Francis Lata
6fdcaa178b
add checkpointing and training resume capabilities
2025-01-22 14:20:17 -08:00
Francis Lata
95cdbbf237
add jit to the training loop
2025-01-22 12:31:29 -08:00
Francis Lata
efe64ebeaf
enable lr scheduler and fix benchmark timing
2025-01-22 09:56:38 -08:00
Francis Lata
d1bc4aef94
do not realize when sharding model weights
2025-01-21 13:45:35 -08:00
Francis Lata
7f331d8836
fix dataloader script
2025-01-21 13:43:59 -08:00
Francis Lata
1bf5ee286b
Revert "debug dataset test failuire"
...
This reverts commit 1b2f9d7f50
2025-01-21 13:30:12 -08:00
Francis Lata
1b2f9d7f50
debug dataset test failuire
2025-01-21 13:23:50 -08:00
Francis Lata
7815d3ddff
Merge branch 'master' into retinanet_mlperf
2025-01-21 13:06:04 -08:00
chenyu
1e283c33d3
remove realize in bert model init [pr] ( #8707 )
2025-01-21 14:11:03 -05:00
Francis Lata
bf36006ff0
set seed
2025-01-20 22:54:54 -08:00
Francis Lata
5d9a604963
add support for BENCHMARK
2025-01-20 22:47:23 -08:00
Francis Lata
be2e97260d
fix dtype for anchor inside dataloader and fix horizontal flip transformation
2025-01-20 22:45:25 -08:00
Francis Lata
cd511384e2
move anchors as part of dataloader
2025-01-20 13:13:16 -08:00
geohotstan
dd82b4c913
make onnx runner a class ( #8647 )
...
* this
* clean up
* more clean ups and improve debug msg
* more correct training toggler
* remove manual training toggling
* change some variable names
* actually just add the training toggle for LIMIT envvar too
* more refinement
* __call__ and OnnxRunner
* fix half pylint, other half is importing from onnx while this file is onnx.py, figure out later
* ahhhh found another mistake
* remove limit from __call__
---------
Co-authored-by: chenyu <chenyu@fastmail.com >
2025-01-20 10:11:05 -08:00
Francis Lata
575c748d94
fix wandb resuming feature
2025-01-20 07:22:16 -08:00
Francis Lata
a90a6e624d
add wandb
2025-01-20 07:07:51 -08:00
Francis Lata
9402872d90
Merge branch 'master' into retinanet_mlperf
2025-01-20 06:51:12 -08:00
chenyu
c49e0fca60
GlobalCounters.reset() in sdxl step [pr] ( #8664 )
2025-01-17 21:10:28 -05:00
Francis Lata
4bc762120e
Merge branch 'master' into retinanet_mlperf
2025-01-15 02:45:21 -08:00
chenyu
930728c069
bert BS 72->66 [pr] ( #8621 )
...
72 does not fit now
2025-01-14 18:41:41 -05:00
Francis Lata
b957b023fc
Merge branch 'master' into retinanet_mlperf
2025-01-13 09:33:38 -08:00
geohotstan
4abe631b56
fix onnx mobilenetv2-7-quantized.onnx ( #8574 )
...
* is 67% considered fixed?
* move test up
* share function
* add qgemm too
* make sure qgemm comes out as int
* actually that note is not right
* remove qgemm (I did it wrong) and add it later lol.
2025-01-13 09:25:06 -08:00
Francis Lata
aeecce1d18
Merge branch 'master' into retinanet_mlperf
2025-01-13 07:02:19 -08:00
chenyu
994944920b
simpler batch_load_train_bert [pr] ( #8582 )
...
don't think that buffer is really beneficial. 5% faster data_time and 1ms faster per step.
https://wandb.ai/chenyuxyz/MLPerf-BERT/runs/69c9lx8y/overview
2025-01-12 20:25:05 -05:00
George Hotz
4ac4c1415a
free intermediate buffers in the jit [pr] ( #8581 )
...
* free intermediate buffers in the jit [pr]
* intermediates_freed
* deallocate if not allocated
* self._first_run is simpler
2025-01-12 15:41:41 -08:00
chenyu
def90b22f6
EVAL_BS=36 for bert [pr] ( #8576 )
...
