Adobe Systems (United States)

Language-guided segmentation usually compresses the referent's location into a single [SEG] hidden-state vector handed to the decoder, a vector with no spatial index while the decoder never sees the expression. AttnSeg instead passes the [SEG] token's attention over image patches as the decoder's dense prompt and supervises that map directly with an attention-guidance objective in two training stages. On Sa2VA, the 4B model raises ReasonSeg gIoU from 57.70 to 64.11 and GS-Eval from 61.91 to 72.11, with gains concentrated on object parts, multiple objects and single objects, and negligible change on stuff regions.

Language-guided segmentation usually compresses the referent's location into a single [SEG] hidden-state vector handed to the decoder, a vector with no spatial index while the decoder never sees the expression. AttnSeg instead passes the [SEG] token's attention over image patches as the decoder's dense prompt and supervises that map directly with an attention-guidance objective in two training stages. On Sa2VA, the 4B model raises ReasonSeg gIoU from 57.70 to 64.11 and GS-Eval from 61.91 to 72.11, with gains concentrated on object parts, multiple objects and single objects, and negligible change on stuff regions.

Prompt2Skill is a multi-agent framework that, from a natural-language task description alone, has a Data Agent derive a task specification and build a proxy data pool through retrieval, adaptation, and synthesis, and a Skill Agent refine the skill in a closed loop of execution feedback, reflective editing, and paired statistical screening, improving over direct prompting by 36.1% and over off-the-shelf skills by 83.0% on average across question answering, reading comprehension, spreadsheet manipulation, and mathematical reasoning with five open-source and commercial target models, without significant regressions.

Prompt2Skill is a multi-agent framework that, from a natural-language task description alone, has a Data Agent derive a task specification and build a proxy data pool through retrieval, adaptation, and synthesis, and a Skill Agent refine the skill in a closed loop of execution feedback, reflective editing, and paired statistical screening, improving over direct prompting by 36.1% and over off-the-shelf skills by 83.0% on average across question answering, reading comprehension, spreadsheet manipulation, and mathematical reasoning with five open-source and commercial target models, without significant regressions.

Observing that video and audio are perceived together yet most generative models treat them in isolation, this survey casts joint generation, cross-modal generation (one modality from the other) and joint editing as three problems defined on a single distribution over audio-visual pairs, organizes the field around one question—how the output is kept coherent across modalities in time and semantics—proposes a taxonomy along five design axes (generation strategy, audio representation, video representation, alignment enforcement, pretraining reuse), and, to the authors' knowledge, provides the first systematic taxonomy of joint audio-visual editing, mapped as nine edit categories spanning 28 edit types, together with methods, datasets and metrics for each setting and a closing list of open pr

Observing that video and audio are perceived together yet most generative models treat them in isolation, this survey casts joint generation, cross-modal generation (one modality from the other) and joint editing as three problems defined on a single distribution over audio-visual pairs, organizes the field around one question—how the output is kept coherent across modalities in time and semantics—proposes a taxonomy along five design axes (generation strategy, audio representation, video representation, alignment enforcement, pretraining reuse), and, to the authors' knowledge, provides the first systematic taxonomy of joint audio-visual editing, mapped as nine edit categories spanning 28 edit types, together with methods, datasets and metrics for each setting and a closing list of open pr

The work reframes LLM routing as coverage-oriented subset selection, uses a determinantal point process to capture both model competence and inter-model redundancy, and trains with a failure-set-marginalization objective that directly maximizes the probability that at least one selected model answers correctly, achieving higher average Success@10 (0.8632 and 0.9914) and more diverse subsets on the medium-pool (3811 candidate LLMs) and large-pool (5000 candidate LLMs) RouterEval settings, while an adaptive greedy stopping rule sets subset size per query.

The work reframes LLM routing as coverage-oriented subset selection, uses a determinantal point process to capture both model competence and inter-model redundancy, and trains with a failure-set-marginalization objective that directly maximizes the probability that at least one selected model answers correctly, achieving higher average Success@10 (0.8632 and 0.9914) and more diverse subsets on the medium-pool (3811 candidate LLMs) and large-pool (5000 candidate LLMs) RouterEval settings, while an adaptive greedy stopping rule sets subset size per query.

The authors present OmniSeek, which turns an Omni-LLM into an active multi-turn tool-using agent (get_audio_clip, get_video_clip), backed by a data engine that synthesizes OmniTraj-170K with 169,725 interleaved audio-visual chain-of-thought trajectories, plus a three-phase training pipeline and an Audio-Visual Necessity reward, achieving leading or competitive results on 10 omni-modal and 4 general video benchmarks.

The authors present OmniSeek, which turns an Omni-LLM into an active multi-turn tool-using agent (get_audio_clip, get_video_clip), backed by a data engine that synthesizes OmniTraj-170K with 169,725 interleaved audio-visual chain-of-thought trajectories, plus a three-phase training pipeline and an Audio-Visual Necessity reward, achieving leading or competitive results on 10 omni-modal and 4 general video benchmarks.

The work proposes DISCO, which separates long-context processing into local evidence grounding executed in parallel by lightweight Worker LLMs and global reasoning handled by a central Driver LLM whose dynamic DAG planning is optimized with GRPO reinforcement learning; with Qwen3-8B it holds 78.4% accuracy on RULER-QA at 1M tokens (where the RAG baseline collapses to 10.9%), improves up to 9.8 points over full-context baselines on LongBench v2, and matches Gemini-3-Pro-Preview while reducing inference cost by over 80%.

The work proposes DISCO, which separates long-context processing into local evidence grounding executed in parallel by lightweight Worker LLMs and global reasoning handled by a central Driver LLM whose dynamic DAG planning is optimized with GRPO reinforcement learning; with Qwen3-8B it holds 78.4% accuracy on RULER-QA at 1M tokens (where the RAG baseline collapses to 10.9%), improves up to 9.8 points over full-context baselines on LongBench v2, and matches Gemini-3-Pro-Preview while reducing inference cost by over 80%.