Mistral-7B-Instruct-v0.3-advisegpt-v0.1

This model is a fine-tuned version of mistralai/Mistral-7B-Instruct-v0.3 on the generator dataset. It achieves the following results on the evaluation set:

• Loss: 0.0762
• Bleu: {'bleu': 0.9585152983456746, 'precisions': [0.9779106264925121, 0.9626412004947897, 0.951895206199588, 0.9430042745426802], 'brevity_penalty': 0.999729358235579, 'length_ratio': 0.9997293948524543, 'translation_length': 1289353, 'reference_length': 1289702}
• Rouge: {'rouge1': 0.9761616288162651, 'rouge2': 0.9590944581779459, 'rougeL': 0.9748018206627191, 'rougeLsum': 0.9758991028742771}

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: 2e-05
• train_batch_size: 3
• eval_batch_size: 1
• seed: 42
• gradient_accumulation_steps: 10
• total_train_batch_size: 30
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• num_epochs: 3
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Bleu	Validation Loss	Rouge
0.0675	0.9998	809	{'bleu': 0.9495787144110293, 'brevity_penalty': 0.9993236461934566, 'length_ratio': 0.9993238748175935, 'precisions': [0.9735806894625358, 0.9547064588389024, 0.9417775802515341, 0.9313417436570839], 'reference_length': 1289702, 'translation_length': 1288830}	0.0936	{'rouge1': 0.9713550622229471, 'rouge2': 0.9502301622796694, 'rougeL': 0.9691228372678113, 'rougeLsum': 0.9708685856330016}
0.0548	1.9998	1618	0.0771	{'bleu': 0.9571495232321637, 'precisions': [0.9773150683231548, 0.9614637013631232, 0.95029828201407, 0.9410538932261553], 'brevity_penalty': 0.9996991100504438, 'length_ratio': 0.9996991553087458, 'translation_length': 1289314, 'reference_length': 1289702}	{'rouge1': 0.9755343391324649, 'rouge2': 0.9577790978374392, 'rougeL': 0.9740177474237091, 'rougeLsum': 0.9752585254668996}
0.0439	2.9995	2427	0.0762	{'bleu': 0.9585152983456746, 'precisions': [0.9779106264925121, 0.9626412004947897, 0.951895206199588, 0.9430042745426802], 'brevity_penalty': 0.999729358235579, 'length_ratio': 0.9997293948524543, 'translation_length': 1289353, 'reference_length': 1289702}	{'rouge1': 0.9761616288162651, 'rouge2': 0.9590944581779459, 'rougeL': 0.9748018206627191, 'rougeLsum': 0.9758991028742771}

Framework versions

• PEFT 0.10.0
• Transformers 4.40.2
• Pytorch 2.3.0+cu121
• Datasets 2.19.1
• Tokenizers 0.19.1