We lifted 12 functions out of this paper's own repositories and ran 9 of them in a sandbox. "Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results.
| Repository | Role | Ran |
|---|---|---|
| jayleicn/ClipBERT | canonical | 9 of 12 |
| Function | Status | Where it lives |
|---|---|---|
| conv3x3 | Ran | jayleicn/ClipBERT/src/modeling/grid_feat.py code served (permissive licence) · get_code("17e89eef72e37efe") |
| get_padding | Ran | jayleicn/ClipBERT/src/datasets/data_utils.py code served (permissive licence) · get_code("0cf387da46369b1a") |
| get_random_sample_indices | Ran | jayleicn/ClipBERT/src/modeling/modeling.py code served (permissive licence) · get_code("f31be4ddaa3ec6c2") |
| get_start_end_idx | Ran | jayleicn/ClipBERT/src/datasets/decoder.py code served (permissive licence) · get_code("707f6cd75833c65d") |
| get_video_decoding_kwargs | Ran | jayleicn/ClipBERT/src/datasets/dataset_base.py code served (permissive licence) · get_code("b6331823c359e9d8") |
| image_to_tensor | Ran | jayleicn/ClipBERT/src/datasets/data_utils.py code served (permissive licence) · get_code("e9242c50098a6ea2") |
| mask_batch_text_tokens | Ran | jayleicn/ClipBERT/src/datasets/data_utils.py code served (permissive licence) · get_code("f828f173e6c2fea2") |
| move_to_cuda | Ran | jayleicn/ClipBERT/src/datasets/dataloader.py code served (permissive licence) · get_code("b68769dd52ad6c44") |
| temporal_sampling | Ran | jayleicn/ClipBERT/src/datasets/decoder.py code served (permissive licence) · get_code("f307b4793e8204b7") |
| instance_bce_with_logits | Not yet run | jayleicn/ClipBERT/src/modeling/modeling.py code served (permissive licence) · get_code("5a4d93d8be97ccd9") |
| parse_with_config | Not yet run | jayleicn/ClipBERT/src/configs/config.py code served (permissive licence) · get_code("3e81145412a40e36") |
| pyav_decode_stream | Not yet run | jayleicn/ClipBERT/src/datasets/decoder.py code served (permissive licence) · get_code("b28b46ef7776c434") |
Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.
Training an effective video-and-language model intuitively requires multiple frames as model inputs. However, it is unclear whether using multiple frames is beneficial to downstream tasks, and if yes, whether the performance gain is worth the drastically-increased computation and memory costs resulting from using more frames. In this work, we explore single-frame models for video-and-language learning. On a diverse set of video-and-language tasks (including text-to-video retrieval and video question answering), we show the surprising result that, with large-scale pre-training and a proper frame ensemble strategy at inference time, a single-frame trained model that does not consider temporal information can achieve better performance than existing methods that use multiple frames for training. This result reveals the existence of a strong "static appearance bias" in popular video-and-language datasets. Therefore, to allow for a more comprehensive evaluation of video-and-language models, we propose two new retrieval tasks based on existing fine-grained action recognition datasets that encourage temporal modeling. Our code is available at https://github.com/jayleicn/singularity
The same record, over MCP at https://syntology.ai/mcp:
get_harvested_code_for_paper("2206.03428")
get_code_for_paper("2206.03428")
have("2206.03428")
Connect an agent — have() is free.