ruadapt_solar_10.7_part2_v3_as2_rsg_lora

This model is a fine-tuned version of RefalMachine/ruadapt_solar_10.7_darulm_unigram_proj_init_part2_v3_alpha_scale_2 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0744

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: 0.00025
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 32
• total_train_batch_size: 128
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 30
• num_epochs: 3
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
0.1	0.45	100	0.0669
0.0837	0.9	200	0.0682
0.0511	1.35	300	0.0762
0.0473	1.8	400	0.0662
0.0188	2.24	500	0.0841
0.0268	2.69	600	0.0744

Framework versions

• Transformers 4.35.0
• Pytorch 2.1.2
• Datasets 2.14.4
• Tokenizers 0.14.1