codeparrot-ds-small

This model is a fine-tuned version of gpt2 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 3.2919

Model description

Due to hardware limitations and an insufficient amount of training data, the model has only been trained for a single epoch and is currently not functional.

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: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 2
• total_train_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: cosine
• lr_scheduler_warmup_steps: 100
• num_epochs: 1
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
5.9286	0.02	500	5.7197
5.2717	0.04	1000	5.1488
4.9972	0.06	1500	4.8345
4.8214	0.08	2000	4.6251
4.6272	0.1	2500	4.4807
4.5241	0.12	3000	4.3484
4.405	0.14	3500	4.2483
4.3189	0.16	4000	4.1680
4.2596	0.18	4500	4.0914
4.2569	0.2	5000	4.0341
4.1614	0.22	5500	3.9615
4.1073	0.24	6000	3.9112
4.0892	0.26	6500	3.8685
4.0151	0.28	7000	3.8277
3.903	0.3	7500	3.7787
3.9248	0.32	8000	3.7447
3.8978	0.34	8500	3.7189
3.9231	0.36	9000	3.6877
3.8936	0.38	9500	3.6479
3.7649	0.4	10000	3.6154
3.8156	0.42	10500	3.6069
3.7588	0.44	11000	3.5772
3.7559	0.46	11500	3.5517
3.814	0.48	12000	3.5230
3.7384	0.5	12500	3.5065
3.6827	0.52	13000	3.4807
3.6679	0.54	13500	3.4585
3.6838	0.56	14000	3.4419
3.7154	0.58	14500	3.4313
3.7117	0.6	15000	3.4156
3.6065	0.62	15500	3.3990
3.613	0.64	16000	3.3820
3.5824	0.66	16500	3.3702
3.6263	0.68	17000	3.3645
3.6073	0.7	17500	3.3529
3.587	0.72	18000	3.3419
3.624	0.74	18500	3.3340
3.583	0.76	19000	3.3273
3.573	0.78	19500	3.3185
3.5576	0.8	20000	3.3123
3.5623	0.82	20500	3.3103
3.5955	0.84	21000	3.3053
3.5947	0.86	21500	3.3015
3.5258	0.88	22000	3.2989
3.5985	0.9	22500	3.2962
3.5723	0.92	23000	3.2938
3.5863	0.94	23500	3.2927
3.5378	0.96	24000	3.2921
3.4952	0.98	24500	3.2919
3.5357	1.0	25000	3.2919

Framework versions

• Transformers 4.51.3
• Pytorch 2.5.1
• Datasets 2.19.1
• Tokenizers 0.21.1