deepseek-coder-arc-agi-finetuned

This model is a fine-tuned version of deepseek-ai/deepseek-coder-6.7b-base on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.2464

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: 2
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 16
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 100
• num_epochs: 4

Training results

Training Loss	Epoch	Step	Validation Loss
0.3729	0.4124	25	0.3884
0.2512	0.8247	50	0.2950
0.2074	1.2474	75	0.2809
0.2416	1.6598	100	0.2723
0.2966	2.0825	125	0.2663
0.1781	2.4948	150	0.2596
0.2821	2.9072	175	0.2534
0.1999	3.3299	200	0.2501
0.1572	3.7423	225	0.2464

Framework versions

• PEFT 0.17.1
• Transformers 4.51.3
• Pytorch 2.7.0+cu126
• Datasets 4.1.1
• Tokenizers 0.21.1