CyberSolve-DeepMind-LinAlg-1D-downsample-v2 / generation_config.json

A second finetuing of the flan-T5-large model on the downsampled DeepMind LingAlg 1D dataset, this time with a GPU batch size of 256 as opposed to 32 used before

f9ec778 verified 2 months ago

raw

history blame contribute delete

142 Bytes

	{
	"_from_model_config": true,
	"decoder_start_token_id": 0,
	"eos_token_id": 1,
	"pad_token_id": 0,
	"transformers_version": "4.47.1"
	}