llama_2_gsm8k_midset_cot_simplest

This model is a fine-tuned version of meta-llama/Llama-2-7b-chat-hf on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.5332

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
• distributed_type: multi-GPU
• gradient_accumulation_steps: 2
• total_train_batch_size: 8
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.1
• training_steps: 200

Training results

Training Loss	Epoch	Step	Validation Loss
1.538	0.04	5	1.5284
1.2607	0.08	10	1.1122
1.0013	0.12	15	0.9311
0.851	0.16	20	0.7617
0.6951	0.2	25	0.6774
0.6362	0.24	30	0.6290
0.6572	0.28	35	0.6103
0.6693	0.32	40	0.5990
0.637	0.36	45	0.5900
0.5662	0.4	50	0.5839
0.6003	0.44	55	0.5782
0.6152	0.48	60	0.5736
0.6012	0.52	65	0.5686
0.5044	0.56	70	0.5696
0.5725	0.6	75	0.5627
0.5592	0.64	80	0.5576
0.5517	0.68	85	0.5548
0.5897	0.72	90	0.5505
0.5353	0.76	95	0.5502
0.5627	0.8	100	0.5454
0.533	0.84	105	0.5426
0.5318	0.88	110	0.5399
0.5323	0.92	115	0.5384
0.5694	0.96	120	0.5378
0.5274	1.0	125	0.5373
0.4683	1.04	130	0.5364
0.5426	1.08	135	0.5358
0.4448	1.12	140	0.5368
0.4947	1.16	145	0.5383
0.4569	1.2	150	0.5388
0.5309	1.24	155	0.5377
0.4978	1.28	160	0.5359
0.4732	1.32	165	0.5350
0.5057	1.3600	170	0.5341
0.487	1.4	175	0.5336
0.4773	1.44	180	0.5334
0.4748	1.48	185	0.5330
0.4889	1.52	190	0.5331
0.4785	1.56	195	0.5331
0.4677	1.6	200	0.5332

Framework versions

• PEFT 0.12.0
• Transformers 4.44.2
• Pytorch 2.4.1+cu121
• Datasets 3.0.0
• Tokenizers 0.19.1