scrapie/dolphin-2.9-llama3-8b-GER-4bit

dolphin-2.9-llama3-8b-GER

This model is a fine-tuned version of cognitivecomputations/dolphin-2.9-llama3-8b on the identity, the alpaca-gpt4_de, the dolphin_de and the airoboros_de datasets. It achieves the following results on the evaluation set:

• Loss: 0.9384

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: 2
• eval_batch_size: 1
• seed: 42
• distributed_type: multi-GPU
• num_devices: 2
• gradient_accumulation_steps: 4
• total_train_batch_size: 16
• total_eval_batch_size: 2
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• lr_scheduler_warmup_steps: 80
• num_epochs: 1.0
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
1.2054	0.12	100	1.0369
1.0667	0.24	200	1.0012
1.0751	0.35	300	0.9849
0.8838	0.47	400	0.9696
0.9846	0.59	500	0.9565
0.9523	0.71	600	0.9486
0.8567	0.82	700	0.9430
0.8284	0.94	800	0.9384

Framework versions

• PEFT 0.10.0
• Transformers 4.39.3
• Pytorch 2.2.2+cu121
• Datasets 2.16.0
• Tokenizers 0.15.2