src/r1-v/local_scripts/zero1_no_optimizer.json

{
	"zero_optimization": {
		"stage": 1,
		"allgather_partitions": true,
		"allgather_bucket_size": 1e9,
		"overlap_comm": false,
		"reduce_scatter": true,
		"reduce_bucket_size": 1e9,
		"contiguous_gradients": true
	},
	"fp16": {
		"enabled": "auto",
		"auto_cast": true,
		"loss_scale": 0,
		"initial_scale_power": 32,
		"loss_scale_window": 1000,
		"hysteresis": 2,
		"min_loss_scale": 1
	},
	"bf16": {
		"enabled": "auto"
	},
	"gradient_accumulation_steps": "auto",
	"gradient_clipping": "auto",
	"steps_per_print": 1,
	"train_batch_size": "auto",
	"train_micro_batch_size_per_gpu": "auto",
	"wall_clock_breakdown": true
}