Training procedure
The following bitsandbytes quantization config was used during training:
- quant_method: bitsandbytes
- _load_in_8bit: False
- _load_in_4bit: True
- llm_int8_threshold: 6.0
- llm_int8_skip_modules: None
- llm_int8_enable_fp32_cpu_offload: False
- llm_int8_has_fp16_weight: False
- bnb_4bit_quant_type: nf4
- bnb_4bit_use_double_quant: True
- bnb_4bit_compute_dtype: float16
- bnb_4bit_quant_storage: uint8
- load_in_4bit: True
- load_in_8bit: False
Framework versions
- PEFT 0.5.0
Open Portuguese LLM Leaderboard Evaluation Results
Detailed results can be found here and on the 🚀 Open Portuguese LLM Leaderboard
| Metric | Value |
|---|---|
| Average | 65.16 |
| ENEM Challenge (No Images) | 57.03 |
| BLUEX (No Images) | 44.92 |
| OAB Exams | 39.64 |
| Assin2 RTE | 90.68 |
| Assin2 STS | 69.97 |
| FaQuAD NLI | 65.14 |
| HateBR Binary | 83.25 |
| PT Hate Speech Binary | 70.36 |
| tweetSentBR | 65.45 |
- Downloads last month
- 15
Inference Providers
NEW
This model is not currently available via any of the supported Inference Providers.
The model cannot be deployed to the HF Inference API:
The model has no pipeline_tag.
Space using recogna-nlp/zephyr_7b_beta_ultraalpaca 1
Evaluation results
- accuracy on ENEM Challenge (No Images)Open Portuguese LLM Leaderboard57.030
- accuracy on BLUEX (No Images)Open Portuguese LLM Leaderboard44.920
- accuracy on OAB ExamsOpen Portuguese LLM Leaderboard39.640
- f1-macro on Assin2 RTEtest set Open Portuguese LLM Leaderboard90.680
- pearson on Assin2 STStest set Open Portuguese LLM Leaderboard69.970
- f1-macro on FaQuAD NLItest set Open Portuguese LLM Leaderboard65.140
- f1-macro on HateBR Binarytest set Open Portuguese LLM Leaderboard83.250
- f1-macro on PT Hate Speech Binarytest set Open Portuguese LLM Leaderboard70.360
- f1-macro on tweetSentBRtest set Open Portuguese LLM Leaderboard65.450