Bruno Rigal

I don't understand the meaning of overlap_mod and overlap easy in get_official_eval_result

5

In the official benchmark, for cars the overlap is supposed to be 70%. Which line corresponds to the official benchmark as given [here](http://www.cvlibs.net/datasets/kitti/eval_object.php?obj_benchmark=3d)?

Generating load profile in linux, libraries required

5

Hi, tanks for this nice software I would like to run the software in linux. I did the chmod command you suggested in a closed issue. However, once it is...

Erreur avec le le logger mlflow

1

### Bug description With the mlflow logger, I have the following error: ```Epoch 9: 100%|█| 23/23 [00:40posixpath.normpath(name) will return a str even for a Path as input and then `norm...

I have a hard time reproducing the benchmark code

1

### 🐛 Describe the bug I get this error: ``` (olmocr) brigal@c6b03d89ae44 /workspace/perso/dolma (main) $ python -m olmocr.bench.convert olmocr_pipeline --dir /data_project/data/docparser/benchmarks/olmocr_bench2/bench_data/ Traceback (most recent call last): File "", line 198,...

Thanks for this very interesting library, I did not see any specific implementations for Grouped Query Attention or Multi Head Latent attention, which seem to be very popular these days.