Ring-lite

Home/Models/inclusionAI/Ring-lite

inclusionAI/Ring-lite

kind: original
quant: bf16 (16-bit)
family: misc
modality: text-generation
root model: inclusionAI/Ling-lite-base-1.5
provider: inclusionAI
license: mit
language: zh, en
context: —
library: transformers
vRAM (GB): —
size on disk: —
downloads: 49
likes: 49
created: —
last modified: 2025-08-18T02:52:11+00:00 (2025)
huggingface: https://huggingface.co/inclusionAI/Ring-lite
tags: transformerssafetensorsbailing_moetext-generationconversationalcustom_codezhenarxiv:2506.14731base_model:inclusionAI/Ling-lite-base-1.5base_model:finetune:inclusionAI/Ling-lite-base-1.5license:mitregion:us

benchmark	category	score	context	conf	evidence
Codeforces	coding	86.5	Codeforces (percentile)	medium	`\| Codeforces (percentile) \| 86.5 \| 88.0 \|`
AIME 2024	math	76.6	AIME 2024	medium	`\| AIME 2024 \| 76.6 \| 79.0 \|`
AIME 2025	math	69.1	AIME 2025	medium	`\| AIME 2025 \| 69.1 \| 69.5 \|`
GPQA Diamond	reasoning	61.1	GPQA Diamond	medium	`\| GPQA Diamond \| 61.1 \| 63.2 \|`
LiveCodeBench	coding	60.7	LiveCodeBench	medium	`\| LiveCodeBench \| 60.7 \| 61.4 \|`
MMLU-Pro	knowledge	60.0	MMLU-Pro	medium	`\| MMLU-Pro \| 60.0 \| 63.0 \|`
C-Eval	knowledge	59.0	C-Eval	medium	`\| C-Eval \| 59.0 \| 65.4 \|`
IFEval	instruction-following	51.6	IF-Eval	medium	`\| IF-Eval \| 51.6 \| 54.3 \|`
Chatbot Arena	chat	27.8	ArenaHard	medium	`\| ArenaHard \| 27.8 \| 62.7 \|`

none scraped.