指數
(Artificial Analysis)暫無 AA 評測資料
LLM Stats 分類評分
(LLM Stats (zeroeval))Psychology70
Reasoning70
Language60
Legal60
Math60
Physics60
Finance60
General60
Healthcare60
Code60
Creativity60
Long Context50
Chat40
Writing40
Biology30
Chemistry30
Summarization20
定價
暫無定價資料
速度
暫無速度資料
供應商價格排行
暫無提供商資料