Tweet_sentiment

This model is a fine-tuned version of TheBloke/Llama-2-7B-Chat-GPTQ on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.6657

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: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 16
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.03
• num_epochs: 1
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
5.422	0.0457	2	5.1721
5.1534	0.0914	4	4.6225
4.5268	0.1371	6	4.0415
3.9455	0.1829	8	3.5294
3.4478	0.2286	10	3.1964
3.1897	0.2743	12	2.9354
2.8951	0.32	14	2.6721
2.5987	0.3657	16	2.4219
2.3629	0.4114	18	2.2606
2.1553	0.4571	20	2.1368
2.109	0.5029	22	2.0289
1.9611	0.5486	24	1.9382
1.8992	0.5943	26	1.8638
1.794	0.64	28	1.8066
1.7374	0.6857	30	1.7621
1.7599	0.7314	32	1.7268
1.769	0.7771	34	1.7006
1.6442	0.8229	36	1.6834
1.6242	0.8686	38	1.6730
1.7367	0.9143	40	1.6677
1.6649	0.96	42	1.6657

Framework versions

• PEFT 0.10.0
• Transformers 4.40.2
• Pytorch 2.1.0+cu121
• Datasets 2.19.1
• Tokenizers 0.19.1