SynLogic-Mix-3-32B

Home/Models/MiniMaxAI/SynLogic-Mix-3-32B

MiniMaxAI/SynLogic-Mix-3-32B

kind: original
quant: bf16 (16-bit)
family: minimax
modality: text-generation
root model: Qwen/Qwen2.5-32B
provider: MiniMaxAI
license: mit
language: en
context: —
library: transformers
vRAM (GB): —
size on disk: —
downloads: 110
likes: 20
created: —
last modified: 2025-06-10T02:53:28+00:00 (2025)
huggingface: https://huggingface.co/MiniMaxAI/SynLogic-Mix-3-32B
tags: transformerssafetensorsqwen2text-generationLLMconversationalendataset:MiniMaxAI/SynLogicarxiv:2505.19641base_model:Qwen/Qwen2.5-32Bbase_model:finetune:Qwen/Qwen2.5-32Blicense:mittext-generation-inferenceendpoints_compatibleregion:us

benchmark	category	score	context	conf	evidence
AIME 2024	math	72.6		low	`\| DeepSeek-R1-Distill-Qwen-32B \| 19.2 \| 66.6 \| 57.2 \| 72.6 \| 63.1 \|`
GPQA Diamond	reasoning	63.1		low	`\| DeepSeek-R1-Distill-Qwen-32B \| 19.2 \| 66.6 \| 57.2 \| 72.6 \| 63.1 \|`
LiveCodeBench	coding	57.2		low	`\| DeepSeek-R1-Distill-Qwen-32B \| 19.2 \| 66.6 \| 57.2 \| 72.6 \| 63.1 \|`
MATH	math	28.6	Zero-Mix-3 (SynLogic+Math+Coding)	medium	`\| Zero-Mix-3 (SynLogic+Math+Coding) \| 28.6 \| 65.0 \| 40.7 \| 35.8 \| 57.5 \|`
MATH	math	18.5	Zero-Mix-2 (Math+Coding)	medium	`\| Zero-Mix-2 (Math+Coding) \| 18.5 \| 58.6 \| 39.5 \| 34.5 \| 55.2 \|`

none scraped.