Releases·Model Release
LNCC releases Manacá-1B, an open ~1.72B-parameter LLM trained from scratch for Brazilian Portuguese with reproducible pipeline
Manacá-1B, an open and reproducible language model for Brazilian Portuguese, was released via GitHub with weights hosted on Hugging Face under the account menezesbruno. The project is licensed CC BY 4.0, positioning it as a fully open resource for Brazilian-Portuguese NLP practitioners.
67 engagement1 sourceMon, Aug 31, 2026, 12:30 AM
AI Overview
- LNCC's Manacá-1B is a decoder-only ~1.72B-parameter model trained from scratch for Brazilian Portuguese, released under CC BY 4.0 on Hugging Face
- Repo ships fully containerized reproducible pipeline: Docker environment, corpus pipeline, tokenizer training, Megatron-LM pretraining, HF conversion, and PT benchmarks with paired significance tests
Related
Releases·Model Release
mii-llm releases Zagreus-0.4B-por, a small open-source Portuguese-English model
Open-source AI lab mii-llm released Zagreus-0.4B-por, a 400-million-parameter bilingual Portuguese-English language model pretrained from scratch. The model is part of the Zagreus family focused on European languages, with training sponsored by Seeweb and Regolo.ai. It is available on Hugging Face.
Open-source AI lab mii-llm released Zagreus-0.4B-por, a 400-million-parameter bilingual Portuguese-English language model pretrained from scratch. The model is part of the Zagreus family focused on European languages, with training sponsored by Seeweb and Regolo.ai. It is available on Hugging Face.
14 engagement1 sourceredditFri, Jul 24, 2026, 02:37 PM
Products·launched
Amalia: First Portuguese LLM hosted for public use
Amalia is the first large language model specifically for Portuguese, hosted for public access. It solves the lack of Portuguese-language AI models for native speakers and developers.
Amalia is the first large language model specifically for Portuguese, hosted for public access. It solves the lack of Portuguese-language AI models for native speakers and developers.
2 engagement1 sourcehackernewsWed, Jul 15, 2026, 02:04 PM
Releases·Model Release
Independent developer releases WARMIND-200M V2, a Portuguese-first 203M LLM
An independent Brazilian developer released WARMIND-200M V2, a 203M-parameter Portuguese-first causal language model trained from scratch. The model runs on CPU and is positioned as a research checkpoint, not a production assistant. The release includes a real demo video showing its capabilities and limitations.
An independent Brazilian developer released WARMIND-200M V2, a 203M-parameter Portuguese-first causal language model trained from scratch. The model runs on CPU and is positioned as a research checkpoint, not a production assistant. The release includes a real demo video showing its capabilities and limitations.
10 engagement2 sourcesredditWed, Aug 5, 2026, 07:48 AM
Releases·Model Release
Indie developer releases 3.7B multilingual reasoning MoE trained from scratch on a single RTX 4090
A solo developer uploaded a checkpoint of Polanka 3.7B, a multilingual reasoning mixture-of-experts model pretrained, mid-trained, and fine-tuned from scratch on one consumer-grade RTX 4090 over many months. The checkpoint is public on Hugging Face as a research artifact. It shows small-scale from-scratch pretraining remains feasible on consumer hardware.
A solo developer uploaded a checkpoint of Polanka 3.7B, a multilingual reasoning mixture-of-experts model pretrained, mid-trained, and fine-tuned from scratch on one consumer-grade RTX 4090 over many months. The checkpoint is public on Hugging Face as a research artifact. It shows small-scale from-scratch pretraining remains feasible on consumer hardware.
57 engagement1 sourceredditTue, Sep 1, 2026, 03:26 PM
Social·Community
Medium article explains how large language models learn
A Medium article provides an educational overview of how large language models learn, from billions of parameters to next-token prediction. It aims to demystify the internal workings of modern language models for a general audience.
A Medium article provides an educational overview of how large language models learn, from billions of parameters to next-token prediction. It aims to demystify the internal workings of modern language models for a general audience.
0 engagement1 sourcerssSat, Aug 22, 2026, 09:11 PM