github
/
minimind-v
mirror of https://github.com/jingyaogong/minimind-v.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Commit Graph

  • 50decbdfeb
         Merge e2332de909 into 1862b633fc ian 2026-09-30 18:20:51 +0800
  • e2332de909 Fix tensor input handling in MiniMindVLM.get_image_embeddings ian 2026-09-30 18:10:37 +0800
  • 1862b633fc [fix] residual step order master jingyaogong 2026-09-23 00:07:30 +0800
  • dacc687889 [fix] fp16 attention overflow jingyaogong 2026-09-20 20:39:51 +0800
  • fa36e4861d [fix] top-1 router gradient jingyaogong 2026-09-18 16:34:37 +0800
  • 740d467ece [update] supporter charts jingyaogong 2026-08-06 22:08:15 +0800
  • b1744b7858 Fix typo in minimind-v bglglzd 2026-07-12 22:42:59 +0300
  • 959561791f [fix] robustness jingyaogong 2026-06-28 18:13:08 +0800
  • db165dfcad
         Merge 2a65e6a675 into e57130e9e7 dynorth 2026-05-21 16:26:09 +0800
  • fa71da3bbf
         Merge 31b57a0569 into e57130e9e7 100apps 2026-05-19 17:47:40 +0800
  • e57130e9e7 [fix] dll jingyaogong 2026-05-19 17:43:47 +0800
  • 31b57a0569
         Merge branch 'master' into master 100apps 2026-05-16 08:36:18 +0800
  • 2b94dcb3d0 [update] num_workers=2 jingyaogong 2026-05-12 09:54:46 +0800
  • 1b5c16f674 feat: use HuggingFace datasets for memory-mapped data loading Shawn Tsai 2026-05-11 18:22:19 +0000
  • 8b96a1efd9 fix(train): 解决Windows平台数据加载线程数问题 100apps 2026-05-11 23:43:45 +0800
  • 2c2f4b5161 remove minimind-3v 100apps 2026-05-11 23:31:40 +0800
  • 8b61e99080 chore(git): 更新.gitignore以忽略minimind-3v目录 100apps 2026-05-11 23:30:46 +0800
  • 8fb354789b chore(deps): 使用pyproject.toml和uv管理依赖环境 100apps 2026-05-11 23:28:03 +0800
  • ee18831dce [fix] repetition_penalty jingyaogong 2026-05-07 19:08:14 +0800
  • b6a2532801 [update] readme jingyaogong 2026-05-02 00:35:40 +0800
  • 1e86f201a2 [update] readme jingyaogong 2026-05-01 12:11:48 +0800
  • 6fa6127b69 [update] readme jingyaogong 2026-04-28 17:26:02 +0800
  • 0a9c06fa68 [update] readme jingyaogong 2026-04-27 10:57:35 +0800
  • fb38500054 [update] readme jingyaogong 2026-04-27 10:51:46 +0800
  • c9c99a96c0 [update] website content website jingyaogong 2026-04-20 16:55:17 +0800
  • ff81f88833 [update] minimind-3v jingyaogong 2026-04-20 16:54:49 +0800
  • b759dd8b20 [update] website jingyaogong 2026-04-20 09:01:00 +0800
  • 92ae3f963a [update] readme jingyaogong 2026-04-20 00:06:27 +0800
  • 9c3343dd88 [update] minimind-3v jingyaogong 2026-04-19 23:51:48 +0800
  • db9667c3a2 [update] image jingyaogong 2026-04-04 14:58:21 +0800
  • 333654d6e9 [update] image jingyaogong 2026-04-04 14:57:45 +0800
  • f74124f6b9 [fix] bugs jingyaogong 2026-04-04 12:43:13 +0800
  • 5098d67d9f [fix] issue#114 jingyaogong 2026-04-04 12:13:28 +0800
  • 41c6d9e2a5
         Merge pull request #115 from CgggggY/fix-single-batch-multi-image-shape jingyaogong 2026-04-04 12:06:04 +0800
  • 9e59a77080 Fix vision_tensors shape for bs=1 multi-image inputs CgggggY 2026-04-04 11:32:38 +0800
  • c840cdea35 Personal study notes for model_vlm (local only) CgggggY 2026-04-04 11:22:27 +0800
  • 4bc6524ac2 [fix] sync minimind jingyaogong 2026-04-02 16:34:38 +0800
  • 887a42d9cb [update] model jingyaogong 2026-04-02 15:29:50 +0800
  • 28acd77d42 [fix] dummy grad jingyaogong 2026-04-01 17:56:44 +0800
  • 48adb0fda0 [fix] pr113 jingyaogong 2026-04-01 15:40:25 +0800
  • ac4c506eeb
         Merge pull request #113 from JX446/master jingyaogong 2026-04-01 15:31:33 +0800
  • eac29745da fix: 修复 eval_vlm.py 中 f-string 引号嵌套导致的语法错误 DJY 2026-04-01 15:11:38 +0800
  • 66ad1c4b7f fix: add find_unused_parameters=True to DDP for mixed text-image batches 王路宝 2026-04-01 14:20:30 +0800
  • 14cfa88629 [update] requirement jingyaogong 2026-03-30 11:51:18 +0800
  • 9089310ede [update] fp16 inference jingyaogong 2026-03-27 16:22:38 +0800
  • 1a6d650e9a [update] readme jingyaogong 2026-03-26 11:46:05 +0800
  • e51dcf0fda [update] minimind-3v jingyaogong 2026-03-25 17:35:05 +0800
  • e23503235b [update] model jingyaogong 2026-03-26 00:07:04 +0800