3X faster eval compared to BS=6.
green https://wandb.ai/chenyuxyz/MLPerf-BERT/runs/ka5p5sm9/overview
red https://wandb.ai/chenyuxyz/MLPerf-BERT/runs/a7maxsxd/overview
2025-01-12 09:43:56 -05:00
Francis Lata
f7537e4db2
Merge branch 'master' into retinanet_mlperf
2025-01-10 05:46:04 -08:00
George Hotz
9833fe83d8
more work on onnx imagenet [pr] ( #8552 )
...
* more work on onnx imagenet [pr]
* working quantization
* static quant
* benchmark onnx 0 dim
2025-01-09 20:28:18 -08:00
George Hotz
e172b759f0
more working ( #8550 )
2025-01-09 18:40:08 -08:00
chenyu
b6be407bc6
fix handcode_opt bert [pr] ( #8509 )
...
* fix handcode_opt bert [pr]
* too slow
2025-01-05 19:14:12 -05:00
George Hotz
24de25b52f
example to benchmark onnx [pr] ( #8459 )
...
* example to benchmark onnx [pr]
* reset global count
2024-12-31 11:38:33 -05:00
qazal
866dfa1f23
create_schedule([x.lazydata]) -> x.schedule() in tests ( #8449 )
2024-12-31 03:15:52 +08:00
Francis Lata
40d6752854
adjust regression loss to mask after L1 loss is calculated
2024-12-27 17:41:12 +00:00
Francis Lata
cc4a673aa9
Merge branch 'master' into retinanet_mlperf
2024-12-26 21:11:55 +00:00
Calum
d8b08790b9
Fix examples/conversation.py ( #8425 )
...
* fix: conversation example
* remove slice func
* remove unused import
* use Tensor.split
2024-12-26 12:45:19 -05:00
chenyu
4712847766
make self_tokenize output more like a python file ( #8411 )
...
use comment for file name and join with newline instead of null byte when export to file
2024-12-25 14:16:30 -05:00
chenyu
a35eef8d58
optionally output to file in self_tokenize.py ( #8399 )
...
can paste the whole tinygrad in gemini this way
2024-12-24 21:09:26 -05:00
Francis Lata
c1a18e13ef
make training work
2024-12-23 21:48:55 +00:00
Francis Lata
d1627d0b1b
start re-enabling training step
2024-12-23 19:43:20 +00:00
Francis Lata
44abfbcacb
Merge branch 'master' into retinanet_mlperf
2024-12-23 05:36:37 +00:00
Francis Lata
8defd337d8
fixes after helper refactor cleanup
2024-12-23 05:30:03 +00:00
Francis Lata
fb689f7097
move BoxCoder to MLPerf helpers
2024-12-23 05:12:51 +00:00
Francis Lata
d57f7cc209
fix regression loss
2024-12-21 09:34:10 +00:00
Francis Lata
630267914f
implement regression loss
2024-12-20 23:38:25 +00:00
Francis Lata
971d10361f
revert anchors to use np
2024-12-20 16:56:50 +00:00
Francis Lata
67da37e390
simplify anchors batching
2024-12-20 15:47:41 +00:00
Harald Schäfer
7059459648
Openpilot compile: fix for openpilot use ( #8338 )
...
* compile3 changes
* merge conflict
* merge conflict
* give dm npy for now
* Revert "give dm npy for now"
This reverts commit bfd980da7d2c2bab5b073127442c361922032ba1.
* updates
* Always float32 floats
* Update compile3.py
* Update compile3.py
---------
Co-authored-by: ZwX1616 <zwx1616@gmail.com >
2024-12-19 19:43:15 -05:00
Francis Lata
759e1d6cbc
make anchors use Tensors
2024-12-19 22:20:03 +00:00
George Hotz
8f95b578f6
use Estimates class [pr] ( #8319 )
...
* use Estimates class [pr]
* frozen dataclass
2024-12-18 10:19:32 -08:00
George Hotz
37fa38d272
Revert "switch beautiful_mnist to use new optimizer [pr] ( #8231 )" ( #8233 )
...
This reverts commit e9ee39df22
2024-12-13 19:07:09 -08:00
