[EMNLP 2025 main] Code for "Stop Looking for Important Tokens in Multimodal Language Models: Duplication Matters More"
active 2025-02-07 → 2026-05-03 (UTC)
Activity over time
Daily event counts in the loaded window
Line chart, 451 days from 2025-02-07 to 2026-05-03. Pushes: 24 total, peak 7 in a day. Pull requests: 0 total, peak 0 in a day. Issues: 15 total, peak 2 in a day. Comments: 15 total, peak 2 in a day. Stars: 66 total, peak 3 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 |
|---|---|---|---|---|
| ZichenWen1 | 34 | 24 | 0 | 7 |
| coolKeen | 5 | 0 | 0 | 1 |
| L-JunJie | 3 | 0 | 0 | 2 |
| Luo-Zhongwei | 2 | 0 | 0 | 0 |
| yiru0410 | 1 | 0 | 0 | 1 |
| BIIIIU | 1 | 0 | 0 | 0 |
| RewindL | 1 | 0 | 0 | 0 |
| SamithVa | 1 | 0 | 0 | 0 |
| Mr-xiu | 1 | 0 | 0 | 1 |
| Korn-No1 | 1 | 0 | 0 | 1 |
| Sugar929 | 1 | 0 | 0 | 1 |
| shanenian | 1 | 0 | 0 | 1 |
| danzel-crazy | 1 | 0 | 0 | 0 |
| ZhuYe0729 | 1 | 0 | 0 | 0 |
Recent activity
Latest issues, pull requests and releases
- Issue#18danzel-crazy2026-05-03 09:19Correct parameter setup for LLaVA-NeXT results (640, 320, 160 tokens)
- Issue comment#11Korn-No12026-02-03 12:13Could you please guide me to find the code for evaluating the Total Time ↓ Prefilling Time ↓ FLOPs ↓ KV Cache ↓ POPE ↑ Speedup ↑ (Min:Sec) (Min:Sec) (MB) (F1-Score) (Total) (Prefilling)?
- Issue comment#13Sugar9292025-11-10 16:34RuntimeError: selected index k out of range
- Issue comment#13L-JunJie2025-09-25 01:42RuntimeError: selected index k out of range
- Issue comment#13L-JunJie2025-09-24 07:48RuntimeError: selected index k out of range
- Issue comment#14ZichenWen12025-09-24 04:36Does your model support multi-image input?
- Issue#12ZichenWen12025-09-24 04:33Does Qwen2.5-VL DART have to use transformers==4.55.4 ?
- Issue#14BIIIIU2025-09-23 12:53Does your model support multi-image input?
- Issue#13L-JunJie2025-09-23 04:18RuntimeError: selected index k out of range
- Issue comment#12shanenian2025-09-03 05:20Does Qwen2.5-VL DART have to use transformers==4.55.4 ?
- Issue#9ZhuYe07292025-06-26 01:29speedup performance testing help
- Issue#7coolKeen2025-04-30 09:35batch >= 2
- Issue#6coolKeen2025-04-30 09:33how to calculate the time like table 2 in the paper?
- Issue#6coolKeen2025-04-27 08:24how to calculate the time like table 2 in the paper?
- Issue comment#5ZichenWen12025-04-25 11:202025-04-23 17:16:36.897 | ERROR | __main__:cli_evaluate:349 - Error during evaluation: Unexpected kwargs
- Issue comment#5ZichenWen12025-04-25 11:112025-04-23 17:16:36.897 | ERROR | __main__:cli_evaluate:349 - Error during evaluation: Unexpected kwargs
- Issue comment#4coolKeen2025-04-23 10:48Minimal Speed Improvement on Qwen2-VL After Pruning
- Issue#5coolKeen2025-04-23 09:232025-04-23 17:16:36.897 | ERROR | __main__:cli_evaluate:349 - Error during evaluation: Unexpected kwargs
- Issue#4ZichenWen12025-04-16 05:44Minimal Speed Improvement on Qwen2-VL After Pruning
- Issue#1ZichenWen12025-04-16 05:43Request for Qwen2VL code
- Issue#3Luo-Zhongwei2025-04-08 21:36HI， I am a new learner about deep learning , I want to know how to apply the token pruning method you proposed to other VLM?
- Issue comment#3ZichenWen12025-04-08 05:05HI， I am a new learner about deep learning , I want to know how to apply the token pruning method you proposed to other VLM?
- Issue comment#4ZichenWen12025-04-05 11:54Minimal Speed Improvement on Qwen2-VL After Pruning
- Issue#4SamithVa2025-04-01 02:05Minimal Speed Improvement on Qwen2-VL After Pruning
- Issue#3Luo-Zhongwei2025-03-29 09:34HI， I am a new learner about deep learning , I want to know how to apply the token pruning method you proposed to other VLM?
Totals cover only the window loaded into ClickHouse and count events, not GitHub's lifetime totals — 66 stars here means stars gained during the window, not the repo's star count.