Mixtral_R2_v0

This model is a fine-tuned version of mistralai/Mixtral-8x7B-Instruct-v0.1 on the generator dataset. It achieves the following results on the evaluation set:

• Loss: 0.7671

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: 2.5e-05
• train_batch_size: 32
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 0.03
• num_epochs: 25

Training results

Training Loss	Epoch	Step	Validation Loss
3.0101	1.0	18	2.1741
2.1612	2.0	36	1.5962
1.6591	3.0	54	1.4202
1.4985	4.0	72	1.3035
1.3585	5.0	90	1.1977
1.294	6.0	108	1.0993
1.1823	7.0	126	1.0139
1.0983	8.0	144	0.9641
1.0371	9.0	162	0.9293
0.9868	10.0	180	0.8961
0.9535	11.0	198	0.8655
0.9259	12.0	216	0.8358
0.882	13.0	234	0.8067
0.8472	14.0	252	0.7938
0.8484	15.0	270	0.7872
0.8215	16.0	288	0.7826
0.8167	17.0	306	0.7779
0.8199	18.0	324	0.7751
0.8042	19.0	342	0.7730
0.8186	20.0	360	0.7710
0.794	21.0	378	0.7698
0.7958	22.0	396	0.7685
0.7858	23.0	414	0.7677
0.7857	24.0	432	0.7671
0.7843	25.0	450	0.7671

Framework versions

• PEFT 0.7.1
• Transformers 4.36.2
• Pytorch 2.1.2+cu121
• Datasets 2.16.1
• Tokenizers 0.15.0