metadata

license: mit
library_name: peft
tags:
  - generated_from_trainer
base_model: HuggingFaceH4/zephyr-7b-beta
model-index:
  - name: zephyr-7b-b-cosmosqa-ptbr-text-v3
    results: []

zephyr-7b-b-cosmosqa-ptbr-text-v3

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: 1.0961

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: cosine
lr_scheduler_warmup_steps: 100
num_epochs: 2

Training results

Training Loss	Epoch	Step	Validation Loss
1.4897	0.13	100	1.1561
1.1452	0.25	200	1.1251
1.1185	0.38	300	1.1146
1.1079	0.51	400	1.1038
1.0906	0.63	500	1.0998
1.0849	0.76	600	1.0967
1.0721	0.89	700	1.0959
1.0623	1.01	800	1.0941
1.0217	1.14	900	1.0987
1.0122	1.27	1000	1.0977
1.0095	1.39	1100	1.0980
1.0088	1.52	1200	1.0966
0.9952	1.65	1300	1.0964
1.0018	1.77	1400	1.0967
0.9991	1.9	1500	1.0961

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	62.46
ENEM Challenge (No Images)	56.89
BLUEX (No Images)	45.76
OAB Exams	38.22
Assin2 RTE	89.05
Assin2 STS	67.49
FaQuAD NLI	64.55
HateBR Binary	77.14
PT Hate Speech Binary	67.20
tweetSentBR	55.87