Datasets:
Update README.md (#4)
Browse files- Update README.md (9962fae83c85f85a099a0b328a638c87ab54551a)
Co-authored-by: Wentao Ma <[email protected]>
README.md
CHANGED
@@ -142,4 +142,32 @@ Each example in the dataset contains:
|
|
142 |
--num_frames 32 \
|
143 |
--max_retries 10 \
|
144 |
--num_threads 1
|
145 |
-
```
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
142 |
--num_frames 32 \
|
143 |
--max_retries 10 \
|
144 |
--num_threads 1
|
145 |
+
```
|
146 |
+
|
147 |
+
5. **Judge the results**
|
148 |
+
```bash
|
149 |
+
cd VideoEval-Pro
|
150 |
+
|
151 |
+
# Set PYTHONPATH
|
152 |
+
export PYTHONPATH=.
|
153 |
+
|
154 |
+
# Run judge script *gpt4o_judge.py* with the following parameters:
|
155 |
+
# --input_path: Path to save output results
|
156 |
+
# --output_path: Path to judged results
|
157 |
+
# --model_name: Version of the judge model
|
158 |
+
# --num_threads: Number of threads for parallel processing
|
159 |
+
|
160 |
+
python tools/gpt4o_judge.py \
|
161 |
+
--input_path <path_to_saved_results> \
|
162 |
+
--output_path <path_to_judged_results> \
|
163 |
+
--model_name <model_version> \
|
164 |
+
--num_threads <num_threads>
|
165 |
+
|
166 |
+
E.g.:
|
167 |
+
python tools/gpt4o_judge.py \
|
168 |
+
--input_path ./results/qwen_results.jsonl \
|
169 |
+
--output_path ./results/qwen_results_judged.jsonl \
|
170 |
+
--model_name gpt-4o-2024-08-06 \
|
171 |
+
--num_threads 1
|
172 |
+
```
|
173 |
+
**Note: the released results are judged by *gpt-4o-2024-08-06***
|