You can deploy this model and start earning money today!

ahmed-ouka/
lora-qwen-eniad

Name: lora-qwen-eniad
Availability: InStock

generated_from_trainer

Qwen/Qwen2.5-1.5B-Instruct

adapter

apache-2.0

Views

No views yet

model

This model is a fine-tuned version of Qwen/Qwen2.5-1.5B-Instruct on the eniad-Assistant_finetune_train dataset. It achieves the following results on the evaluation set:

Loss: 0.1725

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: 4
total_train_batch_size: 4
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_ratio: 0.1
num_epochs: 3.0

Training results

Training Loss	Epoch	Step	Validation Loss
0.5547	0.1	100	0.6038
0.4868	0.2	200	0.4695
0.3905	0.3	300	0.3936
0.3247	0.4	400	0.3415
0.3518	0.5	500	0.3102
0.2619	0.6	600	0.2942
0.281	0.7	700	0.2660
0.263	0.8	800	0.2504
0.2754	0.9	900	0.2386
0.2543	1.0	1000	0.2262
0.1792	1.1	1100	0.2293
0.1864	1.2	1200	0.2134
0.1701	1.3	1300	0.2083
0.1811	1.4	1400	0.2060
0.1629	1.5	1500	0.1994
0.1873	1.6	1600	0.1953
0.157	1.7	1700	0.1944
0.1532	1.8	1800	0.1856
0.1491	1.9	1900	0.1875
0.1392	2.0	2000	0.1786
0.0951	2.1	2100	0.1798
0.0922	2.2	2200	0.1827
0.0942	2.3	2300	0.1772
0.0847	2.4	2400	0.1756
0.0781	2.5	2500	0.1762
0.0806	2.6	2600	0.1737
0.0812	2.7	2700	0.1738
0.0979	2.8	2800	0.1722
0.0697	2.9	2900	0.1723
0.0941	3.0	3000	0.1725

Framework versions

PEFT 0.15.1
Transformers 4.51.3
Pytorch 2.6.0+cu124
Datasets 3.5.0
Tokenizers 0.21.1

ahmed-ouka/lora-qwen-eniad

model

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

ahmed-ouka/
lora-qwen-eniad