pk-bloomz

This model is a fine-tuned version of bigscience/bloomz-3b on the None dataset. It achieves the following results on the evaluation set:

• Loss: 2.9338

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 0.0002
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 5

Training results

Training Loss	Epoch	Step	Validation Loss
3.5665	0.3556	100	3.1386
3.0398	0.7111	200	3.0567
2.9252	1.064	300	3.0102
2.8475	1.4196	400	2.9824
2.8068	1.7751	500	2.9564
2.736	2.128	600	2.9535
2.6993	2.4836	700	2.9417
2.6785	2.8391	800	2.9309
2.6215	3.192	900	2.9332
2.575	3.5476	1000	2.9244
2.5789	3.9031	1100	2.9233
2.4963	4.256	1200	2.9345
2.5072	4.6116	1300	2.9338
2.4719	4.9671	1400	2.9338

Framework versions

• PEFT 0.15.2
• Transformers 4.52.4
• Pytorch 2.6.0+cu124
• Datasets 3.6.0
• Tokenizers 0.21.2