nimlgen/tinygrad
1
0
Fork 0
forked from tinygrad/tinygrad
Code Pull Requests Activity
11,453 Commits 334 Branches 15 Tags
remove_programspec
Commit Graph
1427 Commits
Author SHA1 Message Date
KirillandGitHub 0532025b04 Fix llama 13B weights loading (#700) 
* Fix llama 13B weights loading

* refactor more

* add test

* test storage offset

* fix spacing

* fix strides

* llama 13B working?

* yolo?

* better test for seeks
 2023-03-15 08:59:52 -07:00
George HotzandGitHub 15e0b56e39 compile works (#688) 
* compile works

* runtimes

* line count

* fix custom, to tg dtype

* meh, that's fine with lazy import
 2023-03-12 11:01:25 -07:00
KirillandGitHub af7745073f Add comments to SD (#686) 
* Add explanation for empty lambdas

* Fix my_unpickle if pytorch_lightning is installed

* oops
 2023-03-12 10:56:49 -07:00
geohot 6c3675c01c _mmap loads to gpu fast 2023-03-11 23:00:13 -08:00
geohot 803b0aef28 track memory for numpy/torch 2023-03-11 20:39:10 -08:00
DiogoandGitHub 784afc6c6f Eq magic function support (#683) 
* add eq magic func

* changed from eq to __eq__

* ignore type for linter

* mypy doenst like descriptions :(
 2023-03-11 10:31:46 -08:00
geohot 01f39b19dc move to shapetracker.py 2023-03-11 07:50:07 -08:00
George HotzandGitHub f3ac52aee8 Mypyc (#680) 
* building shapetracker

* default ENABLE_METHOD_CACHE

* symbolic compiles

* improve types

* tensor compiles

* oops, that's a bug

* best of both worlds

* find legit typing bugs

* pad2d can take list or tuple

* sub 200ms when compiled
 2023-03-11 07:33:30 -08:00
geohot d7cb8e3e56 multithreaded fake_torch_load_zipped 2023-03-10 19:16:27 -08:00
George HotzandGitHub b1206bcb18 third try at torch loading (#677) 
* third try at torch loading

* numpy fixed

* fix enet compile

* load_single_weight supports empty weights

* oops, CPU wasn't the default

* so many bugs
 2023-03-10 19:11:29 -08:00
geohot 4780f9a6df llama runs (slowly) in master 2023-03-10 17:36:51 -08:00
George HotzandGitHub 1826ff6b89 dtypes nice and clean (#673) 
* add dtype class

* dtypes

* buffers are lazy

* dtype is tracked by lazybuffer and GenericShape

* fix types in llvm

* llvm store

* dtype tests

* fix tests maybe

* fix flop counter

* fix CI

* CI fix and check format

* fix dtype and dtype check

* fix custom test

* fix test graph
 2023-03-10 16:56:07 -08:00
geohot d26345595d more llama stuff 2023-03-10 10:48:10 -08:00
George HotzandGitHub 1a039306d2 good changes from llama branch (#671) 
* good changes from llama

* transpose behavior changed
 2023-03-09 20:51:22 -08:00
George Hotz d8dda2af3a openpilot fixups 2023-03-06 14:14:44 -08:00
George HotzandGitHub a77d792aff Codegen gpu cleanups (#640) 
* cleanups

* fixups

* handle pre upcasted global buffers

* early is just required

* delete junk from hand coded opt

* implicit upcast_in_mid_reduce

* speedup

* fix exec w validhacks

* reorder opt

* only need to check the output for that

* return total runtime from kernels if debugging
 2023-03-04 15:31:51 -08:00
Patrick GenevaandGitHub 117111825c Fix windows file permission error (#634) 2023-03-04 09:23:55 -08:00
geohot 528cb3b3b9 fix ast test 2023-03-04 07:49:25 -08:00
geohot 893f136fe0 lines from helpers 2023-03-03 23:07:46 -08:00
George HotzandGitHub c53efb3635 optimize for CL (#633) 
* required opt

* simplify

* works

* shift_to_last

* required is fine

* print shape in colored

* better shape

* args was wrong

* debugs

* fix empty shape

* colored shape printer
 2023-03-03 22:00:09 -08:00
DiogoandGitHub 52204a7b88 adding comparison operators (#616) 
* Less, LessOrEqual, Greater, GreaterOrEqual, Equal

