Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

MixEval-X Image / Video #434

Merged
merged 10 commits into from
Dec 3, 2024
Merged

MixEval-X Image / Video #434

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
10 commits
Select commit Hold shift + click to select a range
1fad5d9
reformat mix eval
pufanyi Nov 30, 2024
715df25
video2text fix
pufanyi Nov 30, 2024
cc232f7
video
pufanyi Dec 1, 2024
b89847a
Add image-to-text evaluation tasks and templates
pufanyi Dec 1, 2024
5e23413
Refactor image-to-text evaluation tasks and update configurations
pufanyi Dec 1, 2024
4fddfff
Enhance LlamaVision model with video loading improvements and configu…
pufanyi Dec 1, 2024
b4ccbe1
fix internvl2
pufanyi Dec 1, 2024
aa5c5bf
llava vid default numframe too small
pufanyi Dec 1, 2024
11540d1
add max_new_tokens
pufanyi Dec 2, 2024
57f8c43
remove "with a short phrase" for `gpt4v`
pufanyi Dec 2, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Prev Previous commit
Next Next commit
video2text fix
  • Loading branch information
@pufanyi
pufanyi committed Nov 30, 2024
commit 715df25ce4463381f4881a728f0bb6f583fbedd2
6 changes: 3 additions & 3 deletions lmms_eval/tasks/mix_evals/video2text/mix_evals_video2text_freeform.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,10 +1,10 @@
dataset_name: "video2text_closeended_free-form"
task: "mix_evals_video2text_freeform"
test_split: test
dataset_name: "video2text"
test_split: free_form
output_type: generate_until
doc_to_visual: !function utils.mix_evals_video2text_doc_to_visual
doc_to_text: !function utils.mix_evals_video2text_doc_to_text
doc_to_target: "{{target}}"
doc_to_target: "{{reference_answer}}"
process_results: !function utils.mix_evals_video2text_process_results_freeform
metric_list:
- metric: gpt_eval
Expand Down
6 changes: 3 additions & 3 deletions lmms_eval/tasks/mix_evals/video2text/mix_evals_video2text_mc.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,11 +1,11 @@
include: _default_template_yaml
dataset_name: "video2text_closeended_multiple-choice"
task: "mix_evals_video2text_mc"
test_split: test
dataset_name: "video2text"
test_split: multiple_choice
output_type: generate_until
doc_to_visual: !function utils.mix_evals_video2text_doc_to_visual
doc_to_text: !function utils.mix_evals_video2text_doc_to_text
doc_to_target: "{{target}}"
doc_to_target: "{{reference_answer}}"

generation_kwargs:
max_new_tokens: 5
Expand Down
6 changes: 3 additions & 3 deletions lmms_eval/tasks/mix_evals/video2text/utils.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -232,13 +232,13 @@ def mix_evals_video2text_process_results_open_convs(doc, result):

def mix_evals_video2text_process_results_freeform(doc, result):
pred = result[0]
ground_truth_str = ", ".join([f'"{gt}"' for gt in doc["target"]])
ground_truth_str = ", ".join([f'"{gt}"' for gt in doc["reference_answer"]])
ground_truth_str = f"[{ground_truth_str}]"
content = eval_prompt.format(model_response=pred, ground_truth=ground_truth_str)
eval_answer, model_name = get_eval(model_response=pred, ground_truth=ground_truth_str, max_tokens=1024)
return {
"submission": {"pred": pred, "question_idx": doc["question_index"], "target": doc["target"], "eval_answer": eval_answer, "gpt_prompt": content},
"gpt_eval": {"pred": pred, "question_idx": doc["question_index"], "target": doc["target"], "eval_answer": eval_answer, "gpt_prompt": content},
"submission": {"pred": pred, "question_idx": doc["id"], "target": doc["reference_answer"], "eval_answer": eval_answer, "gpt_prompt": content},
"gpt_eval": {"pred": pred, "question_idx": doc["id"], "target": doc["reference_answer"], "eval_answer": eval_answer, "gpt_prompt": content},
}


Expand Down