llama2-7b-journal-finetune

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: 0.7325

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: 2.5e-05
• train_batch_size: 2
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 1
• training_steps: 1500

Training results

Training Loss	Epoch	Step	Validation Loss
No log	0.01	50	0.5191
No log	0.03	100	0.4746
No log	0.04	150	0.4858
No log	0.06	200	0.5082
No log	0.07	250	0.4972
No log	0.09	300	0.5109
No log	0.1	350	0.5296
No log	0.12	400	0.5666
No log	0.13	450	0.5548
0.2374	0.15	500	0.5662
0.2374	0.16	550	0.5943
0.2374	0.18	600	0.5817
0.2374	0.19	650	0.6102
0.2374	0.2	700	0.5825
0.2374	0.22	750	0.5742
0.2374	0.23	800	0.6160
0.2374	0.25	850	0.6234
0.2374	0.26	900	0.6419
0.2374	0.28	950	0.6401
0.0698	0.29	1000	0.6395
0.0698	0.31	1050	0.6549
0.0698	0.32	1100	0.6706
0.0698	0.34	1150	0.6914
0.0698	0.35	1200	0.7024
0.0698	0.37	1250	0.7284
0.0698	0.38	1300	0.7114
0.0698	0.4	1350	0.7215
0.0698	0.41	1400	0.7304
0.0698	0.42	1450	0.7315
0.0479	0.44	1500	0.7325

Framework versions

• PEFT 0.10.1.dev0
• Transformers 4.40.0.dev0
• Pytorch 2.1.2
• Datasets 2.16.0
• Tokenizers 0.15.2