Support · Tucano 2
Will it last?
Ownership levelPartialnone·limited·partial·substantial·fullAnalytical input D ยท 54.4/100
This page is a projection of the one entry record, the Transparency factor that Support covers. The full verdict is set by all four factors together, floor-weighted so the weakest caps the whole.
Which domain expands which factor
- AssessUse & modify + Transparency
- ImplementData control + Reliability
- UseReliability
- SupportTransparency
Common problems & fixes
- Repetition or run-on output: keep
repetition_penalty near 1.2 and set a hard max_new_tokens (ev-card-instruct).
- Wrong format or poor answers: build the prompt with the repo's chat template, not hand-written strings (
ev-configs).
- Failures beyond 4,096 tokens: split the input and retrieve.
- Invented facts: the cards and a third-party run both report hallucination, so verify outputs (
ev-community).
- Weak instruction following or code: pair the model with tooling, or fine-tune. The paper says the SFT data lacks IFEval-style prompts and has few coding samples (
ev-paper).
Versions, changelog & cadence
The models were uploaded in February 2026, and the paper is dated 3 March 2026 (ev-paper). The post-trained repos were last edited on 2026-08-13, and the GigaVerbo-v2 dataset card on 2026-09-30 (ev-hf-platform, ev-gv2). There is no changelog file or release notes. The commit hash is the anchor, so pin it. The publisher has since released similar suites for Hindi and Bengali, and no roadmap for Tucano 2 is stated.
Security / vulnerability disclosure
The llm-foundry code repository has a SECURITY.md that asks reporters to email a maintainer privately and calls the project research software without a security team (ev-code). The model repos publish no security policy or contact.
Hugging Face discussions are open and empty on the repos inspected (ev-hf-platform). The llm-foundry repository takes GitHub issues. The dataset cards list a Polygl0t project address and maintainer emails (ev-gv2-post).
Tracked known issues
The publisher lists hallucination, bias, repetition and language limits on every card, and the paper lists weak instruction following, coding and long-context retrieval (ev-card-instruct, ev-paper). The corpus licence is mixed with two non-commercial sources, and Tucano2-qwen-1.5B-Base has no LICENSE file (ev-gv2, ev-licence-files). The models have no public issue tracker.
How this scores
The ownership factor this domain covers, drawn from the one entry record.
2
TransparencyDo you know what it is: weights, training, behaviour, and legible terms?
ModerateThe Tucano stages are documented down to configs, logs and datasets (ev-paper, ev-configs). The Qwen3 base stage and its 36T-token data are undisclosed (ev-qwen3-licence), and the corpus carries no single licence (ev-gv2).
How this scores (AOI sub-dimensions)
Provenance2/5how well we can trace and verify what went into the modelSafetensors only, with a Hugging Face LFS sha256 per file, and a clear lineage to Qwen3 base models.
Governance3/5how accountable and well-documented the publisher isA named University of Bonn team with a funded project (BMBF and the NRW science ministry), an arXiv paper, a prior Tucano series and a 2026-08-13 update to the post-trained repos.
What this means for adoptionYou own the weights: Apache-2.0, ungated, commercial use allowed, with the code and data to retrain the Tucano stages on your own hardware. You do not own a verified safe assistant, because no safety evaluation exists, and you cannot rebuild the Qwen3 stage underneath. The level rises if Polygl0t publishes a safety evaluation and resolves the mixed licence of the training corpus.
Sources
The same evidence records as the entry sheet. Read means the text was verified; unverified means it is known to exist but not yet read.
Model cardread2026-11-19
The 3.7B Instruct card declares Apache-2.0, 3,759,341,056 parameters and a 4,096-token context, and describes one SFT round and one APO round on gigaverbo-v2-sft and gigaverbo-v2-preferences.
Model cardread2026-11-19
The 3.7B Base card says the model is continually pretrained from Qwen3-4B-Base on about 50 billion tokens, with the Tucano2 tokenizer transplanted by Orthogonal Matching Pursuit.
Model cardread2026-11-19
The 3.7B Think card declares Apache-2.0 and describes a reasoning model trained with SFT and APO that writes chain-of-thought inside think tags.
Model cardread2026-11-19
The Tucano2-0.6B-Base card declares Apache-2.0, a Llama architecture with 670,127,616 parameters, about 400 billion training tokens and three published stage configs.
Licenceread2026-11-19
Eight of the nine Qwen-derived repos and the 0.6B Base repo carry an identical Apache-2.0 LICENSE file that names four individuals as copyright holders.
Licenceread2026-11-19
Qwen3-4B-Base, 1.7B-Base and 0.6B-Base each carry an Apache-2.0 LICENSE file and are ungated.
Licenceread2026-11-19
Qwen2.5-7B, 14B and 32B-Instruct, which generated the synthetic data, and the abliterated 32B variant that wrote the rejected responses, are Apache-2.0 with LICENSE files and ungated.
Technical_reportread2026-11-19
The paper 'Tucano 2 Cool' (arXiv 2603.03543, 3 March 2026) describes continual pretraining from Qwen3 bases, a from-scratch 0.6B model, the data, the post-training recipe and the publisher's benchmarks.
Documentationread2026-11-19
The GigaVerbo-v2 card lists its source datasets with their licences and states that the combined dataset has no single unified licence (license: other).
Documentationread2026-11-19
The Synth, SFT and Preferences cards declare Apache-2.0 and are ungated.
Documentationread2026-11-19
llm-foundry (training, data, alignment and evaluation code) is Apache-2.0 and the Portuguese lm-evaluation-harness fork is MIT.
Documentationread2026-11-19
The Polygl0t Hugging Face organisation is not verified and has seven members.
Documentationread2026-11-19
The Instruct and Think repos publish SFT and APO YAML configs, training logs, config.json and a ChatML chat template.
Third-party analysisread2026-11-19
A third-party repo ran the 3.7B Instruct in bf16 on vLLM 0.33.0 on an RTX 3080 with 8.7 GiB of VRAM and scored 74.6% on 393 text-only ENEM questions.