---
library_name: peft
tags:
- trl
- sft
- generated_from_trainer
datasets:
- dialogstudio
base_model: NousResearch/Llama-2-7b-hf
model-index:
- name: llama2-7bb-tweet-summarization-gradnorm-0.3
  results: []
---

<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->

# llama2-7bb-tweet-summarization-gradnorm-0.3

This model is a fine-tuned version of [NousResearch/Llama-2-7b-hf](https://huggingface.co/NousResearch/Llama-2-7b-hf) on the dialogstudio dataset.
It achieves the following results on the evaluation set:
- Loss: 2.8160
- Rouge Scores: {'rouge1': 93.719779910895, 'rouge2': 78.0799701185797, 'rougeL': 64.91384075272471, 'rougeLsum': 93.71249369436103}
- Bleu Scores: [0.9468715981421053, 0.9340571158071639, 0.906767913949756, 0.8753561378232885]
- Gen Len: 463.0182

## 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: 4
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- num_epochs: 7
- mixed_precision_training: Native AMP

### Training results

| Training Loss | Epoch | Step | Validation Loss | Rouge Scores                                                                                                            | Bleu Scores                                                                      | Gen Len  |
|:-------------:|:-----:|:----:|:---------------:|:-----------------------------------------------------------------------------------------------------------------------:|:--------------------------------------------------------------------------------:|:--------:|
| 1.9246        | 1.0   | 220  | 1.8384          | {'rouge1': 92.78080137059182, 'rouge2': 78.71532643138437, 'rougeL': 68.0616149947273, 'rougeLsum': 92.78702835703021}  | [0.9079275318266272, 0.8970741286020552, 0.8736002135507472, 0.8466150307526832] | 463.0182 |
| 1.6564        | 2.0   | 440  | 1.8335          | {'rouge1': 93.62527163754612, 'rouge2': 79.14899366889107, 'rougeL': 68.02122989340602, 'rougeLsum': 93.62676386700348} | [0.9282164809556785, 0.9171615801879893, 0.892709310950969, 0.8645188775345913]  | 463.0182 |
| 1.3403        | 3.0   | 660  | 1.9481          | {'rouge1': 93.70688850262614, 'rouge2': 78.96026100012381, 'rougeL': 67.37638965440908, 'rougeLsum': 93.70399692691778} | [0.9342903619020663, 0.9225682522334384, 0.8972845918789121, 0.8681853449069523] | 463.0182 |
| 0.9984        | 4.0   | 880  | 2.1537          | {'rouge1': 93.77800041953847, 'rouge2': 78.72204799373465, 'rougeL': 66.56763131340682, 'rougeLsum': 93.77100407824561} | [0.9425931953005738, 0.9302863494509406, 0.9040669212466305, 0.8739193334758137] | 463.0182 |
| 0.7           | 5.0   | 1100 | 2.3692          | {'rouge1': 93.74639046979189, 'rouge2': 78.51569240275262, 'rougeL': 65.93032986525995, 'rougeLsum': 93.73745084400457} | [0.9440175755443134, 0.93171453625075, 0.9052208696375351, 0.8747208115562404]   | 463.0182 |
| 0.4947        | 6.0   | 1320 | 2.6590          | {'rouge1': 93.75661844384149, 'rouge2': 78.18805763398609, 'rougeL': 65.29243896759789, 'rougeLsum': 93.75034348574664} | [0.9470358425741272, 0.9342995624545122, 0.9070823690393129, 0.8757451333358709] | 463.0182 |
| 0.3922        | 7.0   | 1540 | 2.8160          | {'rouge1': 93.719779910895, 'rouge2': 78.0799701185797, 'rougeL': 64.91384075272471, 'rougeLsum': 93.71249369436103}    | [0.9468715981421053, 0.9340571158071639, 0.906767913949756, 0.8753561378232885]  | 463.0182 |


### Framework versions

- PEFT 0.8.2.dev0
- Transformers 4.38.0.dev0
- Pytorch 2.1.0+cu121
- Datasets 2.16.2.dev0
- Tokenizers 0.15.1