SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning | Paper Citation Record

Observation 34e54f63-db9d-4116-9061-15db86c94e90 · outbound

This paper cites Flamingo: a visual language model for few-shot learning.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Flamingo: a visual language model for few-shot learning

Reference 1

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 54c7000b-c7f6-4b4b-8c2d-6c6747185dec · outbound

This paper cites Qwen Technical Report.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Qwen Technical Report

Reference 2

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 9635897f-d81f-4526-af72-c268849e872a · outbound

This paper cites Token Merging: Your ViT But Faster.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Token Merging: Your ViT But Faster

Reference 3

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 46ee54ee-686e-4a6c-98ba-a983bea94e1a · outbound

This paper cites Medusa: Simple LLM Inference Acceleration Framework with Multiple Decoding Heads.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Medusa: Simple LLM Inference Acceleration Framework with Multiple Decoding Heads

Reference 4

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation c58291ce-3fb6-4e7c-b5a1-5c3d7bfa37a0 · outbound

This paper cites Accelerating Large Language Model Decoding with Speculative Sampling.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Accelerating Large Language Model Decoding with Speculative Sampling

Reference 5

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation d5e4f275-a1da-4fe6-948e-c3f13720c793 · outbound

This paper cites EE-LLM: Large-Scale Training and Inference of Early-Exit Large Language Models with 3D Parallelism.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning EE-LLM: Large-Scale Training and Inference of Early-Exit Large Language Models with 3D Parallelism

Reference 6

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 470689ce-c0bf-446b-b395-60530e85f5be · outbound

This paper cites Sensenova-mars: Empowering multimodal agentic reasoning and search via reinforcement learning.arXiv preprint arXiv:2512.24330, 2025.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Sensenova-mars: Empowering multimodal agentic reasoning and search via reinforcement learning.arXiv preprint arXiv:2512.24330, 2025

Reference 7

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 61464e67-6d1f-4c14-b154-ba829024aeca · outbound

This paper cites Instructblip: Towards general-purpose vision-language models with instruction tuning.Advances in neural information processing systems, 36:49250–49267, 2023.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Instructblip: Towards general-purpose vision-language models with instruction tuning.Advances in neural information processing systems, 36:49250–49267, 2023

Reference 8

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation ad7d7b0e-1c20-4616-ac1d-5ffd55ea5570 · outbound

This paper cites Introducing Agentic Vision in Gemini 3 Flash.https://blog.google/innovation-and-ai/tech nology/developers-tools/agentic-vision-gemini-3-flash/, January 2026.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Introducing Agentic Vision in Gemini 3 Flash.https://blog.google/innovation-and-ai/tech nology/developers-tools/agentic-vision-gemini-3-flash/, January 2026

Reference 9

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 3837c31c-48ad-4fd8-bf45-3c743b5c4dbb · outbound

This paper cites Feather the throttle: Revisiting visual token pruning for vision-language model acceleration.ICCV, 2025.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Feather the throttle: Revisiting visual token pruning for vision-language model acceleration.ICCV, 2025

Reference 10

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation cc2b33b2-b2ca-4e61-ad20-7f2e725eeba2 · outbound

This paper cites Not All Layers of LLMs Are Necessary During Inference.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Not All Layers of LLMs Are Necessary During Inference

Reference 11

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 21086179-5cc2-4cc4-bcb4-dfe6dc198c08 · outbound

This paper cites Framefusion: Combining similarity and importance for video token reduction on large vision language models.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Framefusion: Combining similarity and importance for video token reduction on large vision language models

Reference 12

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 9c2bab70-f4cc-4cea-8fdc-e1c106154e02 · outbound

This paper cites Deep Think with Confidence.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Deep Think with Confidence

Reference 13

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation e8ff4a20-f182-41dc-86fe-a9e413a6b2ce · outbound

This paper cites Thinking with programming vision: Towards a unified view for thinking with images.arXiv preprint arXiv:2512.03746, 2025.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Thinking with programming vision: Towards a unified view for thinking with images.arXiv preprint arXiv:2512.03746, 2025

