README.md

---
license: apache-2.0
language:
- en
- es
datasets:
- NickyNicky/oasst2_orpo_mix_tokenizer_phi_3_v1
---


![image/png](https://cdn-uploads.huggingface.co/production/uploads/641b435ba5f876fe30c5ae0a/Dr2QTdUfXKKxvbaARNerT.png)


![image/png](https://cdn-uploads.huggingface.co/production/uploads/641b435ba5f876fe30c5ae0a/YiH8B9QpGTvEd81Q5Y8sq.png)


![image/png](https://cdn-uploads.huggingface.co/production/uploads/641b435ba5f876fe30c5ae0a/7iUh2GlSeylJ21CEtbQlR.png)


![image/png](https://cdn-uploads.huggingface.co/production/uploads/641b435ba5f876fe30c5ae0a/Rzax5SOedYpI5O9vfcXJV.png)

## Metrics
```
TrainOutput(
  global_step=1526,
  training_loss=0.40326238030062433,
  metrics={
    'train_runtime': 129566.5492,
    'train_samples_per_second': 0.848,
    'train_steps_per_second': 0.012,
    'total_flos': 0.0,
    'train_loss': 0.40326238030062433,
    'epoch': 2.023872679045093
  }
)

max_seq_length= 4096
```


## colab examples.
```
model_id= "NickyNicky/Phi-3-mini-4k-instruct_orpo_V2"

https://colab.research.google.com/drive/16qS7NMSu20LzcwvYCrBGVI7rd9Hr-vpN?usp=sharing
```