Poro 2 70B Instruct
Poro 2 70B Instruct is the Finnish and English conversation release built from the Llama 3.1-based Poro 2 model. It adds supervised instruction training and Direct Preference Optimization.
Model design and training
The instruction phase uses about 1.4 million examples, including bilingual Tulu prompts with Llama 3.3 70B responses, Magpie conversations and EuroParl translation pairs. DPO uses HelpSteer3. Its Llama 3.3 license is a release-specific condition; the underlying architecture remains based on Llama 3.1.Llama-Poro-2-70B-Instruct: official model card
Practical assessment
Practical guidance. The 70B release targets the same bilingual conversation tasks at a larger scale. Assess whether its answer quality justifies its memory needs for the intended workload. Keep source passages inside the documented context rather than assuming the original Llama context limit applies.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Access | Downloadable |
| Model identifiers | Llama-Poro-2-70B-Instruct |
| Context window | 8,192 tokens Sequence length reported by this release's model card; this is not a guaranteed useful retrieval length. Llama-Poro-2-70B-Instruct: official model card |
| Model size | 70.55 billion parameters; 80 decoder layers.Llama-Poro-2-70B-Instruct: official model card |
| License | Llama 3.3 Community License.Llama-Poro-2-70B-Instruct: official model card |
Related models
- Poro 2 70B SFT
LumiOpen · Poro 2
An intermediate instruction-tuned Poro 2 70B checkpoint before the final preference-training stage.
Launch date not confirmedResearch releaseDownloadable - Poro 2 8B Instruct
LumiOpen · Poro 2
An 8B Finnish and English conversation model trained with supervised instructions and Direct Preference Optimization.
Launch date not confirmedDownloadable