---
language:
- pt
license: apache-2.0
task_categories:
- text-generation
- question-answering
- summarization
tags:
- vivamais
- minicpm5
- brazilian-portuguese
- sft
- qlora
- distilled
- synthetic
- redacted
- grounded-qa
size_categories:
- 1K<n<10K
pretty_name: MiniCPM5 Viva Mais text SFT v4
---

# MiniCPM5 Viva Mais text SFT v4

This dataset contains the redacted training and evaluation artifacts used for
`marinarosa/minicpm5-1b-vivamais-v4`. It was built for Viva Mais, a local-first Portuguese WhatsApp
travel-agency copilot that answers grounded questions from an extracted CRM
context.

## Files

- `data/train.jsonl`: 4000 chat-format SFT rows.
- `data/eval/vivamais_qa_eval.jsonl`: 158 dashboard QA eval
  rows.
- `data/teacher/rio31_teacher_distill.jsonl`: 80 accepted
  rows distilled from `prefeitura-rio/Rio-3.1-Open-4B-Instruct`.
- `metadata.json`: source volume, row counts, training recipe, and eval summary.

Each row is JSONL. Training rows use a `messages` list with chat roles. Eval rows
contain redacted Viva Mais dashboard contexts, questions, and scoring metadata.

## Dataset Mix

| Bucket | Rows |
| --- | ---: |
| Grounding negatives / unknown-answer cases | 2381 |
| Viva Mais domain QA | 1511 |
| Rio 3.1 teacher-distilled chat | 80 |
| Generic PT-BR conversation remainder | 28 |

The v4 recipe deliberately emphasized grounded refusal and entity isolation over
open-ended chat style. The target weights were negative `0.60`, domain QA `0.38`,
teacher distillation `0.02`, and generic/public/travel-vocab buckets `0.0`.
A small 28-row generic remainder was still included by the dataset builder's fill
logic.

## Teacher Distillation

- Teacher model: `prefeitura-rio/Rio-3.1-Open-4B-Instruct`
- Accepted teacher rows: 80
- Rejected teacher rows: 5
- Prompting goal: final-answer-only Brazilian Portuguese responses, avoiding
  hidden-reasoning wrappers and verbose assistant preambles.
- Filtering: removed rows with wrapper phrases, hidden reasoning artifacts,
  weak/unsafe travel claims, and answers not matching the allowed v4 categories.

## Privacy

This dataset is synthetic or redacted. Customer names, phone numbers, emails,
CPF values, card numbers, transaction identifiers, hotels, and localizers use
placeholders such as `[CLIENTE_1]`, `[TELEFONE_1]`, `[HOTEL_1]`, and
`[LOCALIZADOR_1]`. No raw WhatsApp exports, identity documents, client names,
full private transcriptions, or local private data are included.

## Evaluation Summary

The model trained from this dataset improved over `marinarosa/minicpm5-1b-vivamais-v1` on both eval
suites but did not pass the internal publish gate because leakage failures stayed
high.

| Suite | Model | Avg score | Pass rate | Count |
| --- | --- | ---: | ---: | ---: |
| Viva Mais dashboard QA | `marinarosa/minicpm5-1b-vivamais-v1` | 0.5042 | 0.4051 | 158 |
| Viva Mais dashboard QA | `marinarosa/minicpm5-1b-vivamais-v4` | 0.5570 | 0.4494 | 158 |
| Generic PT-BR conversational QA | `marinarosa/minicpm5-1b-vivamais-v1` | 0.5814 | 0.5814 | 43 |
| Generic PT-BR conversational QA | `marinarosa/minicpm5-1b-vivamais-v4` | 0.6279 | 0.6279 | 43 |

Known weakness: the candidate often starts with the correct answer but continues
generating unrelated customer facts. This causes cross-customer leakage failures.