Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2605.27379 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866910268673490944 |
|---|---|
| author | Liashkov, Stanislav Borde, Haitz Sáez de Ocáriz Azimi, Azizjon Shoymardonov, Khushbakht Khalilbekov, Shuhratjon Boboeva, Bonu |
| author_facet | Liashkov, Stanislav Borde, Haitz Sáez de Ocáriz Azimi, Azizjon Shoymardonov, Khushbakht Khalilbekov, Shuhratjon Boboeva, Bonu |
| contents | We present Soro, a family of Tajik-specialized conversational large language models (LLMs) designed for real-world deployment under tight compute and connectivity constraints in Tajikistan. Starting from open-weight Gemma 3 checkpoints, we perform Tajik-only continual pretraining on a curated 1.9-billion-token corpus spanning filtered web text, PDF documents, and curriculum-aligned educational materials, followed by supervised instruction tuning on 40K Tajik teacher-style examples. To enable rigorous evaluation despite the limited coverage of Tajik in standard benchmarks, we introduce a suite of Tajik benchmarks covering general knowledge, linguistic competence, and school- and university entrance-exam domains, and we open-source them on Hugging Face. Across these Tajik benchmarks, Soro substantially outperforms same-size Gemma 3 baselines while retaining strong English performance on standard datasets. We further show that FP8 and INT4 quantization of Soro preserves most Tajik-language gains while reducing memory requirements for edge deployment, supporting an ongoing education-sector pilot and planned scale-out across schools in Tajikistan. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_27379 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Soro: A Lightweight Foundation Model and Chatbot for Tajik Liashkov, Stanislav Borde, Haitz Sáez de Ocáriz Azimi, Azizjon Shoymardonov, Khushbakht Khalilbekov, Shuhratjon Boboeva, Bonu Artificial Intelligence Computation and Language We present Soro, a family of Tajik-specialized conversational large language models (LLMs) designed for real-world deployment under tight compute and connectivity constraints in Tajikistan. Starting from open-weight Gemma 3 checkpoints, we perform Tajik-only continual pretraining on a curated 1.9-billion-token corpus spanning filtered web text, PDF documents, and curriculum-aligned educational materials, followed by supervised instruction tuning on 40K Tajik teacher-style examples. To enable rigorous evaluation despite the limited coverage of Tajik in standard benchmarks, we introduce a suite of Tajik benchmarks covering general knowledge, linguistic competence, and school- and university entrance-exam domains, and we open-source them on Hugging Face. Across these Tajik benchmarks, Soro substantially outperforms same-size Gemma 3 baselines while retaining strong English performance on standard datasets. We further show that FP8 and INT4 quantization of Soro preserves most Tajik-language gains while reducing memory requirements for edge deployment, supporting an ongoing education-sector pilot and planned scale-out across schools in Tajikistan. |
| title | Soro: A Lightweight Foundation Model and Chatbot for Tajik |
| topic | Artificial Intelligence Computation and Language |
| url | https://arxiv.org/abs/2605.27379 |