Mistral-7B-Instruct-v0.2-miracl-raft-sft-v2.0

This model is a fine-tuned version of mistralai/Mistral-7B-Instruct-v0.2 on the nthakur/miracl-raft-sft-instruct-v0.2 dataset. It achieves the following results on the evaluation set:

• Loss: 1.2086

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: 1e-05
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• distributed_type: multi-GPU
• num_devices: 3
• gradient_accumulation_steps: 2
• total_train_batch_size: 24
• total_eval_batch_size: 12
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 1

Training results

Training Loss	Epoch	Step	Validation Loss
1.3095	0.0987	200	1.2800
1.249	0.1975	400	1.2516
1.2514	0.2962	600	1.2369
1.275	0.3950	800	1.2263
1.1984	0.4937	1000	1.2197
1.1556	0.5924	1200	1.2149
1.2386	0.6912	1400	1.2116
1.2661	0.7899	1600	1.2096
1.2752	0.8887	1800	1.2088
1.2701	0.9874	2000	1.2086

Framework versions

• PEFT 0.7.1
• Transformers 4.40.1
• Pytorch 2.2.1+cu121
• Datasets 2.19.0
• Tokenizers 0.19.1