1
0
mirror of https://github.com/karpathy/minGPT synced 2024-09-20 19:05:21 +02:00

Commit Graph

  • 20ad0e8920 Restructure data preprocessing umertens 2024-09-15 22:13:02 +0200
  • de024f3dbb don't use the lr scheduler for now Felix Wick 2024-09-13 13:48:34 +0200
  • f76018592e more conservative scheduling Felix Wick 2024-09-13 12:28:54 +0200
  • 90dd88f202 add learning rate scheduler to suppress jumps at training end Felix Wick 2024-09-13 12:12:49 +0200
  • ce691efaa9 remove accidential commit of some testing Felix Wick 2024-09-13 11:09:43 +0200
  • 9ad506d53f drop the causal mask to enable full permutation invariance of columns Felix Wick 2024-09-12 21:22:16 +0200
  • b189b465d3 commented config for experiments Felix Wick 2024-09-04 08:50:44 +0200
  • b02a3f14d7 stupid padding bug Felix Wick 2024-09-01 23:04:25 +0200
  • a3309058a8 small fix Felix Wick 2024-09-01 07:57:52 +0200
  • a5cf057780 some project updates Felix Wick 2024-09-01 00:19:45 +0200
  • 83ee5f2675 updated simulated demand project Felix Wick 2024-08-31 21:12:34 +0200
  • 59a7b8d726 significant speedup in value embeddings Felix Wick 2024-08-31 12:51:36 +0200
  • 63201f2430 added simulated demand data set to multi-task Felix Wick 2024-08-30 21:55:44 +0200
  • b2f4fa7abb use ewma feature in demand forecasting Felix Wick 2024-08-30 20:15:31 +0200
  • c069969288 added some dependencies Felix Wick 2024-08-30 18:23:01 +0200
  • bb20c247f5 updates for demand forecasting and multi task models Felix Wick 2024-08-30 18:05:58 +0200
  • e72c88f88a updated bicycles count project Felix Wick 2024-08-30 15:39:34 +0200
  • d549569ad7 added dependencies Felix Wick 2024-08-30 12:21:22 +0200
  • 44ad841097 python version Felix Wick 2024-08-30 11:42:03 +0200
  • fb731a4d89 added pyproject.toml Felix Wick 2024-08-30 11:20:22 +0200
  • 48252e4591 updated store sales project Felix Wick 2024-08-29 20:46:30 +0200
  • d3c1a2b1fa scale from train only Felix Wick 2024-08-29 13:07:49 +0200
  • 3fd4ee7a6c fix Felix Wick 2024-08-29 08:45:32 +0200
  • df215a1252 test fix Felix Wick 2024-08-28 23:45:33 +0200
  • 7369ca5d98 use all columns for house prices Felix Wick 2024-08-28 18:32:25 +0200
  • a08142717c back to shuffling :), optional train loss logging Felix Wick 2024-08-28 17:20:10 +0200
  • 4edb2639e1 more fixes Felix Wick 2024-08-28 15:52:13 +0200
  • 2760ce896a fix from last commit Felix Wick 2024-08-28 11:40:25 +0200
  • 8b56b9d94e do not shuffle, use remaining batch samples Felix Wick 2024-08-28 09:42:21 +0200
  • 668286042f scaling, loss logging, etc Felix Wick 2024-08-26 19:37:28 +0200
  • 6d6bd23d10 different options for 0 treatment for numerical columns Felix Wick 2024-08-25 11:05:36 +0200
  • f6ddce4549 only generate row value embeddings for unique categories Felix Wick 2024-08-24 17:54:03 +0200
  • 458fed4c26 get it to run on aws ec2 Ubuntu 2024-08-15 04:08:22 +0000
  • 26918ac473 just to get it running Felix Wick 2024-08-09 20:50:12 +0200
  • d54d3cf851 added NY bicycles count project Felix Wick 2024-08-09 19:53:01 +0200
  • 45bc253bfe add mode to train individually Felix Wick 2024-08-09 13:18:43 +0200
  • b027abed81 add target embedding Ulf 2024-08-08 20:15:07 +0200
  • 5254e8c3b3 include target description embedding Felix Wick 2024-08-09 08:22:32 +0200
  • cdd860a72a ewma in multi-task Felix Wick 2024-08-05 22:35:44 +0200
  • 4e53e6aefe pretrained fix Felix Wick 2024-08-05 21:47:30 +0200
  • a25d54b30a include more numerical features Felix Wick 2024-08-05 17:34:32 +0200
  • c7c06ddf88 first version of column embeddings model Felix Wick 2024-08-04 12:29:46 +0200
  • 00cc66d687 added classification mode Felix Wick 2024-07-14 22:43:58 +0200
  • b01d55047e add data enriching and test Felix Wick 2024-07-10 09:54:37 +0200
  • 9f0a3d37cd add multi-task model on store sales and house prices Felix Wick 2024-07-08 23:52:25 +0200
  • 16ec970d8e target leakage :) Felix Wick 2024-07-08 21:15:50 +0200
  • aa26db5d77 added house prices project Felix Wick 2024-07-08 18:13:52 +0200
  • f23592c1b7 enable initialization with language-pretrained gpt2 Felix Wick 2024-07-08 13:50:19 +0200
  • 9ffd9f97ce initial regression model Felix Wick 2024-07-07 13:35:07 +0200
  • 9d83f810f2
    Merge b8979686c3 into 37baab71b9 Victor Ziliang Peng 2024-04-28 09:32:56 -0700
  • b8979686c3 hotz Ziliang Peng 2024-04-28 09:32:28 -0700
  • 0db60b78cc awkward typo Ziliang Peng 2024-04-28 00:37:00 -0700
  • 34f36c89f8
    Merge pull request #4 from ziliangpeng/v--README Victor Ziliang Peng 2024-04-28 00:35:37 -0700
  • 548599923e organize top level files and metadata Ziliang Peng 2024-04-28 00:33:28 -0700
  • 8d17e92199 update README Ziliang Peng 2024-04-28 00:27:45 -0700
  • cf4549ed6f
    Merge pull request #3 from ziliangpeng/tinygradify Victor Ziliang Peng 2024-04-28 00:11:45 -0700
  • 146dfb5709 nit Ziliang Peng 2024-04-28 00:07:15 -0700
  • 45c1a5113f move comment Ziliang Peng 2024-04-28 00:04:53 -0700
  • d81e154dcb more cleanup Ziliang Peng 2024-04-27 23:58:04 -0700
  • 8405765843 bug fix Ziliang Peng 2024-04-27 22:26:45 -0700
  • c447ee0e9f fix the no_grad bias issue Ziliang Peng 2024-04-27 22:10:45 -0700
  • db3301a1e1 tidy import Ziliang Peng 2024-04-27 21:55:24 -0700
  • 2a22fff05e super small cleanup Ziliang Peng 2024-04-27 18:38:41 -0700
  • 9568f28e3e cancel clip_grad_norm_ Ziliang Peng 2024-04-27 18:31:35 -0700