Cohere2 text-generation architecture
The captured configuration identifies Cohere2ForCausalLM with model type cohere2 and Transformers support.
Open Source Model Profile · CohereLabs
tiny-aya-global
tiny-aya-global is a 3.35B-parameter Cohere2-family text-generation model from CohereLabs. According to the model card, it targets balanced coverage across 70+ languages.
- Publisher
- CohereLabs
- Task
- text-generation
- Model type
- cohere2
- License
- cc-by-nc-4.0
- Library
- transformers
- Publication status
- Accepted · not indexed
Model overview
tiny-aya-global is published by CohereLabs as a text-generation model. The captured configuration identifies Cohere2ForCausalLM with model type cohere2, and Safetensors metadata reports about 3.35B parameters. According to the model card, it is the global variant of Tiny Aya, optimized for balanced multilingual representation across 70+ languages.
Recorded capabilities
70+ language coverage
According to the model card, the model covers 70+ languages with emphasis on low-resource language performance; hub tags record the same broad language list.
Documented attention design
According to the model card, the architecture uses three sliding-window attention layers with RoPE plus a fourth global-attention layer.
Documented base lineage
Hub tags and the captured page text link this fine-tune to CohereLabs/tiny-aya-base.
Use cases in the source record
- Multilingual adaptation and instruction-tuning experiments across the documented 70+ languages, including lower-resourced ones.
- Translation, summarization, and cross-lingual generation workflows consistent with the publisher-described multilingual strengths.
Limitations and unknowns
- No extracted numeric benchmark scores were included in this record.
- According to the model card, outputs may contain incorrect or outdated statements, with greater variability in the lowest-resource languages and figurative language.
- According to the model card, commercial use is non-commercial under CC-BY-NC and interested commercial users are directed to Cohere sales.
- Provider state is historical snapshot data and should be refreshed before being presented as current.
Source and provenance
Source: CohereLabs/tiny-aya-global
Captured: Unknown. Processed: 2026-09-07T19:35:38.676996+00:00.
Model Card for tiny-aya-global Best balance across languages and regions. For other regions, check tiny-aya-fire , tiny-aya-earth , tiny-aya-water Model Summary Cohere Labs Tiny Aya is an open weights research release of a pretrained 3.35 billion parameter model optimized for efficient, strong, and balanced multilingual representation across 70+ languages, including many lower-resourced ones. The model is designed to support downstream adaptation, instruction tuning, and local deployment under realistic compute constraints. Developed by: Cohere and Cohere Labs Point of Contact: Cohere Labs License: CC-BY-NC , requires also adhering…
F001F002F003F004F005F006F007F008F009F010F012F013F016F017F018F019F020F021F022F024