* lint fix

* using built in functions

* overriding __eq__ breaks things

* backwards pass for less - foward only tests

* one other spot

* removing backwards for comparison ops to match pytorch

* raise runtime error

* more tests for comparison ops

* fixed the lineup

* added number upcast tests
 2023-03-02 08:10:44 -08:00
geohot d062cc82b8 put restrict back 2023-03-01 21:34:45 -08:00
George HotzandGitHub bfcec234a2 Refactor ASTs (#622) 
* ugh worst branch name

* compiler refactor continues

* scc -> cloc

* buf -> _buf

* finish _buf, and program -> runtime

* gpu is still working, clang isn't

* clang in new style

* ops_metal

* something broke it

* improve metal

* clean up tons of cl crap

* hack fix sync

* cleaner gpu

* gpu metal clang

* cleanups

* minor refactor

* GPUCodegen

* fix up LLVM

* blind CUDA refactor

* codegen / runtime

* keep ops naming

* linter passes

* woah, llvm was allocing 4x what it needed to

* bugfixes

* fix openpilot compiler

* fix compile_efficientnet

* method cache should fix tests

* deal with duped functions
 2023-03-01 18:57:29 -08:00
George Hotz 7e6edfbc64 unbreak onnx conv padding 2023-02-28 13:55:03 -08:00
George Hotz 7d556ca7e0 avg/max pool work in N-D 2023-02-28 13:38:27 -08:00
geohot d584bae5c0 fine, openpilot can have 197 kernels 2023-02-27 11:48:36 -08:00
geohot 7b999add1d all onnx model tests pass 2023-02-27 11:22:45 -08:00
geohot 652d48ccec onnx : openpilot expand issue was fixed yesterday. remove hack 2023-02-27 11:04:42 -08:00
geohot 9d6b63f043 add ConstantOfShape 2023-02-27 10:57:50 -08:00
geohot 082134952b CastLike works with one type hack 2023-02-27 10:51:26 -08:00
Jacky LeeandGitHub 1ffe8d68d5 Add more onnx ops (#615) 
* Add Celu

* Add thresholded relu

* Add softsign
 2023-02-27 10:43:41 -08:00
geohot 643e8b0388 fix tests, test bn evaluate too 2023-02-27 10:39:47 -08:00
DiogoandGitHub 07e643431c added onnx group norm (#614) 2023-02-27 08:11:01 -08:00
DiogoandGitHub e68fa18c9b layer norm support in onnx (#607) 
* layer norm support

* switched to 1e-05
 2023-02-26 22:04:02 -08:00
geohot 3a2a500e90 prevent race condition, external yolo test for now 2023-02-26 17:08:24 -08:00
SohaibandGitHub 71ae6e5605 fix: avgpool without counting padding (#605) 2023-02-26 07:13:00 -08:00
George HotzandGitHub a8de233e12 only div, no reciprocal (#601) 
* only div, no reciprocal

* remove reciprocal

* fix pad shuffling
 2023-02-25 09:35:03 -08:00
SohaibGitHubSohaib Errabii
 		d581a99d90 onnx: lrn (#602) 
Co-authored-by: Sohaib Errabii <[email protected]>
 2023-02-25 09:24:53 -08:00
voidzandGitHub 94bec40110 moved extras/jit.py -> tinygrad/jit.py (#599) 
* moved extras/jit.py to tinygrad/jit.py

* fixed indent

* removed tinygrad.helpers.DEBUG from jit.py
 2023-02-25 08:32:33 -08:00
George HotzandGitHub 2c5e13a513 Reluless (#600) 
* replace relu for maximum

* fix for other backend

* clean up RELU and GT0

* tests for maximum

* had to clean that up

* why reverse a maximum?
 2023-02-25 01:21:16 -08:00
George Hotz 176ad29974 retain support for old onnx 2023-02-24 22:29:54 -08:00
geohot da5643d024 rest of tests shouid be made to pass 2023-02-24 12:52:23 -08:00
geohot 85452fbaf3 onnx 58/109/208 2023-02-24 12:19:05 -08:00
geohot e8a153e4e9 onnx : add a whole bunch of ops 2023-02-24 12:00:03 -08:00
geohot f2486a7248 more onnx ops 2023-02-24 10:55:58 -08:00
geohot 4d0a3dd653 openpilot expand is bugged 2023-02-24 10:25:59 -08:00
geohot 2e56a4793e rename log_softmax, support dim, fix onnx Softmax 2023-02-24 10:11:24 -08:00
geohot 5cdfeffe2c fix shape test 2023-02-24 09:36:32 -08:00
geohot 3becefa218 fix onnx tests 2023-02-24 09:27:18 -08:00
geohot e263c0c628 onnx : another model test is passing 2023-02-24 09:22:58 -08:00