Eagle2-2B

Home/Models/nvidia/Eagle2-2B

nvidia/Eagle2-2B

kind: merge
quant: merge
family: misc
modality: image-text-to-text
root model: google/paligemma-3b-mix-448
provider: nvidia
license: cc-by-nc-4.0
language: multilingual
context: —
library: transformers
vRAM (GB): —
size on disk: —
downloads: 10,455
likes: 34
created: —
last modified: 2025-04-27T14:06:12+00:00 (2025)
huggingface: https://huggingface.co/nvidia/Eagle2-2B
tags: transformerssafetensorseagle_2_5_vlfeature-extractioneagleVLMimage-text-to-textconversationalcustom_codemultilingualarxiv:2501.14818base_model:Qwen/Qwen2.5-1.5B-Instructbase_model:merge:Qwen/Qwen2.5-1.5B-Instructbase_model:google/paligemma-3b-mix-448base_model:merge:google/paligemma-3b-mix-448base_model:google/siglip-so400m-patch14-384base_model:merge:google/siglip-so400m-patch14-384license:cc-by-nc-4.0region:us

benchmark	category	score	context	conf	evidence
DocVQA	multimodal	86.9	DocVQA<sub>test</sub>	medium	`\| DocVQA<sub>test</sub> \| 86.9 \| 88.7 \| 89.2 \| 90.1 \| 88.0 \|`
ChartQA	multimodal	76.2	ChartQA<sub>test</sub>	medium	`\| ChartQA<sub>test</sub> \| 76.2 \| 79.2 \| 81.5 \| 73.0 \| 82.0 \|`

none scraped.