lmind_nq_train6000_eval6489_v1_doc_qa_v3_5e-4_lora2

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: nan
• Accuracy: 0.1684

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: 0.0005
• train_batch_size: 2
• eval_batch_size: 2
• seed: 42
• distributed_type: multi-GPU
• num_devices: 4
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• total_eval_batch_size: 8
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: constant
• lr_scheduler_warmup_ratio: 0.05
• num_epochs: 50.0

Training results

Training Loss	Epoch	Step	Accuracy	Validation Loss
1.3974	1.0	529	0.6156	1.3078
1.2394	2.0	1058	0.6122	1.3467
1.0416	3.0	1587	0.6110	1.3860
0.8277	4.0	2116	0.6089	1.4837
0.6549	5.0	2645	0.6064	1.5426
0.5022	6.0	3174	0.6093	1.5873
0.4309	7.0	3703	0.6067	1.6567
7.4292	8.0	4232	0.1972	6.4572
9.3441	9.0	4761	0.2368	7.3289
8.2455	10.0	5290	0.1801	8.0851
0.0	11.0	5819	0.1684	nan
0.0	12.0	6348	0.1684	nan
0.0	13.0	6877	0.1684	nan
0.0	14.0	7406	0.1684	nan
0.0	15.0	7935	0.1684	nan
0.0	16.0	8464	0.1684	nan
0.0	17.0	8993	0.1684	nan
0.0	18.0	9522	0.1684	nan
0.0	19.0	10051	0.1684	nan
0.0	20.0	10580	0.1684	nan
0.0	21.0	11109	0.1684	nan
0.0	22.0	11638	0.1684	nan
0.0	23.0	12167	0.1684	nan
0.0	24.0	12696	0.1684	nan
0.0	25.0	13225	0.1684	nan
0.0	26.0	13754	0.1684	nan
0.0	27.0	14283	0.1684	nan
0.0	28.0	14812	0.1684	nan
0.0	29.0	15341	0.1684	nan
0.0	30.0	15870	0.1684	nan
0.0	31.0	16399	0.1684	nan
0.0	32.0	16928	0.1684	nan
0.0	33.0	17457	0.1684	nan
0.0	34.0	17986	0.1684	nan
0.0	35.0	18515	0.1684	nan
0.0	36.0	19044	0.1684	nan
0.0	37.0	19573	0.1684	nan
0.0	38.0	20102	0.1684	nan
0.0	39.0	20631	0.1684	nan
0.0	40.0	21160	0.1684	nan
0.0	41.0	21689	0.1684	nan
0.0	42.0	22218	0.1684	nan
0.0	43.0	22747	0.1684	nan
0.0	44.0	23276	0.1684	nan
0.0	45.0	23805	0.1684	nan
0.0	46.0	24334	0.1684	nan
0.0	47.0	24863	0.1684	nan
0.0	48.0	25392	0.1684	nan
0.0	49.0	25921	nan	0.1684
0.0	50.0	26450	nan	0.1684

Framework versions

• Transformers 4.34.0
• Pytorch 2.1.0+cu121
• Datasets 2.18.0
• Tokenizers 0.14.1