metadata
license: apache-2.0
base_model: mnoukhov/pythia160m-sft-tldr
tags:
- trl
- reward-trainer
- generated_from_trainer
metrics:
- accuracy
model-index:
- name: pythia160m-rm-tldr6.9b
results: []
pythia160m-rm-tldr6.9b
This model is a fine-tuned version of mnoukhov/pythia160m-sft-tldr on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.5486
- Accuracy: 0.7121
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: 8
- eval_batch_size: 8
- seed: 42
- distributed_type: multi-GPU
- num_devices: 4
- gradient_accumulation_steps: 8
- total_train_batch_size: 256
- total_eval_batch_size: 32
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- num_epochs: 1
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|---|---|---|---|---|
| No log | 0.2012 | 73 | 0.5667 | 0.6998 |
| 0.6204 | 0.4025 | 146 | 0.5560 | 0.7120 |
| 0.5593 | 0.6037 | 219 | 0.5532 | 0.7078 |
| 0.5593 | 0.8050 | 292 | 0.5486 | 0.7121 |
Framework versions
- Transformers 4.41.1
- Pytorch 2.2.1+cu121
- Datasets 2.19.0
- Tokenizers 0.19.1