---
language:
- pt
license: apache-2.0
library_name: transformers
tags:
- text-generation-inference
- TensorBlock
- GGUF
datasets:
- nicholasKluge/instruct-aira-dataset-v3
- cnmoro/GPT4-500k-Augmented-PTBR-Clean
- rhaymison/orca-math-portuguese-64k
- nicholasKluge/reward-aira-dataset
metrics:
- perplexity
pipeline_tag: text-generation
widget:
- text: <instruction>Cite algumas bandas de rock brasileiras famosas.</instruction>
  example_title: Exemplo
- text: <instruction>Invente uma história sobre um encanador com poderes mágicos.</instruction>
  example_title: Exemplo
- text: <instruction>Qual cidade é a capital do estado do Rio Grande do Sul?</instruction>
  example_title: Exemplo
- text: <instruction>Diga o nome de uma maravilha culinária característica da cosinha
    Portuguesa?</instruction>
  example_title: Exemplo
inference:
  parameters:
    repetition_penalty: 1.2
    temperature: 0.2
    top_k: 20
    top_p: 0.2
    max_new_tokens: 150
co2_eq_emissions:
  emissions: 42270
  source: CodeCarbon
  training_type: pre-training
  geographical_location: Germany
  hardware_used: NVIDIA A100-SXM4-80GB
base_model: TucanoBR/Tucano-2b4-Instruct
model-index:
- name: Tucano-2b4-Instruct
  results:
  - task:
      type: text-generation
      name: Text Generation
    dataset:
      name: CALAME-PT
      type: NOVA-vision-language/calame-pt
      split: all
      args:
        num_few_shot: 0
    metrics:
    - type: acc
      value: 57.66
      name: accuracy
    source:
      url: https://huggingface.co/datasets/NOVA-vision-language/calame-pt
      name: Context-Aware LAnguage Modeling Evaluation for Portuguese
  - task:
      type: text-generation
      name: Text Generation
    dataset:
      name: LAMBADA-PT
      type: TucanoBR/lambada-pt
      split: train
      args:
        num_few_shot: 0
    metrics:
    - type: acc
      value: 39.92
      name: accuracy
    source:
      url: https://huggingface.co/datasets/TucanoBR/lambada-pt
      name: LAMBADA-PT
  - task:
      type: text-generation
      name: Text Generation
    dataset:
      name: ENEM Challenge (No Images)
      type: eduagarcia/enem_challenge
      split: train
      args:
        num_few_shot: 3
    metrics:
    - type: acc
      value: 20.43
      name: accuracy
    source:
      url: https://huggingface.co/spaces/eduagarcia/open_pt_llm_leaderboard
      name: Open Portuguese LLM Leaderboard
  - task:
      type: text-generation
      name: Text Generation
    dataset:
      name: BLUEX (No Images)
      type: eduagarcia-temp/BLUEX_without_images
      split: train
      args:
        num_few_shot: 3
    metrics:
    - type: acc
      value: 22.81
      name: accuracy
    source:
      url: https://huggingface.co/spaces/eduagarcia/open_pt_llm_leaderboard
      name: Open Portuguese LLM Leaderboard
  - task:
      type: text-generation
      name: Text Generation
    dataset:
      name: OAB Exams
      type: eduagarcia/oab_exams
      split: train
      args:
        num_few_shot: 3
    metrics:
    - type: acc
      value: 24.83
      name: accuracy
    source:
      url: https://huggingface.co/spaces/eduagarcia/open_pt_llm_leaderboard
      name: Open Portuguese LLM Leaderboard
  - task:
      type: text-generation
      name: Text Generation
    dataset:
      name: Assin2 RTE
      type: assin2
      split: test
      args:
        num_few_shot: 15
    metrics:
    - type: f1_macro
      value: 43.39
      name: f1-macro
    source:
      url: https://huggingface.co/spaces/eduagarcia/open_pt_llm_leaderboard