  • 2a65e6a675 [eval] Add DPO preference accuracy evaluation for DriveMind-V MiniMind-V User 2026-03-07 00:26:12 +0000
  • 823ce1e4fc [feat] Implement DPO training for driving safety alignment MiniMind-V User 2026-03-06 05:00:42 +0000
  • 62f47fb844 [fix] LoRA/RAG: correct target modules, vlm_checkpoint call, and gitignore MiniMind-V User 2026-03-06 04:49:40 +0000
  • a86648ed66 [feat] Implement LoRA + RAG for DriveMind-V MiniMind-V User 2026-03-05 01:13:59 +0000
  • 56ab6ebd13 [update] readme jingyaogong 2026-02-05 00:23:20 +0800
  • 33db76fe10 [update] dataset name jingyaogong 2026-02-05 00:22:58 +0800
  • b28715e6fd [update] vlm params jingyaogong 2026-02-04 16:41:17 +0800
  • 7e094afa59 [update] shuffle data jingyaogong 2026-01-18 17:02:24 +0800
  • 326b24727e [update] sync minimind jingyaogong 2026-01-15 21:36:28 +0800
  • d2bdff3269 [feat] test parquet jingyaogong 2025-12-27 10:56:07 +0800
  • fef284776f [feat] delete images jingyaogong 2025-12-27 10:53:38 +0800
  • 3684285a91 [feat] support parquet data jingyaogong 2025-12-27 10:49:50 +0800
  • 1863743ee9 [feat] update readme jingyaogong 2025-12-27 10:48:54 +0800
  • 32cf4c5c01 [fix] add gc jingyaogong 2025-12-27 01:09:33 +0800
  • 244e172f75 [fix] table flow jingyaogong 2025-12-24 13:45:41 +0800
  • b7cb98c136 [fix] reduce aux_loss_alpha jingyaogong 2025-12-05 23:07:55 +0800
  • 13c3a2c093 [fix] cuda memory jingyaogong 2025-12-01 16:35:05 +0800
  • e03af46d12 [feat] update yarn jingyaogong 2025-12-01 16:34:00 +0800
  • a157b66ace [feat] del cuda memory jingyaogong 2025-11-27 19:37:39 +0800
  • e613613a13 [feat] update model jingyaogong 2025-11-27 19:36:41 +0800
  • d698d98dc5 [feat] update readme jingyaogong 2025-11-11 11:12:09 +0800
  • 92ca9959bb [feat] update readme jingyaogong 2025-10-30 23:32:12 +0800
  • ca15b0440f [feat] update readme jingyaogong 2025-10-30 23:31:19 +0800
  • 2e548ba08e [feat] update datasets jingyaogong 2025-10-30 11:08:30 +0800
  • 25d25383e7 [feat] update requirements jingyaogong 2025-10-30 11:06:20 +0800
  • 7605850e1d [feat] update args jingyaogong 2025-10-30 10:50:54 +0800
  • 88b0cd89ea [feat] update readme jingyaogong 2025-10-29 12:13:44 +0800
  • 778aeb2946 [feat] update trainer jingyaogong 2025-10-28 23:34:16 +0800
  • 2ce5c31170 [feat] update sft jingyaogong 2025-10-26 21:24:19 +0800
  • c2bfead000 [feat] update readme jingyaogong 2025-10-26 20:55:02 +0800
  • 54fcbb04c8 [feat] update eval-vlm jingyaogong 2025-10-26 20:54:37 +0800
  • aebc6fb191 [feat] pause-training jingyaogong 2025-10-26 20:54:13 +0800
  • e7e2409d8d [feat] update website jingyaogong 2025-10-26 20:51:54 +0800
  • 4c3b5bacbf [fix] 1024-bugs jingyaogong 2025-10-24 00:35:56 +0800
  • 094f88a087 [feat] update readme jingyaogong 2025-10-22 10:58:52 +0800
  • 8bed2ec00b [feat] update readme jingyaogong 2025-10-21 22:53:00 +0800
  • 5361ee9a8c [feat] update website jingyaogong 2025-10-21 22:23:16 +0800
  • a9374500c5
         Merge aefaaedd5c into 974a666741 shanfenglan 2025-10-21 14:11:35 +0800
  • 974a666741 [feat] update readme jingyaogong 2025-10-17 00:04:48 +0800
  • 4d2accbcef [feat] update images jingyaogong 2025-10-17 00:03:58 +0800
  • 58dfb9779c [feat] update minimind model jingyaogong 2025-10-17 00:03:13 +0800
  • 3d5d6aa177 fix cdn jingyaogong 2025-10-12 23:34:25 +0800
  • 5f47204191 init minimind-v docs v2 jingyaogong 2025-10-12 23:02:50 +0800
  • aefaaedd5c add .gitignore and update README files ubuntu_server 2025-06-28 13:11:55 +0800
  • 1ed44bba5d 预训练改为全参数训练，补全模型缺少的generate Jinshijiming 2025-05-02 19:47:13 +0800
  • a8572190a8 update eval jingyaogong 2025-04-27 18:38:51 +0800
  • 623a130c34 250427 jingyaogong 2025-04-27 18:13:22 +0800
  • 6ae1b3e6ae
         Merge b696a7f084 into 26de815253 ckq 2025-04-11 13:07:21 +0800
  • 26de815253 update mask jingyaogong 2025-04-10 22:36:34 +0800
  • 5f130d142d moe note jingyaogong 2025-04-09 17:37:51 +0800
  • 139cc98bb9 moe note jingyaogong 2025-04-09 17:37:06 +0800
  • bb0ec7a420 update readme jingyaogong 2025-04-09 17:03:13 +0800