Reference 14

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 71cb0fa2-72e8-4804-a2ac-c617994cabc5 · outbound

This paper cites ZipVL: Efficient Large Vision-Language Models with Dynamic Token Sparsification.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning ZipVL: Efficient Large Vision-Language Models with Dynamic Token Sparsification

Reference 15

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 5907ab8e-022f-4083-9490-742b9b978eaf · outbound

This paper cites DeepEyesV2: Toward Agentic Multimodal Model.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning DeepEyesV2: Toward Agentic Multimodal Model

Reference 16

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 5adaf7d2-6f7a-4960-8627-9141ec49af55 · outbound

This paper cites Thinking with drafts: Speculative temporal reasoning for efficient long video understanding.ArXiv, abs/2512.00805, 2025.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Thinking with drafts: Speculative temporal reasoning for efficient long video understanding.ArXiv, abs/2512.00805, 2025

Reference 17

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 1770a245-7b8d-4866-ae0e-74e3e5e96826 · outbound

This paper cites Relayllm: Efficient reasoning via collaborative decoding.ArXiv, abs/2601.05167, 2026.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Relayllm: Efficient reasoning via collaborative decoding.ArXiv, abs/2601.05167, 2026

Reference 18

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 7be1eb8d-0728-4f8b-b0dd-8395a292ad02 · outbound

This paper cites Evolver: Chain-of-evolution prompting to boost large multimodal models for hateful meme detection.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Evolver: Chain-of-evolution prompting to boost large multimodal models for hateful meme detection

Reference 19

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 30cc7fc8-5829-4141-93c3-35b8206bb8a9 · outbound

This paper cites Token fusion: Bridging the gap between token pruning and token merging.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Token fusion: Bridging the gap between token pruning and token merging

Reference 20

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation e02c6d94-8efc-40ea-88dd-6571b75fbc34 · outbound

This paper cites Helios: Adaptive model and early-exit selection for efficient llm inference serving.arXiv preprint arXiv:2504.10724, 2025.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Helios: Adaptive model and early-exit selection for efficient llm inference serving.arXiv preprint arXiv:2504.10724, 2025

Reference 21

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation b0e2d2ac-a1aa-4845-ab46-49f8b9db3524 · outbound

This paper cites Mini-o3: Scaling Up Reasoning Patterns and Interaction Turns for Visual Search.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Mini-o3: Scaling Up Reasoning Patterns and Interaction Turns for Visual Search

Reference 22

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 57eb8671-8594-4c3d-9b62-6b218fbbc181 · outbound

This paper cites Fast inference from transformers via speculative decoding.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Fast inference from transformers via speculative decoding

Reference 23

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 8304f4df-66d0-4576-b068-6176a1889d54 · outbound

This paper cites Blip-2: Bootstrapping language-image pre-training with frozen image encoders and large language models.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Blip-2: Bootstrapping language-image pre-training with frozen image encoders and large language models

Reference 24

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 7dffeb42-2ff9-48b8-b07a-fb2b9128e09d · outbound

This paper cites Hero: Rethinking visual token early dropping in high-resolution large vision-language models, 2025.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Hero: Rethinking visual token early dropping in high-resolution large vision-language models, 2025

Reference 25

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 2962bf72-0d68-4f07-a9ad-ef73252a0a98 · outbound

This paper cites Evaluating object hallucination in large vision-language models.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Evaluating object hallucination in large vision-language models

Reference 26

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation f4b2ba77-373f-49ce-ad84-8bb63e273a85 · outbound

This paper cites doi: 10.18653/v1/2023.emnlp-main.20.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning doi: 10.18653/v1/2023.emnlp-main.20

Reference 27

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 865dc1a7-07b6-4cbd-b977-bc6601ca1654 · outbound

This paper cites EAGLE: Speculative Sampling Requires Rethinking Feature Uncertainty.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning EAGLE: Speculative Sampling Requires Rethinking Feature Uncertainty

