assistant_marocain_francais_Qwen05-08
This model is a fine-tuned version of Qwen/Qwen1.5-1.8B-Chat on the None dataset. It achieves the following results on the evaluation set:
- Loss: 5.3864
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: 1e-05
- train_batch_size: 2
- eval_batch_size: 2
- seed: 42
- gradient_accumulation_steps: 4
- 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: cosine
- lr_scheduler_warmup_ratio: 0.1
- lr_scheduler_warmup_steps: 100
- num_epochs: 10
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 5.212 | 0.7692 | 10 | 5.3864 |
| 5.2714 | 1.5385 | 20 | 5.3864 |
| 5.2862 | 2.3077 | 30 | 5.3864 |
| 5.1896 | 3.0769 | 40 | 5.3864 |
| 5.4079 | 3.8462 | 50 | 5.3864 |
| 5.1305 | 4.6154 | 60 | 5.3864 |
| 5.2895 | 5.3846 | 70 | 5.3864 |
| 5.187 | 6.1538 | 80 | 5.3864 |
| 5.3872 | 6.9231 | 90 | 5.3864 |
| 5.1401 | 7.6923 | 100 | 5.3864 |
| 5.2791 | 8.4615 | 110 | 5.3864 |
| 5.2922 | 9.2308 | 120 | 5.3864 |
| 5.1742 | 10.0 | 130 | 5.3864 |
Framework versions
- Transformers 4.53.0
- Pytorch 2.6.0+cu124
- Datasets 2.14.4
- Tokenizers 0.21.2
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support
Model tree for kawkaaaab/assistant_marocain_francais_Qwen05-08
Base model
Qwen/Qwen1.5-1.8B-Chat