Commit Graph

277 Commits

Author SHA1 Message Date
wozeparrot
b979162c5d llama3 eval train (#11706) 2025-08-20 19:56:35 -04:00
chenyu
dbd3b67657 clamp GRAD_CLIP_NORM in llama (#11761) 2025-08-20 19:55:50 -04:00
chenyu
e9d0027591 llama MP realize weight after shard (#11672)
* llama MP realize weight after shard

prevents memory spike on device 0

* empty weight for FAKEDATA
2025-08-14 16:17:46 -04:00
chenyu
ef17af85c6 remove .float call in llama logit (#11598)
* remove .float call in llama logit

* bfloat item
2025-08-10 00:02:18 -04:00
chenyu
45baec1aab model parallel llama (#11588)
MP=8 GRADIENT_ACC_STEPS=3 BS=1 DEFAULT_FLOAT=bfloat16 OPTIM_DTYPE=bfloat16 LLAMA3_SIZE=70B SEQLEN=512 PYTHONPATH=. MODEL=llama3 python3 examples/mlperf/model_train.py
2025-08-09 16:54:27 -04:00
chenyu
702e38dc19 remove FUSE_ARANGE_UINT (#11567)
also add IGNORE_OOB=1 to bert runs. lowered BS on tinybox to 90 since 96 oom during eval without reset
2025-08-07 16:49:06 -04:00
wozeparrot
7ae4335127 feat: generate blend index (#11566) 2025-08-07 14:20:28 -04:00
wozeparrot
2d5bdc939d faster llama3 dataloader (#11540) 2025-08-06 18:25:57 -04:00
chenyu
f7965f85aa Revert "feat: faster index building (#11462)" (#11478)
This reverts commit 3a4deb08d2.
2025-08-02 12:50:48 -04:00
wozeparrot
3a4deb08d2 feat: faster index building (#11462)
* feat: faster index building

* feat: correct training samples
2025-08-02 11:50:18 -04:00
chenyu
9e8e6b45ab grad acc train llama (#11467)
* grad acc train llama

* log step time
2025-08-01 15:54:50 -04:00
chenyu
7ad7329257 data parallel train llama (#11466) 2025-08-01 12:13:51 -04:00
George Hotz
8ff03806e8 add llama layers (#11460)
* add llama layers

* add contig bw for speed
2025-07-31 16:28:04 -07:00
wozeparrot
6252f7770e feat: fake data (#11447) 2025-07-30 17:18:20 -07:00
chenyu
e300451f3a update llama3 (#11446)
`LR=1e-4 TRAIN_ON_VAL=1 DEFAULT_FLOAT=bfloat16 FUSE_ARANGE=1 JITBEAM=2 OPTIM_DTYPE=bfloat16 LLAMA3_SIZE=1B WARMUP_STEPS=36 DECAY_STEPS=360 SEQLEN=512 PYTHONPATH=. AMD=1 AMD_LLVM=0 MODEL=llama3 python3 examples/mlperf/model_train.py` trained to 7
2025-07-30 19:34:21 -04:00
wozeparrot
5fb975351a feat: flag for training on val (#11441) 2025-07-30 14:29:45 -07:00
wozeparrot
825b6a2505 feat: llama3 dataloader (#11340) 2025-07-30 13:27:55 -07:00
chenyu
c14c9a8eff llama3 grad clip (#11003) 2025-06-27 19:14:12 -04:00
chenyu
f2548afeb5 bert grad clipping start with const 0 (#11008)
saved the init kernels
2025-06-27 18:02:23 -04:00
chenyu
6ab5a5cb6c llama3 mlperf train (#10983)
work in progress. now it can overfit small examples and vram roughly matches
2025-06-26 20:24:27 -04:00
chenyu
8751d47985 CosineAnnealingLRWithWarmup (#10981) 2025-06-25 17:45:21 -04:00
chenyu
efad567ebd ruff check whole examples/mlperf/ (#10979) 2025-06-25 12:57:48 -04:00
chenyu
0480139def log_perplexity metrics (#10912) 2025-06-21 10:44:47 -04:00
chenyu
62a540066e remove DEBUG=2 in mi300x bert setup (#10886)
seems fine now, not sure what the issue was
2025-06-19 13:28:53 -04:00
chenyu
f377cc19cd use AM for bert (#10882)
have triained 3 runs and all seem fine
2025-06-19 09:48:54 -04:00
chenyu
b70c7d3631 bert grad accumulation (#10863)
* bert grad accumulation

