This model is a fine-tuned version of Qwen/Qwen3-8B on the all_julia_snippets_format dataset. It achieves the following results on the evaluation set:

Loss: 0.6342

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: 0.0001
train_batch_size: 1
eval_batch_size: 2
seed: 42
distributed_type: multi-GPU
num_devices: 4
gradient_accumulation_steps: 8
total_train_batch_size: 32
total_eval_batch_size: 8
optimizer: Use adamw_torch_fused with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: cosine
lr_scheduler_warmup_ratio: 0.1
num_epochs: 3

Training results

Training Loss	Epoch	Step	Validation Loss
0.7947	0.1113	2000	0.7638
0.7582	0.2226	4000	0.7252
0.7321	0.3339	6000	0.7043
0.7156	0.4452	8000	0.6903
0.7136	0.5565	10000	0.6801
0.6989	0.6678	12000	0.6719
0.6944	0.7791	14000	0.6651
0.6901	0.8904	16000	0.6598
0.6779	1.0017	18000	0.6556
0.6439	1.1130	20000	0.6538
0.645	1.2243	22000	0.6504
0.6387	1.3356	24000	0.6478
0.6005	1.4469	26000	0.6468
0.5976	1.5582	28000	0.6485
0.5934	1.6696	30000	0.6541
0.5978	1.7809	32000	0.6574
0.5965	1.8922	34000	0.6550
0.5961	2.0035	36000	0.6563
0.592	2.1148	38000	0.6541
0.5897	2.2261	40000	0.6513
0.5949	2.3374	42000	0.6498
0.585	2.4487	44000	0.6457
0.5913	2.5600	46000	0.6446
0.5835	2.6713	48000	0.6416
0.5827	2.7826	50000	0.6395
0.5901	2.8939	52000	0.6364

Framework versions

PEFT 0.17.1
Transformers 4.56.2
Pytorch 2.8.0+cu128
Datasets 4.0.0
Tokenizers 0.22.1

CorgiPudding/Qwen3-8B-Julia

sft

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

CorgiPudding/
Qwen3-8B-Julia