Update README.md
Browse files
README.md
CHANGED
@@ -299,4 +299,30 @@ configs:
|
|
299 |
path: preview/WikiVQA_test-*
|
300 |
- split: WikiVQA_adv
|
301 |
path: preview/WikiVQA_adv-*
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
302 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
299 |
path: preview/WikiVQA_test-*
|
300 |
- split: WikiVQA_adv
|
301 |
path: preview/WikiVQA_adv-*
|
302 |
+
task_categories:
|
303 |
+
- visual-question-answering
|
304 |
+
- question-answering
|
305 |
+
- text-generation
|
306 |
+
- image-to-text
|
307 |
+
- video-classification
|
308 |
+
language:
|
309 |
+
- en
|
310 |
+
tags:
|
311 |
+
- Long-context
|
312 |
+
- MLLM
|
313 |
+
- VLM
|
314 |
+
- LLM
|
315 |
+
pretty_name: MileBench
|
316 |
+
size_categories:
|
317 |
+
- 1K<n<10K
|
318 |
---
|
319 |
+
|
320 |
+
|
321 |
+
# MileBench
|
322 |
+
|
323 |
+
We introduce MileBench, a pioneering benchmark designed to test the **M**ult**I**modal **L**ong-cont**E**xt capabilities of MLLMs.
|
324 |
+
This benchmark comprises not only multimodal long contexts, but also multiple tasks requiring both comprehension and generation.
|
325 |
+
We establish two distinct evaluation sets, diagnostic and realistic, to systematically assess MLLMs’ long-context adaptation capacity and their ability to completetasks in long-context scenarios
|
326 |
+
|
327 |
+
To construct our evaluation sets, we gather 6,440 multimodal long-context samples from 21 pre-existing or self-constructed datasets,
|
328 |
+
with an average of 15.2 images and 422.3 words each, as depicted in the figure, and we categorize them into their respective subsets.
|