shawgpt-ft

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: 1.3302

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: 0.0001
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 16
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 5
• num_epochs: 30
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
4.9895	0.8571	3	4.1658
3.4778	2.0	7	3.7207
4.1399	2.8571	10	3.3552
2.747	4.0	14	2.9052
3.2593	4.8571	17	2.6105
2.1018	6.0	21	2.2608
2.4159	6.8571	24	2.0093
1.6044	8.0	28	1.7780
1.9126	8.8571	31	1.6498
1.2575	10.0	35	1.5062
1.6232	10.8571	38	1.4496
1.1013	12.0	42	1.3928
1.4709	12.8571	45	1.3688
1.0337	14.0	49	1.3495
1.4037	14.8571	52	1.3399
1.0328	16.0	56	1.3307
1.3526	16.8571	59	1.3277
0.9745	18.0	63	1.3259
1.299	18.8571	66	1.3254
0.9387	20.0	70	1.3259
1.3275	20.8571	73	1.3271
0.9127	22.0	77	1.3290
1.2574	22.8571	80	1.3300
0.9474	24.0	84	1.3305
1.2843	24.8571	87	1.3302
0.8341	25.7143	90	1.3302

Framework versions

• PEFT 0.13.2
• Transformers 4.44.2
• Pytorch 2.5.0+cu121
• Datasets 3.0.2
• Tokenizers 0.19.1