Datasets:
Update README.md (#4)
Browse files- Update README.md (9962fae83c85f85a099a0b328a638c87ab54551a)
Co-authored-by: Wentao Ma <tonymwt@users.noreply.huggingface.co>
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***
|