translation_llm

This model is a fine-tuned version of NousResearch/Llama-2-7b-hf on an dataset that combines russian sentences from two datasets: parallel-sentences-tatoeba and parallel-sentences-wikimatrix. The model is finetuned for russian language.

It achieves the following results on the evaluation set:

• Loss: 0.0039

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: 4
• eval_batch_size: 8
• 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: linear
• lr_scheduler_warmup_steps: 50
• num_epochs: 5

Training results

Training Loss	Epoch	Step	Validation Loss
0.0063	0.8591	1000	0.0110
0.0041	1.7181	2000	0.0089
0.0016	2.5772	3000	0.0086
0.0006	3.4362	4000	0.0041
0.0023	4.2953	5000	0.0039

Framework versions

• PEFT 0.13.0
• Transformers 4.44.2
• Pytorch 2.4.1+cu121
• Datasets 3.0.1
• Tokenizers 0.19.1