commaai
/
tinygrad
mirror of https://github.com/commaai/tinygrad.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity
693 Commits 10 Branches 0 Tags 52 MiB
Commit Graph

693 Commits

Author SHA1 Message Date
George Hotz 4320c45c4b oops, gpu not available 2021-11-17 12:54:33 -08:00
George Hotz e2a8961a18 less lines, fix bug 2021-11-17 12:52:17 -08:00
George Hotz e8ca3ad053 add Buffer abstraction for each device 2021-11-17 12:46:07 -08:00
George Hotz c162e748f5 fix float64 warning on training 2021-10-30 20:07:31 -07:00
George Hotz b0f14b4af8 move datasets into datasets 2021-10-30 19:55:50 -07:00
George Hotz a95ef16c8c sub 1000 lines 2021-10-30 19:48:24 -07:00
George Hotz 844540a5ed yolo in readme 2021-10-30 19:47:34 -07:00
George Hotz ba28761894 move yolo into examples/yolo 2021-10-30 19:46:00 -07:00
George Hotz 121d5a17ee use tinynn for Conv2d 2021-10-30 19:40:44 -07:00
George Hotz 6bee5bdb7d add torch tests 2021-10-30 18:58:45 -07:00
George Hotz 75260c6e49 unify more of cpu and torch 2021-10-30 18:57:24 -07:00
George Hotz b1dae764d3 more similar between torch and numpy 2021-10-30 18:46:59 -07:00
George Hotz d6cb373c79 import the torch ops that are the same 2021-10-30 18:22:44 -07:00
George Hotz 51a49de299 tests pass for torch ops 2021-10-30 18:00:12 -07:00
George Hotz 641b1dbb40 remove ane, start supporting ops_torch 2021-10-30 17:47:00 -07:00
George Hotz 7d12482d80 refactor efficientnet loading 2021-10-30 17:02:17 -07:00
George Hotz 7472a7ebe2 not forcing 3.9 for a stupid type 2021-10-30 16:52:40 -07:00
George Hotz 114f6ca3fd more readme cleanup 2021-10-30 16:51:25 -07:00
George Hotz fc6597a6d9 only resnet18, it's too slow otherwise 2021-10-30 16:48:39 -07:00
George Hotz c05867dcbb header 2021-10-30 16:41:05 -07:00
George Hotz e10dc90524 tpu docs 2021-10-30 16:38:19 -07:00
George Hotz effd0dc833 update readme 2021-10-30 16:34:00 -07:00
George Hotz f193eeed25 bump all to python 3.9 2021-10-30 16:15:41 -07:00
George Hotz 63f50cff45 move back again 2021-10-30 16:13:29 -07:00
Evan Mays 285621aeda
Cherry backprop for conv2d (#281) 
* quick math: 0 + x = x.

* gradient w.r.t. x using cherry for conv

* gradient w.r.t. w for conv on cherry but doing vector dot products

* small optimization

* [cherry] optimize conv backpass for large channel count

* get rid of numpy einsum
 2021-10-30 16:12:19 -07:00
George Hotz 3d646272d6 move back 2021-10-30 16:12:12 -07:00
George Hotz ac8afd24fa refactor accel 2021-10-30 16:10:59 -07:00
George Hotz 082089d1c7 install requires pillow 2021-10-30 16:00:33 -07:00
Sebastian Kreft 8113eec4cf
feat: add efficientnet test (#285) 
Simple test using the Chicken example from https://upload.wikimedia.org/wikipedia/commons/4/41/Chicken.jpg and the image preprocessing from example/efficientnet.py

Note that EfficientNet loads the weights from the internet so running the tests may be slow the first time. We could speed up the tests by caching the /tmp folder.

Fixes #234
 2021-10-30 15:53:51 -07:00
George Hotz d8e5758c89 needs pillow 2021-10-27 10:14:02 -07:00
Sebastian Kreft 3358770182
chore(efficientnet): don't use eval when loading weights (#286) 
Because the weights are being loaded from a third party internet address, it's unsafe to use eval. Also with the change I think the code became a little bit more clear as now it's clearer which keys are being transformed.

Co-authored-by: Seba Kreft <sebastian.kreft@houm.com>
 2021-10-22 15:10:04 -07:00
Wenjun.Zhang 5402fd7f05
Fix: (#279) 
1. python create model not match swift model in ane build folder.
2. fix ane lib path and add a dummy backward function for use_ane.py
can run in examples folder.
 2021-08-06 07:58:13 -07:00
Zoli Veres 9f732e697a
Fix loading `relu.hwx` from the wrong path caused by moving the `ane` folder into the `accel` folder (#276) 
This is the commit that moved the ane folder 0eda3eb421
 2021-06-29 07:20:54 -07:00
Guglielmo Camporese 2b7589db64
Added ResNet-{18, 34, 50, 101, 152} (#271) 
* added resnets

* fix minor

* fix minor

* resnet in models

* added resnet test

* added resnet train test

* added linear, conv2d nn tests

* fix minor in extra/training

* resnet in models

* fix minor

* fix tolerance for linear in nn test

* fix eval, this causes cpu and gpu UT failing

* revert transformer test

* fix minor for CPU test

* improved model get_params for sequential layer

* fix minor for params counting

* commented broken ops tests

* improved train for resnet
 2021-06-21 09:37:24 -07:00
George Hotz 89798d2f43 some flags 2021-06-19 11:46:31 -07:00
George Hotz d81eae8288 debug cherry crash 2021-06-19 11:41:20 -07:00
George Hotz 0eda3eb421 move ane and fpga to accel folder 2021-06-19 11:26:20 -07:00
George Hotz d3f169b267 move good models to models, add a training step test 2021-06-19 11:24:15 -07:00
George Hotz b48d4bad2e clean up print spam 2021-06-19 10:31:04 -07:00
Jacky Lee 3a91d5434f
Add dropout test (#265) 
* Add dropout test

* Remove condition where training is false

* Skip dropout test when on GPU

* Revert changes to tensor.py and fix test case

* Revert change on whitespace

* Convert Tensor to cpu for testing

* Fix whitespace in tensor.py
 2021-06-19 08:49:13 -07:00
George Hotz ca0a38f2d5 more tpu notes 2021-06-18 13:28:06 -07:00
George Hotz 027535d0b5 microcoded matmul 2021-06-17 21:03:08 -07:00
George Hotz d1dd9b46f6 if i was giving systolic arrays thumbs like siskyl and ebert (RIP) i would give them 0 2021-06-17 19:48:58 -07:00
George Hotz 026e2ae6a7 three registers and a zero command 2021-06-17 17:09:18 -07:00
George Hotz 2e71ae33f6 max op works 2021-06-17 17:01:21 -07:00
George Hotz 9e12c1bbba cherry binop 2021-06-17 16:50:40 -07:00
George Hotz fcdabea880 training mnist with cherry ops 2021-06-17 16:45:35 -07:00
George Hotz 2affd226b3 speed up sum 2021-06-17 16:38:34 -07:00
George Hotz e8eb7d1b7e max op 2021-06-17 16:20:56 -07:00
George Hotz c1d469d440 sum op 2021-06-17 16:19:35 -07:00