The official code of the paper "PyramidDrop: Accelerating Your Large Vision-Language Models via Pyramid Visual Redundancy Reduction".
active 2024-10-17 → 2026-01-26 (UTC)
Activity over time
Daily event counts in the loaded window
Line chart, 467 days from 2024-10-17 to 2026-01-26. Pushes: 29 total, peak 9 in a day. Pull requests: 0 total, peak 0 in a day. Issues: 25 total, peak 2 in a day. Comments: 27 total, peak 2 in a day. Stars: 75 total, peak 8 in a day.
- Pushes
- Pull requests
- Issues
- Comments
- Stars
Stars, PRs, issues and forks are under-captured in the later part of this window. GH Archive progressively stopped capturing non-push events during 2026 — −95% or worse by the end of the window. Every series here except Pushes fades for that reason, so a decline above reflects the archive, not this repository. Pushes stay reliable throughout, so read them, and the contributor counts derived from them, as the real signal. Data health has the measurements.
Top contributors
Pushes, PRs, issues, reviews and comments — stars and forks excluded, so this is contribution rather than popularity
| Contributor | Contributions | Pushes | PRs | Comments |
|---|---|---|---|---|
| Cooperx521 | 40 | 29 | 0 | 10 |
| YuchenLiu98 | 6 | 0 | 0 | 3 |
| ymy-k | 4 | 0 | 0 | 2 |
| tianjiedai | 4 | 0 | 0 | 3 |
| Yxxxb | 3 | 0 | 0 | 1 |
| rosarubus72 | 3 | 0 | 0 | 1 |
| ashun989 | 2 | 0 | 0 | 2 |
| liuyafeng0609 | 2 | 0 | 0 | 1 |
| FanshuoZeng | 2 | 0 | 0 | 0 |
| Man-PaperRejected | 1 | 0 | 0 | 1 |
| HarrisonWu42 | 1 | 0 | 0 | 0 |
| NielsRogge | 1 | 0 | 0 | 0 |
| junkerman2004 | 1 | 0 | 0 | 0 |
| tanghao2118 | 1 | 0 | 0 | 0 |
| zhangce01 | 1 | 0 | 0 | 1 |
| chenghuaWang | 1 | 0 | 0 | 0 |
| Wuzimeng | 1 | 0 | 0 | 0 |
| Maxlium | 1 | 0 | 0 | 0 |
| 123998809 | 1 | 0 | 0 | 1 |
| qq1332427275 | 1 | 0 | 0 | 0 |
Recent activity
Latest issues, pull requests and releases
- Issue#30jiuyixu252026-01-26 19:37The version of Transformers
- Issue comment#29liuyafeng06092025-11-11 06:40How do I set different compression versions?
- Issue#29liuyafeng06092025-11-10 14:13How do I set different compression versions?
- Issue#28Wuzimeng2025-10-20 09:52Settings of Table 2
- Issue#27Maxlium2025-09-15 05:28the "layer_list" and "image_token_ratio_list" in the Table 1
- Issue comment#7Man-PaperRejected2025-08-15 09:26IndexError: too many indices for tensor of dimension 2 Pdrop_rank_drop
- Issue comment#26tianjiedai2025-06-25 07:19Have anyone tried evaluate PDrop under lmms-eval framework?
- Issue comment#12Cooperx5212025-06-25 04:03[Author] Please place any models that need to be compatible with pdrop here.
- Issue comment#12zhangce012025-06-20 07:37[Author] Please place any models that need to be compatible with pdrop here.
- Issue comment#26ashun9892025-06-19 11:45Have anyone tried evaluate PDrop under lmms-eval framework?
- Issue comment#12ashun9892025-06-19 11:09[Author] Please place any models that need to be compatible with pdrop here.
- Issue comment#201239988092025-06-11 07:16how to calculate the TFLOPs of Model
- Issue#26tianjiedai2025-06-11 07:08Have anyone tried evaluate PDrop under lmms-eval framework?
- Issue comment#20tianjiedai2025-06-11 06:41how to calculate the TFLOPs of Model
- Issue comment#22liaolea2025-05-31 06:20about average tokens
- Issue comment#22tianjiedai2025-05-31 06:04about average tokens
- Issue#24FanshuoZeng2025-05-09 03:22在推理时，当max_new_tokens较大的时候，推理时间相比于基线方法减少并不显著。
- Issue#21chenghuaWang2025-04-27 08:57Question about decoding stage.
- Issue#19FanshuoZeng2025-04-14 02:40使用PDrop进行LoRA微调自定义数据的train_flos
- Issue comment#18Cooperx5212025-04-10 04:06During the model inference process, does LLaVA-NeXT-7B use the llava-v1.6-vicuna-7b model or the open-llava-next-vicuna-7b model?
- Issue comment#17Cooperx5212025-04-06 11:05Support for FlashAttention-2 and SDPA？
- Issue#17HarrisonWu422025-04-06 06:29Support for FlashAttention-2 and SDPA？
- Issue comment#16Cooperx5212025-04-02 02:02Does pyramidDrop support multiple images?
- Issue#15tanghao21182025-03-23 10:47citation
- Issue#14qq13324272752025-03-21 01:56attention map visualization
Totals cover only the window loaded into ClickHouse and count events, not GitHub's lifetime totals — 75 stars here means stars gained during the window, not the repo's star count.