github/tinygrad
1
1
Fork 0
mirror of https://github.com/tinygrad/tinygrad.git synced 2026-02-11 23:25:04 -05:00
Code Issues Packages Projects Releases Wiki Activity
3,534 Commits 220 Branches 14 Tags
ea74856d99f3b30e17e4bec82200a3536052088f
Commit Graph

158 Commits

Author SHA1 Message Date
chenyu
97275101e9 fix safetensor load uint32 and uint64 (#3315) 
the correct keys are U32 and U64.
 2024-02-04 10:46:27 -05:00
Yoshinori Sano
edb74897b2 support safe load bf16 (#3310) 
* support safe load bf16

* fix lint error E501

* add test for loading safetensors

* key should be BOOL

* fix lint
 2024-02-04 10:08:39 -05:00
chenyu
d459956966 move TestGetContraction to test_helpers (#3313) 
also cleaned long lines in test_shapetracker and enabled the line length check
 2024-02-04 06:05:01 -05:00
Felix Wu
021eea3a52 fix UnboundLocalError when running Compiler with DISABLE_COMPILER_CACHE (#3296) 2024-02-01 21:12:33 -05:00
David Hou
3378625773 name upcast variables (#3200) 
* name upcast variables

* typing

* unused
 2024-01-22 11:37:28 -05:00
George Hotz
d2aab65958 remove unused expr node (#3170) 
* remove unused expr node

* still works

* simple expr_idxs

* fixup typing
 2024-01-18 14:18:43 -08:00
George Hotz
f0c178b7e9 move get_contraction to helpers (#3162) 
* move get_contraction to helpers

* move simplify

* lines

* to_movement_ops is not generic
 2024-01-17 19:13:11 -08:00
George Hotz
a464909d79 fast resnet eval (#3135) 
* fast resnet eval

* fix HIP multidevice graph

* neater expression for devices

* lines

* add decorator test
 2024-01-15 14:15:18 -08:00
Paul Gustafson
6bb65cd02e fix off-by-one error in st_equal (#3131) 
* fix off by one error

* whitespace
 2024-01-15 11:32:13 -08:00
chenyu
c658aa4fbf minor cleanup of test_disk_tensor (#3112) 2024-01-13 20:54:58 -05:00
chenyu
a300fea2a4 failed test case due to cast resets shapetracker (#3109) 
cast implicitly resets shapetracker and makes it contiguous (for disk tensor), which fails for Interpreted backend if inputs contain non-contiguous st.
 2024-01-13 12:46:51 -05:00
chenyu
f018a55ea1 update NumNode.__hash__ to be hash(self.b) (#3105) 
with this, `a:=NumNode(x) == b` implies `hash(a) == hash(b)`
 2024-01-12 19:46:21 -05:00
chenyu
dab8214103 unit tests for Device.canonicalize (#3055) 2024-01-09 12:47:20 -05:00
George Hotz
655c6f61d3 St real size (#3046) 
* track the size in the lazybuffer

* shapetracker real size

* lint
 2024-01-08 14:44:53 -08:00
George Hotz
c003be7309 Revert "track size in shapetracker" (#3043) 
* Revert "track size in shapetracker (#3026)"

This reverts commit a8ba1ac08f.

* st.size
 2024-01-08 13:13:39 -08:00
George Hotz
a8ba1ac08f track size in shapetracker (#3026) 
* track size in shapetracker

* shapetracker adapter

* size is an int

* create Buffer with st.size

* only compare the views for the jit

* fix webgpu
 2024-01-05 20:15:53 -08:00
George Hotz
60abc62a3f fast hip read (#3014) 
* fast hip read

* hip read faster

* fix tests

* to_mv

* simplify

* bump to 6k lines
 2024-01-05 10:33:13 -08:00
George Hotz
9699c8c90b don't alloc for InterpretedASTRunner (#2999) 2024-01-03 17:05:53 -08:00
chenyu
74cc6fd3c2 remove AndNode.__floordiv__ special case (#2996) 
* remove AndNode.__floordiv__

AndNode produces a Node that min/max is bounded by [0, 1] so `//` on top of that is almost always 0.
we don't really use that either

* keep the test
 2024-01-03 17:44:55 -05:00
chenyu
ff5399f053 move one last dtype test from test_helpers to test_dtype (#2975) 2024-01-02 12:37:56 -05:00
George Hotz
a280cfe169 move dtypes to dtype.py (#2964) 
* move dtypes to dtype.py

* fix urllib
 2024-01-01 14:58:48 -08:00
George Hotz
c81ce9643d move globalcounters to ops (#2960) 
* move globalcounters to ops

* missed a few

* sick of that failing
 2024-01-01 14:21:02 -08:00
chenyu
8291986959 Variable.sum -> Node.sum, Variable.ands -> Node.ands (#2961) 2024-01-01 16:21:28 -05:00
chenyu
3d720b5761 move expand_idx, iter_idxs and expand_node from symbolic to linearizer (#2959) 2024-01-01 14:41:21 -05:00
George Hotz
5cac6338a4 apply the multitensor optimizations in lazy.py (#2901) 
* apply the multitensor optimizations in lazy.py

* less lines

* hack for webgpu

* save a line
 2023-12-21 13:55:49 -08:00
George Hotz
1765849937 new lazy, benchmark (#2878) 
* lazy rewrite, try 2