      name: Open Portuguese LLM Leaderboard
  - task:
      type: text-generation
      name: Text Generation
    dataset:
      name: Assin2 STS
      type: eduagarcia/portuguese_benchmark
      split: test
      args:
        num_few_shot: 10
    metrics:
    - type: pearson
      value: 6.31
      name: pearson
    source:
      url: https://huggingface.co/spaces/eduagarcia/open_pt_llm_leaderboard
      name: Open Portuguese LLM Leaderboard
  - task:
      type: text-generation
      name: Text Generation
    dataset:
      name: FaQuAD NLI
      type: ruanchaves/faquad-nli
      split: test
      args:
        num_few_shot: 15
    metrics:
    - type: f1_macro
      value: 43.97
      name: f1-macro
    source:
      url: https://huggingface.co/spaces/eduagarcia/open_pt_llm_leaderboard
      name: Open Portuguese LLM Leaderboard
  - task:
      type: text-generation
      name: Text Generation
    dataset:
      name: HateBR Binary
      type: ruanchaves/hatebr
      split: test
      args:
        num_few_shot: 25
    metrics:
    - type: f1_macro
      value: 27.7
      name: f1-macro
    source:
      url: https://huggingface.co/spaces/eduagarcia/open_pt_llm_leaderboard
      name: Open Portuguese LLM Leaderboard
  - task:
      type: text-generation
      name: Text Generation
    dataset:
      name: PT Hate Speech Binary
      type: hate_speech_portuguese
      split: test
      args:
        num_few_shot: 25
    metrics:
    - type: f1_macro
      value: 29.18
      name: f1-macro
    source:
      url: https://huggingface.co/spaces/eduagarcia/open_pt_llm_leaderboard
      name: Open Portuguese LLM Leaderboard
  - task:
      type: text-generation
      name: Text Generation
    dataset:
      name: tweetSentBR
      type: eduagarcia-temp/tweetsentbr
      split: test
      args:
        num_few_shot: 25
    metrics:
    - type: f1_macro
      value: 43.11
      name: f1-macro
    source:
      url: https://huggingface.co/spaces/eduagarcia/open_pt_llm_leaderboard
      name: Open Portuguese LLM Leaderboard
  - task:
      type: text-generation
      name: Text Generation
    dataset:
      name: ARC-Challenge (PT)
      type: arc_pt
      args:
        num_few_shot: 25
    metrics:
    - type: acc_norm
      value: 32.05
      name: normalized accuracy
    source:
      url: https://github.com/nlp-uoregon/mlmm-evaluation
      name: Evaluation Framework for Multilingual Large Language Models
  - task:
      type: text-generation
      name: Text Generation
    dataset:
      name: HellaSwag (PT)
      type: hellaswag_pt
      args:
        num_few_shot: 10
    metrics:
    - type: acc_norm
      value: 48.28
      name: normalized accuracy
    source:
      url: https://github.com/nlp-uoregon/mlmm-evaluation
      name: Evaluation Framework for Multilingual Large Language Models
  - task:
      type: text-generation
      name: Text Generation
    dataset:
      name: TruthfulQA (PT)
      type: truthfulqa_pt
      args:
        num_few_shot: 0
    metrics:
    - type: mc2
      value: 38.44
      name: bleurt
    source:
      url: https://github.com/nlp-uoregon/mlmm-evaluation
      name: Evaluation Framework for Multilingual Large Language Models
  - task:
      type: text-generation
      name: Text Generation
    dataset:
      name: Alpaca-Eval (PT)
      type: alpaca_eval_pt
      args:
        num_few_shot: 0
    metrics:
    - type: lc_winrate
      value: 13.0
      name: length controlled winrate
    source:
      url: https://github.com/tatsu-lab/alpaca_eval
      name: AlpacaEval
---

