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: 1.4358

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: 16
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 2
• num_epochs: 40
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
3.6983	1.0	1	3.6528
3.5663	2.0	2	3.5662
3.554	3.0	3	3.3946
3.3511	4.0	4	3.2324
3.271	5.0	5	3.0797
3.0504	6.0	6	2.9379
2.9238	7.0	7	2.8082
2.7588	8.0	8	2.6900
2.6477	9.0	9	2.5812
2.5407	10.0	10	2.4791
2.4989	11.0	11	2.3837
2.3429	12.0	12	2.3018
2.2806	13.0	13	2.2213
2.1603	14.0	14	2.1381
2.1297	15.0	15	2.0542
2.0215	16.0	16	1.9790
1.9543	17.0	17	1.9113
1.8405	18.0	18	1.8506
1.7458	19.0	19	1.7989
1.6685	20.0	20	1.7535
1.7441	21.0	21	1.7127
1.5727	22.0	22	1.6772
1.6027	23.0	23	1.6452
1.5803	24.0	24	1.6143
1.589	25.0	25	1.5838
1.512	26.0	26	1.5563
1.4089	27.0	27	1.5346
1.4374	28.0	28	1.5160
1.3226	29.0	29	1.4998
1.3265	30.0	30	1.4862
1.3214	31.0	31	1.4750
1.4226	32.0	32	1.4657
1.4556	33.0	33	1.4583
1.3064	34.0	34	1.4522
1.2501	35.0	35	1.4472
1.3525	36.0	36	1.4432
1.2416	37.0	37	1.4402
1.3267	38.0	38	1.4380
1.2629	39.0	39	1.4365
1.2634	40.0	40	1.4358

Framework versions

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