metadata
base_model: meta-llama/Meta-Llama-3.1-8B
datasets:
- llama-duo/synth_closed_qa_dataset_dedup
library_name: peft
license: llama3.1
tags:
- alignment-handbook
- trl
- sft
- generated_from_trainer
model-index:
- name: llama3.1-8b-closedqa-gpt4o-100k
results: []
llama3.1-8b-closedqa-gpt4o-100k
This model is a fine-tuned version of meta-llama/Meta-Llama-3.1-8B on the llama-duo/synth_closed_qa_dataset_dedup dataset. It achieves the following results on the evaluation set:
- Loss: 3.8424
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: 4
- eval_batch_size: 2
- seed: 42
- distributed_type: multi-GPU
- num_devices: 4
- gradient_accumulation_steps: 2
- total_train_batch_size: 32
- total_eval_batch_size: 8
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 10
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 0.9422 | 0.9991 | 582 | 1.9762 |
| 0.8939 | 2.0 | 1165 | 2.0232 |
| 0.8255 | 2.9991 | 1747 | 2.1086 |
| 0.7584 | 4.0 | 2330 | 2.2541 |
| 0.6928 | 4.9991 | 2912 | 2.4424 |
| 0.6102 | 6.0 | 3495 | 2.7089 |
| 0.5466 | 6.9991 | 4077 | 3.0554 |
| 0.5038 | 8.0 | 4660 | 3.4053 |
| 0.4624 | 8.9991 | 5242 | 3.6952 |
| 0.454 | 9.9914 | 5820 | 3.8424 |
Framework versions
- PEFT 0.12.0
- Transformers 4.44.0
- Pytorch 2.4.0+cu121
- Datasets 2.20.0
- Tokenizers 0.19.1