Eagle2.5-8B

Home/Models/nvidia/Eagle2.5-8B

nvidia/Eagle2.5-8B

kind: merge
quant: merge
family: misc
modality: image-text-to-text
root model: Qwen/Qwen2.5-7B-Instruct
provider: nvidia
license: other
language: multilingual
context: 128K
library: transformers
vRAM (GB): —
size on disk: —
downloads: 129,409
likes: 47
created: —
last modified: 2025-11-29T04:34:57+00:00 (2025)
huggingface: https://huggingface.co/nvidia/Eagle2.5-8B
tags: transformerssafetensorseagle_2_5_vlfeature-extractioneagleVLMimage-text-to-textconversationalcustom_codemultilingualarxiv:2504.15271base_model:Qwen/Qwen2.5-7B-Instructbase_model:merge:Qwen/Qwen2.5-7B-Instructbase_model:google/siglip2-so400m-patch16-512base_model:merge:google/siglip2-so400m-patch16-512license:otherregion:us

benchmark	category	score	context	conf	evidence
DocVQA	multimodal	92.8	DocVQA<sub>test</sub>	medium	`\| DocVQA<sub>test</sub> \| 92.8 \| 93.1 \| 93.0 \| 95.7 \| 94.1 \|`
ChartQA	multimodal	85.7	ChartQA<sub>test</sub>	medium	`\| ChartQA<sub>test</sub> \| 85.7 \| 87.2 \| 84.8 \| 87.3 \| 87.5 \|`
MMBench	multimodal	83.1	MMBench_V11<sub>test</sub>	medium	`\| MMBench_V11<sub>test</sub> \| 83.1 \| 74.6 \| 83.2 \| 82.6 \| 81.7 \|`
MMMU	multimodal	69.1	MMMU<sub>val</sub>	medium	`\| MMMU<sub>val</sub> \| 69.1 \| 62.2 \| 56.0 \| 58.6 \| 55.8 \|`
MathVista	multimodal	63.8	MathVista<sub>testmini</sub>	medium	`\| MathVista<sub>testmini</sub> \| 63.8 \| 63.9 \| 64.4 \| 68.2 \| 67.8 \|`

none scraped.