Reference 28

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation aeca9d6b-6911-460a-8817-b66377ca052f · outbound

This paper cites Eagle-2: Faster inference of language models with dynamic draft trees.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Eagle-2: Faster inference of language models with dynamic draft trees

Reference 29

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation d2091b90-f24d-49c8-8bdf-6a008faf3a68 · outbound

This paper cites EAGLE-3: Scaling up Inference Acceleration of Large Language Models via Training-Time Test.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning EAGLE-3: Scaling up Inference Acceleration of Large Language Models via Training-Time Test

Reference 30

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation b468e171-18ce-477c-bfb3-1a1ef890b242 · outbound

This paper cites Moe-llava: Mixture of experts for large vision-language models.IEEE Transactions on Multimedia, 2026.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Moe-llava: Mixture of experts for large vision-language models.IEEE Transactions on Multimedia, 2026

Reference 31

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 8ef8e9bd-1a19-4d4c-a9c4-16e926bbd613 · outbound

This paper cites Speculative Decoding Reimagined for Multimodal Large Language Models.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Speculative Decoding Reimagined for Multimodal Large Language Models

Reference 32

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation c28b0408-a7a4-48c1-b2f8-d954127d7d56 · outbound

This paper cites Accelerating multi-modal llm gaming performance via input prediction and mishit correction.arXiv preprint arXiv:2512.17250, 2025.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Accelerating multi-modal llm gaming performance via input prediction and mishit correction.arXiv preprint arXiv:2512.17250, 2025

Reference 33

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 5113e20f-ef45-4ef2-a4ec-dd415cb88929 · outbound

This paper cites Efficient inference of vision instruction-following models with elastic cache.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Efficient inference of vision instruction-following models with elastic cache

Reference 34

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation ed59f10a-b025-44c8-8478-96155b13dfcf · outbound

This paper cites Video-rag: Visually-aligned retrieval-augmented long video comprehension.arXiv preprint arXiv:2411.13093, 2024.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Video-rag: Visually-aligned retrieval-augmented long video comprehension.arXiv preprint arXiv:2411.13093, 2024

Reference 35

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 92c3a50f-89bb-4581-a27b-0d86d16ce91c · outbound

This paper cites Quota: Query-oriented token assignment via cot query decouple for long video comprehension.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Quota: Query-oriented token assignment via cot query decouple for long video comprehension

Reference 36

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 91a9334f-8741-45d2-8aa6-4df40d3e0291 · outbound

This paper cites Introducing OpenAI o3 and o4-mini.https://openai.com/index/introducing-o3-and-o4-mini/ , April 2025.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Introducing OpenAI o3 and o4-mini.https://openai.com/index/introducing-o3-and-o4-mini/ , April 2025

Reference 37

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 505f063e-809c-4352-a29c-3175fb56c820 · outbound

This paper cites SpecReason: Fast and Accurate Inference-Time Compute via Speculative Reasoning.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning SpecReason: Fast and Accurate Inference-Time Compute via Speculative Reasoning

Reference 38

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 68baa483-9c90-4e76-8b4e-2da6fe045052 · outbound

This paper cites Skywork R1V: Pioneering Multimodal Reasoning with Chain-of-Thought.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Skywork R1V: Pioneering Multimodal Reasoning with Chain-of-Thought

Reference 39

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 5cec84e8-696c-426f-aa9f-973b5c627484 · outbound

This paper cites Toolformer: Language models can teach themselves to use tools.Advances in neural information processing systems, 36:68539–68551, 2023.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Toolformer: Language models can teach themselves to use tools.Advances in neural information processing systems, 36:68539–68551, 2023

Reference 40

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation e04f0807-dfd7-4296-8400-b8e374934018 · outbound

This paper cites Mmspec: Benchmarking speculative decoding for vision-language models.arXiv preprint arXiv:2603.14989, 2026.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Mmspec: Benchmarking speculative decoding for vision-language models.arXiv preprint arXiv:2603.14989, 2026

