telugu-colloquial-translator

This model is a fine-tuned version of unsloth/tinyllama-chat-bnb-4bit on the None dataset. It achieves the following results on the evaluation set:

• Loss: 9.2129

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: 0.0003
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 8
• 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: linear
• lr_scheduler_warmup_steps: 2
• num_epochs: 10
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
14.6163	0.2667	2	9.2468
14.5969	0.5333	4	9.2468
14.676	0.8	6	9.2468
14.5883	1.0	8	9.2468
14.6246	1.2667	10	9.2468
14.5945	1.5333	12	9.2468
14.6381	1.8	14	9.2468
14.5547	2.0	16	9.2468
12.8615	2.2667	18	9.2478
9.3963	2.5333	20	9.2486
13.8496	2.8	22	9.2491
9.2063	3.0	24	9.2487
6.4105	3.2667	26	9.2479
5.6432	3.5333	28	9.2472
5.2603	3.8	30	9.2459
4.8088	4.0	32	9.2440
4.5624	4.2667	34	9.2414
4.313	4.5333	36	9.2390
4.1646	4.8	38	9.2368
4.0837	5.0	40	9.2344
3.9956	5.2667	42	9.2319
3.9756	5.5333	44	9.2298
3.8652	5.8	46	9.2285
3.9191	6.0	48	9.2272
3.8893	6.2667	50	9.2251
3.8461	6.5333	52	9.2231
3.7788	6.8	54	9.2219
3.8249	7.0	56	9.2203
3.7855	7.2667	58	9.2183
3.7378	7.5333	60	9.2165
3.7678	7.8	62	9.2157
3.7418	8.0	64	9.2151
3.726	8.2667	66	9.2143
3.6778	8.5333	68	9.2134
3.7152	8.8	70	9.2129

Framework versions

• PEFT 0.14.0
• Transformers 4.48.3
• Pytorch 2.6.0+cu124
• Datasets 3.3.1
• Tokenizers 0.21.0