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: 1.3292

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: 3
• total_train_batch_size: 24
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 2
• num_epochs: 25
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
2.279	0.8571	2	1.9568
2.0827	1.7143	4	1.7515
1.1669	3.0	7	1.5483
1.5868	3.8571	9	1.4687
1.587	4.7143	11	1.4263
0.948	6.0	14	1.3811
1.3703	6.8571	16	1.3654
1.4095	7.7143	18	1.3524
0.9208	9.0	21	1.3367
1.3025	9.8571	23	1.3315
1.3563	10.7143	25	1.3298
0.8394	12.0	28	1.3292
1.3065	12.8571	30	1.3307
1.2415	13.7143	32	1.3309
0.85	15.0	35	1.3344
1.2656	15.8571	37	1.3335
1.2467	16.7143	39	1.3326
0.8036	18.0	42	1.3319
1.2126	18.8571	44	1.3326
1.1744	19.7143	46	1.3328
0.8366	21.0	49	1.3333
1.0147	21.4286	50	1.3333

Framework versions

• PEFT 0.13.2
• Transformers 4.45.2
• Pytorch 2.4.1+cu121
• Datasets 3.0.1
• Tokenizers 0.20.1