* use at least float32 for optim.lr
when doing mixed precision training (float32 weight, default_float=half), still use float32 to store lr.
it would have been upcasted later in actual weight update, but would have lost precision.
this improved resnet convergence significantly
* undo type annotation
* no ret value and just force contiguous
* ok revert contiguous stuff
* actually do force it contiguous
* revert again lol
* add simple regression test
* add assert for MLB
* guess we're contiguous everything from now on
* lol ugly af empty return...
* don't change order cuz i don't get disk
* Preprocessing script
* short seq prob
* comments + env vars
* Add preprocessing reference. Add test
* lint fix + add eval test support
* whitespaces
* point to commit
* comment
* rename
* better comments
* pm4 kernel launch works
* disable USE_THREAD_DIMENSIONS
* add kernel code
* work on real pm4
* pm4 signal
* same
* gate pm4
* hcq tests pass
* ops passes
* pm4 is closer
* pm4 debug (#4165)
* start debug tests passing
* prg
* smth
* hdp flush
* cleaner 1
* do not need this
* logs not need
* small things
* linter
* remove AQL
* test hcq
* fix tests
* it's subtracting, it shouldn't be -1
* pm4 changes (#4251)
* not need this anymore
* sdma signal with non atomic
---------
Co-authored-by: nimlgen <138685161+nimlgen@users.noreply.github.com>
* wmma: widen TC usage in search by using PADTO on TC axes when possible
* test: start tests for the new padding TC behavior
* search: upgrade padded TC search to TC_OPT >= 2
* test: add behavior and correctness test for padded TC
added optional argument to apply_tensor_core to set TC_OPT level
* linearizer: add tests for the PADTO behvaior and docs
* WIP: clean up update stats
* line savings now
* fix graphs
* fix tests
* tighter prints
* remove extra jit=false
* debug=2 means wait
* that won't update stats
* still wait
* fuzz schedule context vars
* fuzz unique toposorts
* merge ground truth with the rest
* Revert "merge ground truth with the rest"
This reverts commit 1f3463bb57.
* readability>
* can override
* lol does this work
* some more changes
* a tiny note
* rename a variable
* add test for data const and add TODO comment
* make type correct
make type correct
* add DICE loss and metrics
* update dice to include reference implementation's link
* remove unused imports
* remove unnecessary test file and update pred + label for metrics and losses test
* add tests to CI + add exclusion of mlperf_unet3d
---------
Co-authored-by: chenyu <chenyu@fastmail.com>
* in forced_realize, unchase last op if it is upcast
* start on test
* flesh out test
* more test
* comment
* comment out parallel reduce test
* reorder
* unused
* resnet individual layer benchmarks!
* small
* 1 and 2
* mem_used
* no ci
* better conv print
* defaults
* prints
* adjust
* adjust
* adjust
* benchmark only one layer example
* tensor.training, zero_grad, sum instead of mean, last mem, last kernel count
* default jitcnt=1
* scale flops/kernels with jitcnt
* add note about jitcnt memory
* touchup
* write llm.c and add a few new methods to tensor
* training works
* add jit
* tests for new functions
* test tolist
* simple fix for onnx test failures (#4186)
* write llm.c and add a few new methods to tensor
* training works
* add jit
* tests for new functions
* bump line count to 7500
* simplest fix
* safenumpy tolist for now
---------
Co-authored-by: George Hotz <geohot@gmail.com>
Co-authored-by: George Hotz <72895+geohot@users.noreply.github.com>
---------
Co-authored-by: geohotstan <135171913+geohotstan@users.noreply.github.com>
* rewrite the jit in the context of new schedule
* mypy better
* fix placeholder
* tests
* all functionality should work
* fix tests
* no CacheCollector
