colin
/
Witllm
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Commit Graph

  • f74a5d29bd Update unsuper minist. master Colin 2024-11-04 14:23:29 +0800
  • df05002c90 Update unsuper minist. Colin 2024-11-03 15:15:33 +0800
  • 1bb41f0ee7 Update. Colin 2024-10-31 15:15:13 +0800
  • 2ad977a072 Update show. Colin 2024-10-31 00:01:57 +0800
  • 59449df047 Refine minist unsuper. Colin 2024-10-28 18:52:28 +0800
  • 5b2cd4da61 Add model.conv1.weight normal after update grad. Colin 2024-10-28 16:31:42 +0800
  • 6a0b47c674 Update show tools. Colin 2024-10-27 19:48:53 +0800
  • a0da2565fe Update minist unsuper. Colin 2024-10-22 19:13:19 +0800
  • 385c438c1c Update unsuper. Colin 2024-10-22 13:54:26 +0800
  • f3690fd47f Update unsuper learning. Colin 2024-10-07 16:39:29 +0800
  • 22464e7724 Update unsuper. Colin 2024-10-05 16:17:41 +0800
  • 45d5701835 Unsuper train with max confidense of conv output Colin 2024-10-04 01:30:18 +0800
  • 81f203ce59 update the max grad's weight. Colin 2024-09-22 15:18:08 +0800
  • 03516f6302 Remove unsuper pool layer. Colin 2024-09-21 17:44:57 +0800
  • 39cecd1146 mov optimizer to grad sub. Colin 2024-09-17 01:59:00 +0800
  • ce64a2f7aa Update show. Colin 2024-09-16 18:46:09 +0800
  • 8583bc56d7 Delete png. Colin 2024-09-16 18:29:15 +0800
  • d798295ace Update. Colin 2024-09-16 17:27:37 +0800
  • ad246c6c7f Update mnist unsuper learning. Colin 2024-09-08 15:22:12 +0800
  • dd07e54edd Update image. Colin 2024-09-07 16:29:01 +0800
  • a052fceaeb Update high parameter of minist. Colin 2024-09-02 18:06:47 +0800
  • e48e724ca9 Refine minist dump. Colin 2024-09-02 17:52:33 +0800
  • 4ad4c01403 Refine train. Colin 2024-09-01 18:15:07 +0800
  • c395fa7baa Refine to dump heat image. Colin 2024-08-29 16:47:06 +0800
  • 4a8846390b Add device set. 梁鸿 2024-08-18 23:47:47 +0800
  • f2ee49a639 Add dump in minist. Colin 2024-08-18 17:42:00 +0800
  • 950055c210 Refine show. Colin 2024-08-18 16:38:13 +0800
  • b860d794a6 refine code Colin 2024-08-18 00:46:36 +0800
  • d50cb798b6 Update more. Colin 2024-07-31 22:04:01 +0800
  • 50e502ae96 Add auto size support in meaning dataset. Colin 2024-04-23 19:05:06 +0800
  • 24957aa2ae Refine train. Colin 2024-04-19 19:12:38 +0800
  • 2d415d9e44 Reduce memory cost when build dataset. Colin 2024-04-19 16:57:59 +0800
  • 43883be692 Speedup meaning dataset build speed. Colin 2024-04-19 15:08:40 +0800
  • b062cc9c94 Save memory cost when meaning dataset build by np array. Colin 2024-04-19 00:50:40 +0800
  • a524d01ac3 Add multithread dataloder support. Colin 2024-04-17 23:59:09 +0800
  • 17de117bda Refine train, Colin 2024-04-17 20:22:49 +0800
  • ef08359a94 Refine meaning dataset memory cost when building. Colin 2024-04-14 23:35:55 +0800
  • c907210fc1 Update document. Colin 2024-04-14 17:41:30 +0800
  • 9926b893a4 Fix dataset. Colin 2024-04-14 01:27:58 +0800
  • 6791264987 Update meaning dataset function. Colin 2024-04-12 20:04:04 +0800
  • 43e486aa1c Add mask when validation. Colin 2024-04-10 14:43:16 +0800
  • 7434427ec9 Update meaning dataset. Colin 2024-04-10 00:34:47 +0800
  • 7560166b76 Dump dataset to single file. Colin 2024-04-07 17:32:21 +0800
  • 9d3b9a210a Speedup dataset generate. Colin 2024-04-07 17:03:35 +0800
  • 33d1e22655 Refine meaning dataset. Colin 2024-04-07 00:25:21 +0800
  • 2bc9e3b57e Refine train dataset. Colin 2024-04-03 17:09:30 +0800
  • 3c774983d4 Refine mapping print. Colin 2024-04-03 13:03:59 +0800
  • 1642a91d80 Add meaning map print. Colin 2024-04-03 11:24:00 +0800
  • 89c12380cb Delete no used files. Colin 2024-04-02 22:34:58 +0800
  • a15e55bead Add mapping output. Colin 2024-04-02 19:59:05 +0800
