Update README.md
Browse files
README.md
CHANGED
|
@@ -21,15 +21,16 @@ dataset_info:
|
|
| 21 |
dtype: string
|
| 22 |
splits:
|
| 23 |
- name: test
|
| 24 |
-
num_bytes: 568698994
|
| 25 |
num_examples: 227
|
| 26 |
download_size: 568517703
|
| 27 |
-
dataset_size: 568698994
|
| 28 |
configs:
|
| 29 |
- config_name: default
|
| 30 |
data_files:
|
| 31 |
- split: test
|
| 32 |
path: data/test-*
|
|
|
|
| 33 |
---
|
| 34 |
|
| 35 |
|
|
@@ -51,4 +52,4 @@ configs:
|
|
| 51 |
## Benchmarking LMM&VLM
|
| 52 |
Most existing VLMs struggle to select the correct image based on scientific knowledge, with performance often resembling random guessing. Similarly, LMMs face challenges in this area. However, SciScore stands out by demonstrating exceptional performance, achieving human-level accuracy after being trained on Science-T2I.
|
| 53 |
|
| 54 |
-
<img src="./exp.png" align="center">
|
|
|
|
| 21 |
dtype: string
|
| 22 |
splits:
|
| 23 |
- name: test
|
| 24 |
+
num_bytes: 568698994
|
| 25 |
num_examples: 227
|
| 26 |
download_size: 568517703
|
| 27 |
+
dataset_size: 568698994
|
| 28 |
configs:
|
| 29 |
- config_name: default
|
| 30 |
data_files:
|
| 31 |
- split: test
|
| 32 |
path: data/test-*
|
| 33 |
+
license: apache-2.0
|
| 34 |
---
|
| 35 |
|
| 36 |
|
|
|
|
| 52 |
## Benchmarking LMM&VLM
|
| 53 |
Most existing VLMs struggle to select the correct image based on scientific knowledge, with performance often resembling random guessing. Similarly, LMMs face challenges in this area. However, SciScore stands out by demonstrating exceptional performance, achieving human-level accuracy after being trained on Science-T2I.
|
| 54 |
|
| 55 |
+
<img src="./exp.png" align="center">
|