lemexp-task1-v3-lemma_object_small_nodefs-Llama-3.2-1B-8lr-12epochs-no-eos

This model is a fine-tuned version of meta-llama/Llama-3.2-1B on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.3114

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: 0.0008
• train_batch_size: 1
• eval_batch_size: 1
• seed: 42
• distributed_type: multi-GPU
• num_devices: 8
• gradient_accumulation_steps: 2
• total_train_batch_size: 16
• total_eval_batch_size: 8
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 12
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
0.8393	0.2001	720	0.7144
0.6912	0.4002	1440	0.6475
0.6189	0.6003	2160	0.6094
0.5896	0.8003	2880	0.5918
0.566	1.0003	3600	0.5725
0.5303	1.2004	4320	0.5583
0.5265	1.4004	5040	0.5440
0.5189	1.6005	5760	0.5196
0.5122	1.8006	6480	0.5208
0.503	2.0006	7200	0.5195
0.4854	2.2006	7920	0.5059
0.4775	2.4007	8640	0.5010
0.4629	2.6008	9360	0.4847
0.4676	2.8009	10080	0.4837
0.4644	3.0008	10800	0.4814
0.4412	3.2009	11520	0.4813
0.4415	3.4010	12240	0.4584
0.4331	3.6011	12960	0.4591
0.4299	3.8012	13680	0.4594
0.4315	4.0011	14400	0.4487
0.4049	4.2012	15120	0.4382
0.4108	4.4013	15840	0.4438
0.4016	4.6014	16560	0.4380
0.4065	4.8014	17280	0.4301
0.3934	5.0014	18000	0.4301
0.3823	5.2015	18720	0.4155
0.3779	5.4016	19440	0.4208
0.3752	5.6016	20160	0.4183
0.3787	5.8017	20880	0.4027
0.3655	6.0017	21600	0.4029
0.3512	6.2018	22320	0.3936
0.3546	6.4018	23040	0.3980
0.3527	6.6019	23760	0.3936
0.3461	6.8020	24480	0.3854
0.3427	7.0019	25200	0.3776
0.3243	7.2020	25920	0.3837
0.3238	7.4021	26640	0.3760
0.3231	7.6022	27360	0.3704
0.3176	7.8023	28080	0.3619
0.3246	8.0022	28800	0.3671
0.2978	8.2023	29520	0.3680
0.2919	8.4024	30240	0.3621
0.2983	8.6025	30960	0.3530
0.2909	8.8026	31680	0.3471
0.2911	9.0025	32400	0.3455
0.2692	9.2026	33120	0.3394
0.2692	9.4027	33840	0.3401
0.2731	9.6028	34560	0.3342
0.271	9.8028	35280	0.3307
0.257	10.0028	36000	0.3329
0.2408	10.2029	36720	0.3315
0.2376	10.4029	37440	0.3216
0.2404	10.6030	38160	0.3234
0.2371	10.8031	38880	0.3175
0.2418	11.0031	39600	0.3149
0.2229	11.2031	40320	0.3170
0.2201	11.4032	41040	0.3127
0.2165	11.6033	41760	0.3117
0.2161	11.8034	42480	0.3114

Framework versions

• PEFT 0.15.2
• Transformers 4.57.1
• Pytorch 2.7.0+cu126
• Datasets 4.3.0
• Tokenizers 0.22.1