--- base_model: mistralai/Mistral-7B-Instruct-v0.3 datasets: - generator library_name: peft license: apache-2.0 tags: - trl - sft - generated_from_trainer model-index: - name: mistral_7b_cosine_lr_2e-4_bs2 results: [] --- # mistral_7b_cosine_lr_2e-4_bs2 This model is a fine-tuned version of [mistralai/Mistral-7B-Instruct-v0.3](https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.3) on the generator dataset. It achieves the following results on the evaluation set: - Loss: 0.3819 ## 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: 8 - seed: 42 - gradient_accumulation_steps: 8 - total_train_batch_size: 16 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: cosine - lr_scheduler_warmup_ratio: 0.03 - lr_scheduler_warmup_steps: 15 - num_epochs: 4 ### Training results | Training Loss | Epoch | Step | Validation Loss | |:-------------:|:------:|:----:|:---------------:| | 3.5854 | 0.0366 | 10 | 0.7044 | | 0.6542 | 0.0732 | 20 | 0.8683 | | 0.5736 | 0.1098 | 30 | 0.5023 | | 0.4886 | 0.1465 | 40 | 0.4735 | | 0.4757 | 0.1831 | 50 | 0.4552 | | 0.453 | 0.2197 | 60 | 0.4451 | | 0.4494 | 0.2563 | 70 | 0.4380 | | 0.4457 | 0.2929 | 80 | 0.4329 | | 0.4353 | 0.3295 | 90 | 0.4271 | | 0.434 | 0.3661 | 100 | 0.4239 | | 0.4307 | 0.4027 | 110 | 0.4198 | | 0.4256 | 0.4394 | 120 | 0.4167 | | 0.4173 | 0.4760 | 130 | 0.4130 | | 0.4195 | 0.5126 | 140 | 0.4100 | | 0.4159 | 0.5492 | 150 | 0.4075 | | 0.4102 | 0.5858 | 160 | 0.4045 | | 0.4135 | 0.6224 | 170 | 0.4034 | | 0.408 | 0.6590 | 180 | 0.4004 | | 0.405 | 0.6957 | 190 | 0.3992 | | 0.4053 | 0.7323 | 200 | 0.3960 | | 0.3994 | 0.7689 | 210 | 0.3934 | | 0.3968 | 0.8055 | 220 | 0.3914 | | 0.3966 | 0.8421 | 230 | 0.3885 | | 0.3894 | 0.8787 | 240 | 0.3868 | | 0.3896 | 0.9153 | 250 | 0.3860 | | 0.3939 | 0.9519 | 260 | 0.3836 | | 0.387 | 0.9886 | 270 | 0.3818 | | 0.3511 | 1.0252 | 280 | 0.3839 | | 0.3316 | 1.0618 | 290 | 0.3834 | | 0.3281 | 1.0984 | 300 | 0.3819 | ### Framework versions - PEFT 0.13.2 - Transformers 4.45.2 - Pytorch 2.4.1+cu121 - Datasets 3.0.1 - Tokenizers 0.20.0