Phi-3.5-mini-instruct
MicrosoftPhiOpen WeightMIT · Commercial OK
Description
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.
Release Date
2024-08-23
Parameters
3.8B
Context Length
—
Modalities
text
Capability Radar
60
general
60
coding
60
reasoning
26
scienceest.
0
agents
0
multimodal
Science uses a reasoning proxy when dedicated science benchmarks are unavailable.
Rankings
| Domain | #Rank | Score | Source |
|---|---|---|---|
| Reasoning | 51 | 69.0 | LS |
Benchmark Scores (LLM Stats)
Biology
GPQA
30.4%SR
Code
RepoQA
77.0%SR
HumanEval
62.8%SR
Creativity
Social IQa
74.7%SR
Arena Hard
37.0%SR
Finance
MMLU
69.0%SR
TruthfulQA
64.0%SR
MMLU-Pro
47.4%SR
General
ARC-C
84.6%SR
PIQA
81.0%SR
OpenBookQA
79.2%SR
MBPP
0.70 / 100SR
MMMLU
55.4%SR
Language
BoolQ
78.0%SR
MEGA XStoryCloze
73.5%SR
BIG-Bench Hard
69.0%SR
Winogrande
68.5%SR
MEGA XCOPA
63.1%SR
MEGA TyDi QA
62.2%SR
MEGA MLQA
61.7%SR
MEGA UDPOS
46.5%SR
SQuALITY
24.3%SR
Long Context
RULER
84.1%SR
Qasper
41.9%SR
GovReport
25.9%SR
QMSum
21.3%SR
SummScreenFD
16.0%SR
Math
GSM8k
86.2%SR
MATH
48.5%SR
MGSM
47.9%SR
Reasoning
HellaSwag
69.4%SR
AA Evaluation Indices
No AA evaluation data available
LLM Stats Category Scores
Psychology70
Reasoning70
Code60
Creativity60
Finance60
General60
Healthcare60
Language60
Legal60
Math60
Physics60
Long Context50
Writing40
Biology30
Chemistry30
Summarization20
Pricing
No pricing data available
Speed
No speed data available
Available Providers
(LS internal units)No provider data available