Group MMMU images into one image (#83)

* update

* update font

* Add matplotlib.font_manager import in utils.py

* Refactor font handling in add_order_label function in utils.py

* group mmmu

---------

Co-authored-by: Li Bo <[email protected]>

Author: 2 people

lmms_eval/tasks/mmmu/mmmu_group_img.yaml

@@ -0,0 +1,4 @@
+group: mmmu_group_img
+task:
+- mmmu_val_group_img
+- mmmu_test_group_img

lmms_eval/tasks/mmmu/mmmu_group_img_test.yaml

@@ -0,0 +1,19 @@
+dataset_path: lmms-lab/MMMU
+task: "mmmu_test_group_img"
+test_split: test
+output_type: generate_until
+doc_to_visual: !function utils_group_img.mmmu_doc_to_visual
+doc_to_text: !function utils_group_img.mmmu_doc_to_text
+doc_to_target: "answer"
+# The return value of process_results will be used by metrics
+process_results: !function utils_group_img.mmmu_process_results
+# Note that the metric name can be either a registed metric function (such as the case for GQA) or a key name returned by process_results
+generation_kwargs:
+  max_new_tokens: 16
+  image_aspect_ratio: original
+metric_list:
+  - metric: submission
+    aggregation: !function utils_group_img.mmmu_test_aggregate_results_for_submission
+    higher_is_better: true
+metadata:
+  - version: 0.0

lmms_eval/tasks/mmmu/mmmu_group_img_val.yaml

@@ -0,0 +1,21 @@
+dataset_path: lmms-lab/MMMU
+task: "mmmu_val_group_img"
+test_split: validation
+output_type: generate_until
+doc_to_visual: !function utils_group_img.mmmu_doc_to_visual
+doc_to_text: !function utils_group_img.mmmu_doc_to_text
+doc_to_target: "answer"
+# The return value of process_results will be used by metrics
+process_results: !function utils_group_img.mmmu_process_results
+# Note that the metric name can be either a registed metric function (such as the case for GQA) or a key name returned by process_results
+generation_kwargs:
+  max_new_tokens: 16
+model_specific_generation_kwargs:
+  llava:
+    image_aspect_ratio: original
+metric_list:
+  - metric: mmmu_acc
+    aggregation: !function utils_group_img.mmmu_aggregate_results
+    higher_is_better: true
+metadata:
+  - version: 0.0