Datasets:

tiny226
/

small-model-simulation-data

The quick brown fox jumps over the lazy dog.
Language models learn statistical patterns from sequences of tokens.
This is a small smoke test for distributed full fine tuning.
The model should process this text and compute a training loss.
DeepSpeed ZeRO two distributes optimizer states and gradients across GPUs.
We are testing Qwen3 one point seven billion parameter training.
The purpose of this dataset is only to verify the training pipeline.

	The quick brown fox jumps over the lazy dog.
	Language models learn statistical patterns from sequences of tokens.
	This is a small smoke test for distributed full fine tuning.
	The model should process this text and compute a training loss.
	DeepSpeed ZeRO two distributes optimizer states and gradients across GPUs.
	We are testing Qwen3 one point seven billion parameter training.
	The purpose of this dataset is only to verify the training pipeline.