MMSearch-R1 is an end-to-end RL framework that enables LMMs to perform on-demand, multi-turn search with real-world multimodal search tools.
active 2025-04-03 → 2026-05-07 (UTC)
Activity over time
Daily event counts in the loaded window
Line chart, 400 days from 2025-04-03 to 2026-05-07. Pushes: 4 total, peak 1 in a day. Pull requests: 0 total, peak 0 in a day. Issues: 21 total, peak 2 in a day. Comments: 47 total, peak 4 in a day. Stars: 196 total, peak 16 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 |
|---|---|---|---|---|
| vvukimy | 31 | 4 | 0 | 19 |
| vishwassathish | 9 | 0 | 0 | 7 |
| Helen1p | 4 | 0 | 0 | 2 |
| wyd99 | 4 | 0 | 0 | 3 |
| ZZZYYYLL | 3 | 0 | 0 | 3 |
| Haelles | 3 | 0 | 0 | 2 |
| fc20567 | 2 | 0 | 0 | 1 |
| hl0737 | 2 | 0 | 0 | 1 |
| Curry-Christopher | 2 | 0 | 0 | 1 |
| VAN-QIAN | 1 | 0 | 0 | 1 |
| jianpugh | 1 | 0 | 0 | 0 |
| circlePi | 1 | 0 | 0 | 0 |
| zhangxy-2019 | 1 | 0 | 0 | 1 |
| BlueBlueFF | 1 | 0 | 0 | 1 |
| suyan-liang | 1 | 0 | 0 | 1 |
| DietDietDiet | 1 | 0 | 0 | 1 |
| bixie6868 | 1 | 0 | 0 | 0 |
Recent activity
Latest issues, pull requests and releases
- Issue comment#29vishwassathish2026-01-20 06:38Which LiveVQA dataset were the models evaluated on?
- Issue#29vvukimy2026-01-15 06:15Which LiveVQA dataset were the models evaluated on?
- Issue comment#8wyd992025-12-10 06:12training with image search api
- Issue#27wyd992025-12-10 06:07FSDPVLLMShardingManager' is deprecated
- Issue comment#27wyd992025-12-10 06:07FSDPVLLMShardingManager' is deprecated
- Issue comment#23vvukimy2025-12-08 03:52Does the current implementation support GRPO training with LoRA adapters?
- Issue comment#28VAN-QIAN2025-12-08 03:37Questions about the ground truth answers of sampled 2000 Infoseek Test Split
- Issue comment#13wyd992025-12-02 05:55Hello, I would like to know what version of verl you are using. I currently get the error ImportError: cannot import name '_timer' from 'verl.trainer.ppo.ray_trainer'
- Issue#26Curry-Christopher2025-11-05 12:56can't open image search results cache
- Issue comment#26Curry-Christopher2025-11-05 12:56can't open image search results cache
- Issue comment#7Helen1p2025-09-18 11:29meet error when attach model responses to vllm_inputs
- Issue#19vishwassathish2025-09-16 03:38Model Hallucinates after a few steps when training
- Issue#22Helen1p2025-09-15 17:58ZeroDivisionError: integer modulo by zero
- Issue comment#22Helen1p2025-09-15 17:58ZeroDivisionError: integer modulo by zero
- Issue comment#22vvukimy2025-09-15 11:41ZeroDivisionError: integer modulo by zero
- Issue#22Helen1p2025-09-15 09:38ZeroDivisionError: integer modulo by zero
- Issue comment#20ZZZYYYLL2025-09-15 01:54ValueError: Image features and image tokens do not match during training with multi-image inputs and multiple image search calls
- Issue comment#20vvukimy2025-09-14 03:47ValueError: Image features and image tokens do not match during training with multi-image inputs and multiple image search calls
- Issue comment#19vvukimy2025-09-14 03:09Model Hallucinates after a few steps when training
- Issue comment#20ZZZYYYLL2025-09-11 18:25ValueError: Image features and image tokens do not match during training with multi-image inputs and multiple image search calls
- Issue comment#19vvukimy2025-09-10 03:10Model Hallucinates after a few steps when training
- Issue comment#19vishwassathish2025-09-09 22:49Model Hallucinates after a few steps when training
- Issue comment#19vishwassathish2025-09-02 22:34Model Hallucinates after a few steps when training
- Issue comment#20ZZZYYYLL2025-09-02 07:35ValueError: Image features and image tokens do not match during training with multi-image inputs and multiple image search calls
- Issue comment#19vishwassathish2025-09-02 06:31Model Hallucinates after a few steps when training
Totals cover only the window loaded into ClickHouse and count events, not GitHub's lifetime totals — 196 stars here means stars gained during the window, not the repo's star count.