# Qwen2.5-Omni-7B

CRESiST replaces prefix-incremental generation in LLM-based simultaneous speech translation with an interleaved sequence of speech chunks and target tokens, so the KV cache is reused across the whole stream and no voice activity detection is needed to cut sentences. An RNN-T Actor samples read-or-write trajectories, a Qwen2.5-omni-7B Critic with Router and Refiner scores them on semantic consistency and latency, and reverse distillation feeds the decisions back to the Actor. On MuST-C and ACL 60/60, unsegmented streaming lands only 1–2 COMET points below the oracle-segmented setting.

CRESiST replaces prefix-incremental generation in LLM-based simultaneous speech translation with an interleaved sequence of speech chunks and target tokens, so the KV cache is reused across the whole stream and no voice activity detection is needed to cut sentences. An RNN-T Actor samples read-or-write trajectories, a Qwen2.5-omni-7B Critic with Router and Refiner scores them on semantic consistency and latency, and reverse distillation feeds the decisions back to the Actor. On MuST-C and ACL 60/60, unsegmented streaming lands only 1–2 COMET points below the oracle-segmented setting.

Diagnosing multi-hop reasoning difficulties and partial decoupling in the local optimization of perception and reasoning objectives in omni-modal LLMs, this work proposes a text-centric post-training paradigm in which text-only training carries the main reasoning optimization and reduced-data native audio-visual RL then refines perception; the best text-only configuration (supervised fine-tuning followed by RL) raises Qwen2.5-Omni-7B's geometric mean of nine reasoning scores by 25.83% over the base model with 56.6% fewer GPU-hours than the complete native audio-visual route, training on data synthesized entirely by a text-only LLM raises this geometric mean by 21.01%, and refinement restores perception above the base level while retaining 93.5% of the reasoning gain.

Diagnosing multi-hop reasoning difficulties and partial decoupling in the local optimization of perception and reasoning objectives in omni-modal LLMs, this work proposes a text-centric post-training paradigm in which text-only training carries the main reasoning optimization and reduced-data native audio-visual RL then refines perception; the best text-only configuration (supervised fine-tuning followed by RL) raises Qwen2.5-Omni-7B's geometric mean of nine reasoning scores by 25.83% over the base model with 56.6% fewer GPU-hours than the complete native audio-visual route, training on data synthesized entirely by a text-only LLM raises this geometric mean by 21.01%, and refinement restores perception above the base level while retaining 93.5% of the reasoning gain.