engineer1-heavy-barc-llama3.1-8b-instruct-lora64-testtime-finetuning

This model is a fine-tuned version of barc0/Llama-3.1-ARC-Potpourri-Transduction-8B on the tttx/problem0_data, the barc0/transduction_formatted_rearc_dataset_100k and the barc0/transduction_heavy_100k_jsonl datasets. It achieves the following results on the evaluation set:

• Loss: 0.0000

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: 2
• seed: 42
• distributed_type: multi-GPU
• num_devices: 8
• gradient_accumulation_steps: 2
• total_train_batch_size: 32
• total_eval_batch_size: 16
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss
0.0016	1.0	1	0.0005
0.0014	2.0	2	0.0001
0.0001	3.0	3	0.0001
0.0001	4.0	4	0.0000
0.0	5.0	5	0.0000
0.0	6.0	6	0.0000
0.0	7.0	7	0.0000
0.0	8.0	8	0.0000
0.0	9.0	9	0.0000
0.0	10.0	10	0.0000

Framework versions

• PEFT 0.13.2
• Transformers 4.45.2
• Pytorch 2.4.0+cu121
• Datasets 3.0.1
• Tokenizers 0.20.1