Upload train_stats.json with huggingface_hub

8410449 verified about 1 month ago

1.85 kB

{
 "base": "Qwen/Qwen3-14B",
 "method": "fsdp2-custom-loop",
 "n_rows": 39729,
 "n_self": 0,
 "n_sequences": 20171,
 "n_tokens": 32094200,
 "world": 4,
 "micro_batch": 4,
 "accum": 4,
 "steps": 315,
 "lr": 1e-05,
 "optim": "AdamW fused, fp32 master, bf16 compute, fp32 grad reduce",
 "loss_curve": [
3082733154296875,
9415383338928223,
7052346467971802,
621016502380371,
471733570098877,
6130437850952148,
5772490501403809,
52907395362854,
428034782409668,
4627792835235596,
461775302886963,
4702470302581787,
4943097829818726,
4746789932250977,
4906061887741089,
4127854108810425,
4552178382873535,
4429655075073242,
470187783241272,
5004136562347412,
457139015197754,
432055115699768,
47483491897583,
4406695365905762,
4969544410705566,
4111571311950684,
3831615447998047,
3048855066299438,
3899734020233154,
3976258039474487,
4286668300628662,
39436674118042,
422532081604004,
452055811882019,
4810572862625122,
4608440399169922,
4630420207977295,
408216118812561,
4094760417938232,
4694641828536987,
378267765045166,
457957148551941,
380078911781311,
3461487293243408,
374085783958435,
3755388259887695,
5014238357543945,
3779674768447876,
4091764688491821,
3811347484588623,
3926327228546143,
436885118484497,
3584022521972656,
3977388143539429,
4249296188354492,
3708088397979736,
401006817817688,
3582624197006226,
5031040906906128,
4119453430175781,
4745867252349854,
3782695531845093,
4943677186965942
 ],
 "loss_first": [
3082733154296875
 ],
 "loss_last": [
4943677186965942
 ],
 "train_seconds": 3221.413412332535,
 "tokens_per_second": 9962.769720003585,
 "resumed_from": null
}

	{
	"base": "Qwen/Qwen3-14B",
	"method": "fsdp2-custom-loop",
	"n_rows": 39729,
	"n_self": 0,
	"n_sequences": 20171,
	"n_tokens": 32094200,
	"world": 4,
	"micro_batch": 4,
	"accum": 4,
	"steps": 315,
	"lr": 1e-05,
	"optim": "AdamW fused, fp32 master, bf16 compute, fp32 grad reduce",
	"loss_curve": [
	2.3082733154296875,
	1.9415383338928223,
	1.7052346467971802,
	1.621016502380371,
	1.471733570098877,
	1.6130437850952148,
	1.5772490501403809,
	1.52907395362854,
	1.428034782409668,
	1.4627792835235596,
	1.461775302886963,
	1.4702470302581787,
	1.4943097829818726,
	1.4746789932250977,
	1.4906061887741089,
	1.4127854108810425,
	1.4552178382873535,
	1.4429655075073242,
	1.470187783241272,
	1.5004136562347412,
	1.457139015197754,
	1.432055115699768,
	1.47483491897583,
	1.4406695365905762,
	1.4969544410705566,
	1.4111571311950684,
	1.3831615447998047,
	1.3048855066299438,
	1.3899734020233154,
	1.3976258039474487,
	1.4286668300628662,
	1.39436674118042,
	1.422532081604004,
	1.452055811882019,
	1.4810572862625122,
	1.4608440399169922,
	1.4630420207977295,
	1.408216118812561,
	1.4094760417938232,
	1.4694641828536987,
	1.378267765045166,
	1.457957148551941,
	1.380078911781311,
	1.3461487293243408,
	1.374085783958435,
	1.3755388259887695,
	1.5014238357543945,
	1.3779674768447876,
	1.4091764688491821,
	1.3811347484588623,
	1.3926327228546143,
	1.436885118484497,
	1.3584022521972656,
	1.3977388143539429,
	1.4249296188354492,
	1.3708088397979736,
	1.401006817817688,
	1.3582624197006226,
	1.5031040906906128,
	1.4119453430175781,
	1.4745867252349854,
	1.3782695531845093,
	1.4943677186965942
	],
	"loss_first": [
	2.3082733154296875
	],
	"loss_last": [
	1.4943677186965942
	],
	"train_seconds": 3221.413412332535,
	"tokens_per_second": 9962.769720003585,
	"resumed_from": null
	}