MicrosoftMicrosoft

Phi 3.5 Mini 4B

Legacy
705.5KDownloads984LikesAug 2024Veröffentlicht128K TokenKontextMITLizenz39 BasisQualität

Phi 3.5 Mini 4B (4B parameters) requires approximately 10.1 GB of VRAM with Q4_K_M quantization. For the best balance of quality and speed, we recommend hardware with at least 12 GB of VRAM.

Loslegen

— kopieren & einfügen, um lokal auszuführen

Copy-paste commands to run Phi 3.5 Mini 4B on your machine.

Run

ollama run phi3.5

Quick specs

Parameters4B
Architecturedense
Context128K tokens
Modalitytext
Min RAM1.6 GB
Rec. RAM2.4 GB (Q4_K_M)
LicenseMIT
FamilyPhi
Chat

About this model

Phi-3.5-mini is a lightweight, state-of-the-art open model built upon datasets used for Phi-3 - synthetic data and filtered publicly available websites - with a focus on very high-quality, reasoning dense data. The model belongs to the Phi-3 model family and supports 128K token context length. The model underwent a rigorous enhancement process, incorporating both supervised fine-tuning, proximal policy optimization, and direct preference optimization to ensure precise instruction adherence and robust safety measures.

  • Memory/compute constrained environments
  • Latency bound scenarios
  • Strong reasoning (especially code, math and logic)

Verwandte Modelle

Deine Hardware

Erkennung...

Schnellauswahl

Beste Hardware

Top-Empfehlungen für Phi 3.5 Mini 4B

Dieses Modell ausführen

Quantisierungsoptionen

VRAM-Schätzungen nach Quantisierungsstufe

No hardware detected — fit column shows raw VRAM estimates

QuantBitsVRAMQualityFit
Q2_K
2
1.6 GB
Low
Q3_K_S
3
2.0 GB
Low
NVFP4
4
2.2 GB
Medium
Q4_K_M
4
2.4 GB
Medium
Q5_K_M
5
2.9 GB
High
Q6_K
6
3.3 GB
High
Q8_0
8
4.3 GB
Very High
F16
16
8.2 GB
Maximum

Quality benchmarks

Phi 3.5 Mini 4B benchmark scores

Benchmark verified

Coding

SWE-bench Verified
HumanEval+62.8%
Aider Polyglot
LiveCodeBench

Reasoning

MMLU-Pro69.0%
GPQA Diamond12.0%
MATH-50019.6%
ARC Challenge84.6%

General

Chatbot Arena
IFEval57.7%

Source: official · 2024-08-20

Hardware-Kompatibilität

Eignungsschätzungen für alle Hardware

Rechner öffnen

Computing compatibility...

Speicheraufschlüsselung

Reference: RTX 2060 6GB

Weights2.4 GB
KV Cache5.9 GB
Runtime1.2 GB
Headroom0.6 GB

Häufig gestellte Fragen

FAQ — Phi 3.5 Mini 4B

Siehe auch