SHAKTI: A 2.5 Billion Parameter Small Language Model Optimized for Edge AI and Low-Resource Environments
Syed Abdul Gaffar Shakhadri, Kruthika KR, Rakshit Aralimatti
Unverified — Be the first to reproduce this paper.
ReproduceAbstract
We introduce Shakti, a 2.5 billion parameter language model specifically optimized for resource-constrained environments such as edge devices, including smartphones, wearables, and IoT systems. Shakti combines high-performance NLP with optimized efficiency and precision, making it ideal for real-time AI applications where computational resources and memory are limited. With support for vernacular languages and domain-specific tasks, Shakti excels in industries such as healthcare, finance, and customer service. Benchmark evaluations demonstrate that Shakti performs competitively against larger models while maintaining low latency and on-device efficiency, positioning it as a leading solution for edge AI.
Tasks
Benchmark Results
| Dataset | Model | Metric | Claimed | Verified | Status |
|---|---|---|---|---|---|
| BBH | Shakti-LLM (2.5B) | Accuracy | 58.2 | — | Unverified |
| BoolQ | Shakti-LLM (2.5B) | Accuracy | 61.1 | — | Unverified |
| HellaSwag | Shakti-LLM (2.5B) | Accuracy | 52.4 | — | Unverified |
| MedQA | Shakti-LLM (2.5B) | Accuracy | 60.3 | — | Unverified |
| MML | qwen-LLM 7B | Accuracy | 71.8 | — | Unverified |
| PIQA | Shakti-LLM (2.5B) | Accuracy | 86.2 | — | Unverified |
| TriviaQA | Shakti-LLM (2.5B) | EM | 58.2 | — | Unverified |
| TruthfulQA | Shakti-LLM (2.5B) | Accuracy | 68.4 | — | Unverified |