字节放出了一款具备长期记忆能力的多模态智能体：M3-Agent是一款具备长期记忆能力的多模态智能体，能够实时处理视觉和听觉输入，并将其转化为长期记忆。其记忆系统以实体为中心，支持多轮迭代推理，并在基准测试中表现优秀，尤其在需要长期记忆和跨模态推理的任务上。
active 2025-08-12 → 2026-06-10 (UTC)
Activity over time
Daily event counts in the loaded window
Line chart, 303 days from 2025-08-12 to 2026-06-10. Pushes: 10 total, peak 2 in a day. Pull requests: 0 total, peak 0 in a day. Issues: 8 total, peak 1 in a day. Comments: 12 total, peak 2 in a day. Stars: 623 total, peak 59 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 |
|---|---|---|---|---|
| hyc2026 | 11 | 10 | 0 | 1 |
| Celina-love-sweet | 5 | 0 | 0 | 3 |
| jbymy2000 | 4 | 0 | 0 | 1 |
| HarryHsing | 2 | 0 | 0 | 1 |
| jiTest921 | 2 | 0 | 0 | 2 |
| xiexie123 | 1 | 0 | 0 | 1 |
| Prepanpan | 1 | 0 | 0 | 1 |
| hongliang-wei | 1 | 0 | 0 | 0 |
| Happy2Git | 1 | 0 | 0 | 1 |
| gihunsong | 1 | 0 | 0 | 0 |
| NIneeeeeem | 1 | 0 | 0 | 1 |
Recent activity
Latest issues, pull requests and releases
- Issue comment#12Celina-love-sweet2025-12-16 11:25M3-Agent-Memorization是否可以vllm推理？
- Issue comment#5Celina-love-sweet2025-12-16 05:59An error occurred during the speech processing process
- Issue#16Celina-love-sweet2025-12-16 05:43请问您可以提供您实验中关于videomme-long的记忆图吗
- Issue#15Celina-love-sweet2025-12-08 07:51实验结果的疑问
- Issue comment#11Celina-love-sweet2025-11-05 01:12记忆图可视化错误
- Issue#8gihunsong2025-09-19 00:41Question on Achieving Real-Time Performance in M3-Agent
- Issue comment#2Happy2Git2025-09-16 10:23generate_memory_qwen is missing and memorization_intermediate_outputs stuck
- Issue#3hongliang-wei2025-09-10 16:46End-to-end inference code or demo?
- Issue#2jbymy20002025-09-04 09:03generate_memory_qwen is missing and memorization_intermediate_outputs stuck
- Issue comment#2jbymy20002025-09-02 08:37generate_memory_qwen is missing and memorization_intermediate_outputs stuck
- Issue#2jbymy20002025-09-02 08:31generate_memory_qwen is missing and memorization_intermediate_outputs stuck
- Issue comment#6NIneeeeeem2025-08-29 06:20Could you please double check the Video-MME-Long score for MA-LMM in your main result table?
- Issue comment#5jiTest9212025-08-29 03:01An error occurred during the speech processing process
- Issue comment#5jiTest9212025-08-28 09:27An error occurred during the speech processing process
- Issue comment#2Prepanpan2025-08-28 03:21generate_memory_qwen is missing and memorization_intermediate_outputs stuck
- Issue comment#3hyc20262025-08-20 02:38End-to-end inference code or demo?
- Issue comment#1HarryHsing2025-08-19 09:26README file for the training code
- Issue comment#2xiexie1232025-08-19 01:41generate_memory_qwen is missing and memorization_intermediate_outputs stuck
- Issue#2jbymy20002025-08-18 13:11generate_memory_qwen is missing and memorization_intermediate_outputs stuck
- Issue#1HarryHsing2025-08-14 10:29README file for the training code
Totals cover only the window loaded into ClickHouse and count events, not GitHub's lifetime totals — 623 stars here means stars gained during the window, not the repo's star count.