metadata
language:
- pt
license: mit
library_name: peft
tags:
- generated_from_trainer
base_model: HuggingFaceH4/zephyr-7b-beta
datasets:
- heloisy/cosmos_qa_ptbr_text
model-index:
- name: zephyr-7b-cosmosqa-ptbr_text
results: []
zephyr-7b-cosmosqa-ptbr_text
This model is a fine-tuned version of HuggingFaceH4/zephyr-7b-beta on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 1.0951
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: 4e-05
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 32
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 100
- training_steps: 1000
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 1.4903 | 0.13 | 100 | 1.1563 |
| 1.1453 | 0.25 | 200 | 1.1254 |
| 1.1192 | 0.38 | 300 | 1.1149 |
| 1.109 | 0.51 | 400 | 1.1045 |
| 1.0927 | 0.63 | 500 | 1.1003 |
| 1.0883 | 0.76 | 600 | 1.0971 |
| 1.077 | 0.89 | 700 | 1.0955 |
| 1.0694 | 1.01 | 800 | 1.0934 |
| 1.0377 | 1.14 | 900 | 1.0958 |
| 1.03 | 1.27 | 1000 | 1.0951 |
Framework versions
- PEFT 0.7.1.dev0
- Transformers 4.36.0.dev0
- Pytorch 2.0.1+cu117
- Datasets 2.14.7
- Tokenizers 0.15.0
Open Portuguese LLM Leaderboard Evaluation Results
Detailed results can be found here and on the 🚀 Open Portuguese LLM Leaderboard
| Metric | Value |
|---|---|
| Average | 63.16 |
| ENEM Challenge (No Images) | 57.80 |
| BLUEX (No Images) | 47.15 |
| OAB Exams | 38.95 |
| Assin2 RTE | 89.50 |
| Assin2 STS | 68.47 |
| FaQuAD NLI | 63.08 |
| HateBR Binary | 78.43 |
| PT Hate Speech Binary | 68.05 |
| tweetSentBR | 57.01 |