<div style="width: auto; margin-left: auto; margin-right: auto">
<img src="https://i.imgur.com/jC7kdl8.jpeg" alt="TensorBlock" style="width: 100%; min-width: 400px; display: block; margin: auto;">
</div>
<div style="display: flex; justify-content: space-between; width: 100%;">
    <div style="display: flex; flex-direction: column; align-items: flex-start;">
        <p style="margin-top: 0.5em; margin-bottom: 0em;">
            Feedback and support: TensorBlock's  <a href="https://x.com/tensorblock_aoi">Twitter/X</a>, <a href="https://t.me/TensorBlock">Telegram Group</a> and <a href="https://x.com/tensorblock_aoi">Discord server</a>
        </p>
    </div>
</div>

## TucanoBR/Tucano-2b4-Instruct - GGUF

This repo contains GGUF format model files for [TucanoBR/Tucano-2b4-Instruct](https://huggingface.co/TucanoBR/Tucano-2b4-Instruct).

The files were quantized using machines provided by [TensorBlock](https://tensorblock.co/), and they are compatible with llama.cpp as of [commit b4242](https://github.com/ggerganov/llama.cpp/commit/a6744e43e80f4be6398fc7733a01642c846dce1d).

<div style="text-align: left; margin: 20px 0;">
    <a href="https://tensorblock.co/waitlist/client" style="display: inline-block; padding: 10px 20px; background-color: #007bff; color: white; text-decoration: none; border-radius: 5px; font-weight: bold;">
        Run them on the TensorBlock client using your local machine ↗
    </a>
</div>

## Prompt template

```
<instruction>{prompt}</instruction>
```

## Model file specification

| Filename | Quant type | File Size | Description |
| -------- | ---------- | --------- | ----------- |
| [Tucano-2b4-Instruct-Q2_K.gguf](https://huggingface.co/tensorblock/Tucano-2b4-Instruct-GGUF/blob/main/Tucano-2b4-Instruct-Q2_K.gguf) | Q2_K | 0.933 GB | smallest, significant quality loss - not recommended for most purposes |
| [Tucano-2b4-Instruct-Q3_K_S.gguf](https://huggingface.co/tensorblock/Tucano-2b4-Instruct-GGUF/blob/main/Tucano-2b4-Instruct-Q3_K_S.gguf) | Q3_K_S | 1.084 GB | very small, high quality loss |
| [Tucano-2b4-Instruct-Q3_K_M.gguf](https://huggingface.co/tensorblock/Tucano-2b4-Instruct-GGUF/blob/main/Tucano-2b4-Instruct-Q3_K_M.gguf) | Q3_K_M | 1.197 GB | very small, high quality loss |
| [Tucano-2b4-Instruct-Q3_K_L.gguf](https://huggingface.co/tensorblock/Tucano-2b4-Instruct-GGUF/blob/main/Tucano-2b4-Instruct-Q3_K_L.gguf) | Q3_K_L | 1.297 GB | small, substantial quality loss |
| [Tucano-2b4-Instruct-Q4_0.gguf](https://huggingface.co/tensorblock/Tucano-2b4-Instruct-GGUF/blob/main/Tucano-2b4-Instruct-Q4_0.gguf) | Q4_0 | 1.397 GB | legacy; small, very high quality loss - prefer using Q3_K_M |
| [Tucano-2b4-Instruct-Q4_K_S.gguf](https://huggingface.co/tensorblock/Tucano-2b4-Instruct-GGUF/blob/main/Tucano-2b4-Instruct-Q4_K_S.gguf) | Q4_K_S | 1.408 GB | small, greater quality loss |
| [Tucano-2b4-Instruct-Q4_K_M.gguf](https://huggingface.co/tensorblock/Tucano-2b4-Instruct-GGUF/blob/main/Tucano-2b4-Instruct-Q4_K_M.gguf) | Q4_K_M | 1.484 GB | medium, balanced quality - recommended |
| [Tucano-2b4-Instruct-Q5_0.gguf](https://huggingface.co/tensorblock/Tucano-2b4-Instruct-GGUF/blob/main/Tucano-2b4-Instruct-Q5_0.gguf) | Q5_0 | 1.693 GB | legacy; medium, balanced quality - prefer using Q4_K_M |
| [Tucano-2b4-Instruct-Q5_K_S.gguf](https://huggingface.co/tensorblock/Tucano-2b4-Instruct-GGUF/blob/main/Tucano-2b4-Instruct-Q5_K_S.gguf) | Q5_K_S | 1.693 GB | large, low quality loss - recommended |
| [Tucano-2b4-Instruct-Q5_K_M.gguf](https://huggingface.co/tensorblock/Tucano-2b4-Instruct-GGUF/blob/main/Tucano-2b4-Instruct-Q5_K_M.gguf) | Q5_K_M | 1.737 GB | large, very low quality loss - recommended |
| [Tucano-2b4-Instruct-Q6_K.gguf](https://huggingface.co/tensorblock/Tucano-2b4-Instruct-GGUF/blob/main/Tucano-2b4-Instruct-Q6_K.gguf) | Q6_K | 2.007 GB | very large, extremely low quality loss |
| [Tucano-2b4-Instruct-Q8_0.gguf](https://huggingface.co/tensorblock/Tucano-2b4-Instruct-GGUF/blob/main/Tucano-2b4-Instruct-Q8_0.gguf) | Q8_0 | 2.599 GB | very large, extremely low quality loss - not recommended |


## Downloading instruction

### Command line

Firstly, install Huggingface Client

```shell
pip install -U "huggingface_hub[cli]"
```

Then, downoad the individual model file the a local directory

```shell
huggingface-cli download tensorblock/Tucano-2b4-Instruct-GGUF --include "Tucano-2b4-Instruct-Q2_K.gguf" --local-dir MY_LOCAL_DIR
```

If you wanna download multiple model files with a pattern (e.g., `*Q4_K*gguf`), you can try:

```shell
huggingface-cli download tensorblock/Tucano-2b4-Instruct-GGUF --local-dir MY_LOCAL_DIR --local-dir-use-symlinks False --include='*Q4_K*gguf'
```