Llama2-7B-lora-r-32-generic-step-1100-lr-1e-5-labels_40.0

This model is a fine-tuned version of meta-llama/Llama-2-7b-hf on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 2.9175

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: 1e-05
• 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: cosine
• lr_scheduler_warmup_steps: 20
• training_steps: 1100
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
5.501	0.3653	20	5.5640
4.8032	0.7306	40	4.8354
4.2159	1.0959	60	4.3007
3.9751	1.4612	80	4.0222
3.8467	1.8265	100	3.8388
3.5551	2.1918	120	3.7061
3.5184	2.5571	140	3.5941
3.4653	2.9224	160	3.5033
3.3495	3.2877	180	3.4278
3.2385	3.6530	200	3.3642
3.2138	4.0183	220	3.3090
3.1979	4.3836	240	3.2627
3.0379	4.7489	260	3.2206
3.038	5.1142	280	3.1844
3.0155	5.4795	300	3.1540
2.9582	5.8447	320	3.1237
2.9131	6.2100	340	3.0988
2.8304	6.5753	360	3.0768
2.8736	6.9406	380	3.0555
2.7928	7.3059	400	3.0390
2.7968	7.6712	420	3.0242
2.7866	8.0365	440	3.0062
2.7656	8.4018	460	2.9970
2.7002	8.7671	480	2.9867
2.648	9.1324	500	2.9757
2.6285	9.4977	520	2.9711
2.6431	9.8630	540	2.9611
2.6574	10.2283	560	2.9558
2.6849	10.5936	580	2.9502
2.6184	10.9589	600	2.9439
2.9509	11.3242	620	2.9405
2.5995	11.6895	640	2.9381
2.6439	12.0548	660	2.9330
2.6179	12.4201	680	2.9319
2.5906	12.7854	700	2.9285
2.6481	13.1507	720	2.9246
2.5334	13.5160	740	2.9256
2.5575	13.8813	760	2.9221
2.5301	14.2466	780	2.9204
2.4874	14.6119	800	2.9211
2.4977	14.9772	820	2.9189
2.4987	15.3425	840	2.9188
2.5354	15.7078	860	2.9172
2.5015	16.0731	880	2.9177
2.4765	16.4384	900	2.9183
2.5101	16.8037	920	2.9176
2.5042	17.1689	940	2.9167
2.4973	17.5342	960	2.9173
2.6753	17.8995	980	2.9167
2.5334	18.2648	1000	2.9167
2.4684	18.6301	1020	2.9170
2.4829	18.9954	1040	2.9173
2.4656	19.3607	1060	2.9174
2.4721	19.7260	1080	2.9175
2.5127	20.0913	1100	2.9175

Framework versions

• PEFT 0.15.2
• Transformers 4.45.2
• Pytorch 2.5.0+cu121
• Datasets 3.2.0
• Tokenizers 0.20.3