You can deploy this model and start earning money today!

zh27/
sft-datajudge-v4-confident-qwen3-8b-lora

Name: sft-datajudge-v4-confident-qwen3-8b-lora
Availability: InStock

Views

No views yet

scorer_binary_v4_confident_qwen3_8b_lora_e3

This model is a fine-tuned version of Qwen/Qwen3-8B on the scorer_binary_v4_confident_train dataset. It achieves the following results on the evaluation set:

Loss: 0.0493

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: 1
seed: 42
gradient_accumulation_steps: 16
total_train_batch_size: 16
optimizer: Use OptimizerNames.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_steps: 20
num_epochs: 3.0

Training results

Training Loss	Epoch	Step	Validation Loss
0.0639	0.1800	50	0.0579
0.0585	0.3600	100	0.0548
0.0498	0.5401	150	0.0537
0.0501	0.7201	200	0.0514
0.0534	0.9001	250	0.0525
0.0467	1.0792	300	0.0559
0.0337	1.2592	350	0.0505
0.0475	1.4392	400	0.0514
0.0430	1.6193	450	0.0497
0.0486	1.7993	500	0.0493
0.0418	1.9793	550	0.0495
0.0398	2.1584	600	0.0508
0.0262	2.3384	650	0.0563
0.0280	2.5185	700	0.0544
0.0271	2.6985	750	0.0543
0.0422	2.8785	800	0.0540

Framework versions

PEFT 0.18.1
Transformers 5.2.0
Pytorch 2.11.0+cu128
Datasets 4.0.0
Tokenizers 0.22.2

zh27/sft-datajudge-v4-confident-qwen3-8b-lora

scorer_binary_v4_confident_qwen3_8b_lora_e3

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

zh27/
sft-datajudge-v4-confident-qwen3-8b-lora