SAGE Celer 2.6 Technical Card
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2026
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866913036396134400 |
|---|---|
| author | SAGEA Research Team Jha, Basab Paudel, Firoj Puri, Ujjwal Liu, Adrian Henkel, Ethan Yuting, Zhang Kowalczyk, Mateusz Huang, Mei Donghyuk, Choi Junhao, Wang |
| author_facet | SAGEA Research Team Jha, Basab Paudel, Firoj Puri, Ujjwal Liu, Adrian Henkel, Ethan Yuting, Zhang Kowalczyk, Mateusz Huang, Mei Donghyuk, Choi Junhao, Wang |
| contents | We introduce SAGE Celer 2.6, the latest in our line of general-purpose Celer models from SAGEA. Celer 2.6 is available in 5B, 10B, and 27B parameter sizes and benefits from extensive architectural modifications and further pre-training on an undisclosed model. Using our Inverse Reasoning (IR) pipeline, SAGEA natively trains Celer 2.6 to validate its own logic paths, minimizing cascading error and hallucination in complex reasoning tasks. Celer 2.6 also boasts natively integrated multimodal functionality with an end-to-end vision encoder to avoid common pitfalls in adapter-based approaches. Celer 2.6 provides highly competitive results on mathematics, coding, and general intelligence benchmarks (ACUMEN), along with low latency. Most importantly, Celer 2.6 is specifically optimized for South Asian language support, with a custom tokenizer for the Devanagari script and strong performance in both Nepali and Hindi without sacrificing English reasoning ability. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2604_14168 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | SAGE Celer 2.6 Technical Card SAGEA Research Team Jha, Basab Paudel, Firoj Puri, Ujjwal Liu, Adrian Henkel, Ethan Yuting, Zhang Kowalczyk, Mateusz Huang, Mei Donghyuk, Choi Junhao, Wang Computation and Language Artificial Intelligence We introduce SAGE Celer 2.6, the latest in our line of general-purpose Celer models from SAGEA. Celer 2.6 is available in 5B, 10B, and 27B parameter sizes and benefits from extensive architectural modifications and further pre-training on an undisclosed model. Using our Inverse Reasoning (IR) pipeline, SAGEA natively trains Celer 2.6 to validate its own logic paths, minimizing cascading error and hallucination in complex reasoning tasks. Celer 2.6 also boasts natively integrated multimodal functionality with an end-to-end vision encoder to avoid common pitfalls in adapter-based approaches. Celer 2.6 provides highly competitive results on mathematics, coding, and general intelligence benchmarks (ACUMEN), along with low latency. Most importantly, Celer 2.6 is specifically optimized for South Asian language support, with a custom tokenizer for the Devanagari script and strong performance in both Nepali and Hindi without sacrificing English reasoning ability. |
| title | SAGE Celer 2.6 Technical Card |
| topic | Computation and Language Artificial Intelligence |
| url | https://arxiv.org/abs/2604.14168 |