mistral_7b_0-3_oh-dcft-v3.1-qwen-2.5-72b

This model is a fine-tuned version of mistralai/Mistral-7B-v0.3 on the mlfoundations-dev/oh-dcft-v3.1-qwen-2.5-72b dataset. It achieves the following results on the evaluation set:

• Loss: 0.3581

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: 5e-06
• train_batch_size: 16
• eval_batch_size: 8
• seed: 42
• distributed_type: multi-GPU
• num_devices: 32
• total_train_batch_size: 512
• total_eval_batch_size: 256
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine_with_min_lr
• lr_scheduler_warmup_ratio: 0.05
• num_epochs: 3.0

Training results

Training Loss	Epoch	Step	Validation Loss
0.3426	1.0	534	0.3397
0.264	2.0	1068	0.3331
0.1874	3.0	1602	0.3581

Framework versions

• Transformers 4.46.1
• Pytorch 2.4.0
• Datasets 3.0.2
• Tokenizers 0.20.3