  • e2b48c0ab4 Add mamba. Colin 2024-04-02 15:38:49 +0800
  • 7a8815cceb Refine the base code. Colin 2024-03-29 22:10:25 +0800
  • 618d57f23c Update define. Colin 2024-03-26 18:15:55 +0800
  • 33b351ff8a Refine train.py. Colin 2024-03-26 15:01:19 +0800
  • b0ca4dc35d Update meaning dataset define. Colin 2024-03-26 11:32:02 +0800
  • e29c0b9a41 Add python pip required define. Colin 2024-03-25 20:41:41 +0800
  • d10e7a8396 Refine train.py for train. Colin 2024-03-25 19:53:11 +0800
  • 4c7fdbe817 Add GPU stress test. Colin 2024-03-25 13:20:17 +0800
  • c7391b090e Delete unused files. Colin 2024-03-20 23:05:05 +0800
  • c4f7ef2813 Update special dateset. Colin 2024-03-20 23:04:29 +0800
  • 01e5f86e94 Add inference. Colin 2024-03-20 22:27:28 +0800
  • b248d1d890 Fix model bug. Colin 2024-03-20 22:23:52 +0800
  • 72718e6b72 Add Batch dataloader support. Colin 2024-03-18 11:43:41 +0800
  • 9feaafcb7a Apply meaning data train. Colin 2024-03-15 11:16:13 +0800
  • 0ae63298b2 use custom vocab_size. Colin 2024-03-14 13:28:40 +0800
  • 05f17b1221 Refine model config and init. Colin 2024-03-14 11:40:26 +0800
  • 8330cbb036 Add meaning dataset. Colin 2024-03-13 19:41:02 +0800
  • c094afb0f9 Add tensorboard event out. Colin 2024-03-09 16:55:03 +0800
  • f1394d5974 Refine code. Colin 2024-03-08 20:46:42 +0800
  • 601c7f6510 Retest wit. Colin 2024-03-07 16:30:37 +0800
  • a70d12d04d Rename train file. Colin 2024-03-05 22:09:58 +0800
  • 9ef3e92b23 Try model train. Colin 2024-03-05 22:09:28 +0800
  • 11fc8f1d39 Refine label used. Colin 2024-03-05 22:08:37 +0800
  • fdc8c657b3 Add accurancy in loss. Colin 2024-03-05 19:30:15 +0800
  • cf726a5b9f Add loss and logger code. Colin 2024-03-05 15:54:03 +0800
  • 9e8e92ae25 Update trainer to custom data. Colin 2024-03-04 21:41:46 +0800
  • 1622bf3054 add mnbvc dataset . Colin 2024-03-03 23:35:40 +0800
  • 8120be66a6 sperate train and val dataset. Colin 2024-02-26 23:59:00 +0800
  • d1906629ab Enable wit train on cutome dataset and loss down. Colin 2024-02-26 22:42:50 +0800
  • 1ef3e419cb Add custom dataset support. Colin 2024-02-26 00:31:47 +0800
  • e5f97af291 Add wit train support. Colin 2024-02-25 20:20:32 +0800
  • fc071dce70 Remove no use tiktoken. Colin 2024-02-21 21:11:15 +0800
  • fe13f12327 Add wit. Colin 2024-02-06 14:08:45 +0800
  • 6366b52fef Add reaserch sile resault. Colin 2024-02-04 23:48:51 +0800
  • 9d5d590b09 Add dataset and wit. Colin 2024-02-04 23:48:24 +0800
  • b7c27af6c8 Add research_token to dump token relationship in attention layer0. Colin 2024-01-29 00:12:08 +0800
  • 185278f3a9 Update research_attention dump without sum. Colin 2024-01-28 17:55:08 +0800
  • 3f296ccdb2 Update research. Colin 2024-01-26 20:35:25 +0800
  • bba27e3444 Refine prepareInput. Colin 2024-01-25 18:05:08 +0800
  • 19491d1f4a Refine model of qwen. Colin 2024-01-24 21:22:03 +0800
  • 11af10e710 Refine research_attention and forward model. Colin 2024-01-23 13:13:21 +0800
  • 1811b9611a Refine research_attention. Colin 2024-01-22 20:57:27 +0800
  • 5dbac40925 Refien. Colin 2024-01-21 22:43:16 +0800
  • 17a2df2e6f Update show and q@k dump. Colin 2024-01-21 20:50:36 +0800
  • ae6ea67bbe Refine qwen/research_attention.py. Colin 2024-01-21 17:54:05 +0800
  • dab1c94bc6 Refine qwen to module fomater. Colin 2024-01-21 16:46:00 +0800
  • 9d28280cb1 Refine model of qwen and add runner. Colin 2024-01-21 12:45:56 +0800
  • 7c047f0b32 Refine model of qwen. Colin 2024-01-21 02:33:55 +0800
  • 40ae899515 Refine model of qwen. Colin 2024-01-20 20:47:26 +0800
  • 4d493014ba Refine model of qwen. Colin 2024-01-20 20:20:18 +0800