metadata
base_model: NousResearch/Llama-2-7b-hf
tags:
- generated_from_trainer
- math
model-index:
- name: llama2-docsum-adapter
results: []
license: mit
datasets:
- camel-ai/math
language:
- en
metrics:
- bleurt
- bleu
- bertscore
pipeline_tag: text-generation
llama2-docsum-adapter
This model is a fine-tuned version of NousResearch/Llama-2-7b-hf on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.7568
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
- gradient_accumulation_steps: 4
- total_train_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_ratio: 0.05
- num_epochs: 2
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 1.0125 | 0.42 | 13 | 0.9909 |
| 0.8517 | 0.83 | 26 | 0.8135 |
| 0.7423 | 1.25 | 39 | 0.7766 |
| 0.5581 | 1.66 | 52 | 0.7568 |
Framework versions
- Transformers 4.35.2
- Pytorch 2.1.0+cu121
- Datasets 2.15.0
- Tokenizers 0.15.0