* realize grad
2025-06-18 12:17:07 -04:00
chenyu
075a74cf25 add global_batch_size to mlperf bert (#10852)
global_batch_size = grad_acc_steps * batch_size. no-op change to prep grad acc for bert
2025-06-17 17:54:15 -04:00
chenyu
81e296d7b8 remove Tensor.test() in retinanet (#10770)
test was removed
2025-06-10 22:14:57 -04:00
George Hotz
32e9949052 rename lazydata to uop (#10698) 2025-06-08 08:42:22 -07:00
chenyu
4ab3391e6f set -o pipefail for mlperf run_and_time (#10577)
also run the 5.1 script in ci cron job
2025-05-30 16:36:44 -04:00
chenyu
baf482d314 copy mlperf stuff to 5.1 (#10576)
5.0 is finalized, new changes go to 5.1
2025-05-30 16:12:39 -04:00
George Hotz
b3b43a82c4 remove Tensor.no_grad, it's meaningless now [pr] (#10556) 2025-05-28 22:20:02 -07:00
chenyu
74cf5dbd9e mlperf system updates (#10550)
standardized processor and accelerator names
2025-05-28 16:15:46 -04:00
chenyu
51dc7eedb0 correct use AM for resnet run_and_time (#10524) 2025-05-26 15:33:11 -04:00
chenyu
c1919ad55f use AM for resnet run_and_time (#10523) 2025-05-26 14:50:49 -04:00
chenyu
2d50efb92b set -e on mlperf run_and_time scripts (#10519) 2025-05-26 09:22:30 -04:00
chenyu
dc6309242d WallTimeEvent for mlperf ci (#10506) 2025-05-24 10:56:03 -04:00
chenyu
67d1364106 update LOGMLPERF in red resnet run_and_time (#10416) 2025-05-19 13:23:33 -04:00
chenyu
485e80da69 run_and_time for resnet ci (#10405) 2025-05-18 23:39:57 -04:00
wozeparrot
1ed04f993b move benchmark stat tracking to influxdb (#10185) 2025-05-15 16:14:56 -07:00
George Hotz
568d6d96e7 small changes from new multi [pr] (#10318) 2025-05-14 20:50:59 -07:00
George Hotz
bfc30fa6ea hotfix: typo in shm_name 2025-05-14 19:34:52 -07:00
George Hotz
2bc54b3e22 manually handle OSX 2025-05-14 19:17:51 -07:00
George Hotz
ab460486d7 Revert "resnet dataloader osx (#10316)"
This reverts commit aef336930a.
2025-05-14 19:15:07 -07:00
George Hotz
aef336930a resnet dataloader osx (#10316)
* mlperf dataloader on mac

* resnet dataloader [pr]

* simple should work
2025-05-14 18:31:26 -07:00
chenyu
610ee79b22 cherry pick mlperf5.0 branch to master (#10089) 2025-04-28 15:36:56 -04:00
chenyu
74c6cf8be3 lint mlperf model_train (#10038) 2025-04-24 16:19:44 -04:00
chenyu
a25abf55e3 retinanet only call postprocess_detections with RUNMLPERF (#10017)
during setup only need to compile `_eval_step().numpy()`
2025-04-23 20:45:38 -04:00
chenyu
65faa1d94b explicit device in mlperf scripts (#10015) 2025-04-23 17:11:52 -04:00
chenyu
a3f938dbee remove retinanet INITMLPERF from beam script (#10011)
it only controls logging, loading real data or not is solely controlled by RUNMLPERF
2025-04-23 14:32:54 -04:00