Unverified50% confidenceFactExact time
The Swift series of models is trained by the UkisAI team using reinforcement learning, applying a penalty signal to the chain-of-thought length of reasoning models in order to shorten internal reasoning steps.
1
Sources
50%
Confidence
Long-term
Relevance
9/25/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe approach adopted by UkisAI does not directly train the model to shorten its thinking, but instead trains it to think more efficiently, in order to address the overthinking problem in small LLMs.72% similarVerifiedThe reasoning paradigm is essentially the application of reinforcement learning to the post-training stage of language models, allowing the model to autonomously explore reasoning paths and receive reward feedback based on answer correctness, first publicly demonstrated in the o1 series of models.71% similarUnverifiedSwift's optimization occurs at the level of reasoning behavior with no change to the model's parameter count, which is fundamentally different from distillation or quantization compression methods that target model weights.70% similarUnverifiedThe traditional deep learning paradigm divides the model into two stages: a training stage for learning parameters, and an inference stage where parameters are frozen and outputs are produced directly.67% similarUnverifiedo系列针对链式思维推理进行强化训练，引入了蒙特卡洛树搜索（MCTS）与过程奖励模型67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/951063API
curl https://kongchang.com/api/v1/knowledge/claims/951063MCP
get_claim(id=951063)