* min fix tests

* pass contig test

* put broken pads back

* move that to realize

* no contig child fixes array packing

* so wrong

* now that's correct

* base children

* fix bind issues

* disable to_image_idx

* fix tests

* that failure shouldn't break other tests

* more fixes

* fix torch

* skip failing tests in CI

* 1e-7

* half is broken

* 1e-6 margin of error
 2023-12-20 14:33:21 -08:00
Peter Cawley
dae8976889 Fix reshape merging with masks (#2877) 2023-12-20 14:00:58 -08:00
George Hotz
ca59054463 fix shapetracker math (#2861) 
* proper test

* all st math good now

* fix real_strides bug
 2023-12-19 22:17:34 -08:00
chenyu
5a739e8c20 update one skipped pad_reshape test that was fine (#2860) 
* update one skipped pad_reshape test that was fine

had a typo

* this one passed
 2023-12-19 23:25:52 -05:00
chenyu
ad233d557f disable reshape merging with masks (#2858) 
fuzzer found a bug, and it's not complete
 2023-12-19 19:06:16 -05:00
Oleg Rybalko
42a038c83f More readable torch_load ext check (#2853) 
* more readable extension check

* enable tarfile test

* detach tensor if requires grad in torch
 2023-12-19 14:53:15 -05:00
George Hotz
b2192b5400 minor improvements (#2845) 2023-12-18 22:09:08 -08:00
George Hotz
d086325b1b hotfix: failing tests 2023-12-18 21:12:42 -08:00
George Hotz
b6d71b131e hotfix: push broken tests 2023-12-18 21:08:42 -08:00
George Hotz
80f53245e8 shapetracker add and invert (#2828) 
* invert (broken)

* decent invert

* shapetracker invert works

* plus is meh, invert is good

* support invert mask

* a few more invert tests

* shapetracker math invert test
 2023-12-18 16:03:27 -08:00
chenyu
b4fa189c8c Revert "Revert "Make Tensor creation allow multi-dim list of int and bool (#2793)" (#2810)" (#2813) 
This reverts commit 71a60762ed.
 2023-12-17 11:48:27 -05:00
chenyu
71a60762ed Revert "Make Tensor creation allow multi-dim list of int and bool (#2793)" (#2810) 
This reverts commit 798bf813b1.
 2023-12-17 02:03:52 -05:00
geohotstan
798bf813b1 Make Tensor creation allow multi-dim list of int and bool (#2793) 
* the universe is flat as a 2D tensor

* try this

* TESTS

* less lines in test

* don't change all_int since other places use it

* add tests and del noqa by making non-aesthetic spacing LOOOOOL

* some reordering

* fixed empty list and add tests

* more tests

* add list bool tensors

* clearer with least lines added

* added bool

* oops

* more tests

* improved tests

* oops
 2023-12-17 01:58:10 -05:00
George Hotz
877c78b4ce lazy tests (#2796) 
* tests

* mini sd is very mini
 2023-12-16 08:24:21 -08:00
chenyu
5235cdee3d remove _arg_int32 internal type (#2767) 
in DEFINE_GLOBAL, PtrDtype(int32) is buffer and int32 is int
 2023-12-14 14:17:14 -05:00
George Hotz
7e5b3e53fe changes to prep for new lazy (#2748) 
* changes to prep for new lazy

* put those back
 2023-12-13 10:28:22 -08:00
Umut Zengin
8ad7cfeeb1 More simplification in to_image_idx and symbolic (#2679) 
* less valid

* add test

---------

Co-authored-by: chenyu <chenyu@fastmail.com>
 2023-12-13 12:30:44 -05:00
George Hotz
6d6eb9302d ruff checks the max line length is 150 (#2734) 
* ruff checks the max line length is 150

* fix tensor.py

* a lot more

* done
 2023-12-12 17:34:47 -08:00
Guy Leroy
ee9e1d3662 Extend available types for safe_save (#2720) 
* Extend available types to save with

* Linter fix
 2023-12-11 14:50:35 -08:00
George Hotz
0fd44259cd bf16 fix + cleanups from mixtral (#2698) 
* bf16 fix + cleanups from mixtral

* generic bf16 cast
 2023-12-10 16:31:52 -08:00
qazal
73b067f5ce Bitcast p2 bfloat16 tests + clang fix (#2635) 
* add bf16 test support

this model takes me almost a minute to download though:

https://huggingface.co/TinyPixel/Llama-2-7B-bf16-sharded/resolve/main/pytorch_model-00001-of-00014.bin?download=true: 100%|█████████████████████████████| 981M/981M [00:40<00:00, 24.2MB/s]

* ensure we first load if it is bitcast to avoid taking the address of an rvalue

* tiny bf16 in the cloud

skip GPU

* should skip torch

lint

* Revert "ensure we first load if it is bitcast to avoid taking the address of an rvalue"

This reverts commit b86a28ab84.

* break the kernel

* skip LLVM and GPU in CI

* skip CUDA
 2023-12-08 10:30:10 -08:00
chenyu
b931a20882 minor shapetracker cleanup (#2652) 2023-12-06 11:43:52 -05:00
Amrit Sahu
71d989b476 adding test to cover #2644 failure (#2645) 2023-12-06 11:00:30 -05:00
George Hotz
232ed2af3f more test cleanups (#2631) 
* more test cleanups

* move test example back
 2023-12-05 16:17:57 -08:00
George Hotz
35b5e95097 parallel beam search (#2610) 
* better print

* fix beam search with vars

* cleanups

* parallel is not default

* restore that

* bugfix

* cleanups

* bugfix
 2023-12-05 10:09:45 -08:00