You can deploy this model and start earning money today!

xummer/
mistral-7b-belebele-lora-ckb-arab

mistralai/Mistral-7B-Instruct-v0.3

other

Views

No views yet

belebele_ckb_Arab

This model is a fine-tuned version of mistralai/Mistral-7B-Instruct-v0.3 on the belebele_ckb_Arab_train dataset. It achieves the following results on the evaluation set:

Loss: 0.1180
Accuracy: 0.9565
Mcq Accuracy: 0.2317

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: 16
eval_batch_size: 16
seed: 42
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: 0.1
num_epochs: 5.0

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Mcq Accuracy
0.3429	1.1111	40	0.1088	0.9552	0.2195
0.0672	2.2222	80	0.1118	0.9568	0.2561
0.0578	3.3333	120	0.1078	0.9543	0.1951
0.0500	4.4444	160	0.1178	0.9571	0.2561

Framework versions

PEFT 0.18.1
Transformers 5.2.0
Pytorch 2.10.0+cu128
Datasets 4.0.0
Tokenizers 0.22.2

xummer/mistral-7b-belebele-lora-ckb-arab

belebele_ckb_Arab

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

xummer/
mistral-7b-belebele-lora-ckb-arab