coinplusfire_llm_2

This model is a fine-tuned version of TheBloke/Mistral-7B-Instruct-v0.2-GPTQ on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.1450

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: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 2
• num_epochs: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
2.2919	0.99	51	1.8319
1.6082	1.99	103	1.6426
1.4689	3.0	155	1.5522
1.3821	4.0	207	1.4883
1.3406	4.99	258	1.4421
1.2592	5.99	310	1.3900
1.2115	7.0	362	1.3508
1.1705	8.0	414	1.3213
1.1555	8.99	465	1.2913
1.1031	9.99	517	1.2629
1.0727	11.0	569	1.2418
1.0481	12.0	621	1.2208
1.0466	12.99	672	1.1971
1.006	13.99	724	1.1864
0.989	15.0	776	1.1732
0.9719	16.0	828	1.1589
0.979	16.99	879	1.1535
0.9494	17.99	931	1.1469
0.9401	19.0	983	1.1449
0.9302	19.71	1020	1.1450

Framework versions

• PEFT 0.10.0
• Transformers 4.38.2
• Pytorch 2.1.0+cu121
• Datasets 2.18.0
• Tokenizers 0.15.2