Microsoft/Dense

Phi-3-medium-14b

chatcodingreasoning
14B
Parameters
4K
Context length
8
Benchmarks
4
Quantizations
0
Architecture
Dense
Released
2024-05-21
Layers
40
KV Heads
8
Head Dim
128
Family
phi

Quantizations & VRAM

Q4_K_M4.5 bpw
8.4 GB
VRAM required
94%
Quality
Q6_K6.5 bpw
11.9 GB
VRAM required
97%
Quality
Q8_08 bpw
14.5 GB
VRAM required
100%
Quality
FP1616 bpw
28.5 GB
VRAM required
100%
Quality

Benchmarks (8)

Arena Elo1460
IFEval64.2
BBH49.4
MMLU-PRO40.8
BigCodeBench37.6
MATH19.6
MUSR13.1
GPQA11.5

Run with Ollama

$ollama run phi3:14b

GPUs that can run this model

At Q4_K_M quantization. Sorted by minimum VRAM.

Find the best GPU for Phi-3-medium-14b

Build Hardware for Phi-3-medium-14b