Phi-4-mini-flash-reasoning

Home/Models/microsoft/Phi-4-mini-flash-reasoning

microsoft/Phi-4-mini-flash-reasoning

kind: original
quant: bf16 (16-bit)
family: phi
modality: text-generation
root model: microsoft/Phi-4-mini-flash-reasoning
provider: microsoft
license: mit
language: en
context: 64K
library: transformers
vRAM (GB): —
size on disk: —
downloads: 841
likes: 281
created: —
last modified: 2025-12-10T20:24:55+00:00 (2025)
huggingface: https://huggingface.co/microsoft/Phi-4-mini-flash-reasoning
tags: transformerssafetensorsphi4flashtext-generationnlpmathcodeconversationalcustom_codeenarxiv:2507.06607license:mitregion:us

benchmark	category	score	conf	evidence
MATH-500	math	84.5	low	`\| DeepSeek-R1-Distill-Qwen-1.5B \| 29.58 \| 20.78 \| 84.50 \| 37.69 \|`
GPQA Diamond	reasoning	37.69	low	`\| DeepSeek-R1-Distill-Qwen-1.5B \| 29.58 \| 20.78 \| 84.50 \| 37.69 \|`
AIME 2024	math	29.58	low	`\| DeepSeek-R1-Distill-Qwen-1.5B \| 29.58 \| 20.78 \| 84.50 \| 37.69 \|`
AIME 2025	math	20.78	low	`\| DeepSeek-R1-Distill-Qwen-1.5B \| 29.58 \| 20.78 \| 84.50 \| 37.69 \|`

none scraped.