Reference 41

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 768424c4-a3f0-4d6a-9bba-8ada2c25b8ce · outbound

This paper cites Hugginggpt: Solving ai tasks with chatgpt and its friends in hugging face.Advances in Neural Information Processing Systems, 36: 38154–38180, 2023.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Hugginggpt: Solving ai tasks with chatgpt and its friends in hugging face.Advances in Neural Information Processing Systems, 36: 38154–38180, 2023

Reference 42

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 8fb8a25a-fb83-4f89-834a-d4863fd8685d · outbound

This paper cites Codedance: A dynamic tool-integrated mllm for executable visual reasoning.ArXiv, abs/2512.17312, 2025.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Codedance: A dynamic tool-integrated mllm for executable visual reasoning.ArXiv, abs/2512.17312, 2025

Reference 43

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 3af15faf-0a09-45b0-bff6-2e5aa8371daf · outbound

This paper cites Gemini: A Family of Highly Capable Multimodal Models.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Gemini: A Family of Highly Capable Multimodal Models

Reference 44

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 48f721f0-df1e-4110-b945-6c01c828914e · outbound

This paper cites Kimi K2.5: Visual Agentic Intelligence.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Kimi K2.5: Visual Agentic Intelligence

Reference 45

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation c43f5d66-9646-4004-9d86-aba71934fe96 · outbound

This paper cites Qwen3 Technical Report.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Qwen3 Technical Report

Reference 46

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 390f05de-486e-4e6f-9ba0-9688c9851933 · outbound

This paper cites Branchynet: Fast inference via early exiting from deep neural networks.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Branchynet: Fast inference via early exiting from deep neural networks

Reference 47

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 6ca8693a-df7b-4d48-90c9-2baaf131be25 · outbound

This paper cites Look-m: Look-once optimization in kv cache for efficient multimodal long-context inference.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Look-m: Look-once optimization in kv cache for efficient multimodal long-context inference

Reference 48

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 39829c2f-335e-4840-9dc0-a52800e8b6ae · outbound

This paper cites Meda: Dynamic kv cache allocation for efficient multimodal long-context inference.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Meda: Dynamic kv cache allocation for efficient multimodal long-context inference

Reference 49

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation a7f7de56-a813-4e29-8ef1-e6b43ea53ec5 · outbound

This paper cites Fourier-vlm: Compressing vision tokens in the frequency domain for large vision-language models, 2025.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Fourier-vlm: Compressing vision tokens in the frequency domain for large vision-language models, 2025

Reference 50

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 6035e410-db77-413a-b73b-28e2a565e747 · outbound

This paper cites Divide, conquer and combine: A training-free framework for high-resolution image perception in multimodal large language models.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Divide, conquer and combine: A training-free framework for high-resolution image perception in multimodal large language models

Reference 51

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation d7b79076-2a35-4c51-b3a5-e1c7c493d74f · outbound

This paper cites Efficient visual transformer by learnable token merging.IEEE transactions on pattern analysis and machine intelligence, 2025.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Efficient visual transformer by learnable token merging.IEEE transactions on pattern analysis and machine intelligence, 2025

Reference 52

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 1e1dfc9c-74da-4b71-a0dc-5347b54124fa · outbound

This paper cites V*: Guided Visual Search as a Core Mechanism in Multimodal LLMs.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning V*: Guided Visual Search as a Core Mechanism in Multimodal LLMs

Reference 53

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 917e1973-cdee-4155-b008-137144e81711 · outbound

This paper cites Speculative decoding: Exploiting speculative execution for accelerating seq2seq generation.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Speculative decoding: Exploiting speculative execution for accelerating seq2seq generation

Reference 54

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 04da9712-272a-4d07-9bd5-fd6b33f8eac3 · outbound

This paper cites SWIFT: On-the-Fly Self-Speculative Decoding for LLM Inference Acceleration.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning SWIFT: On-the-Fly Self-Speculative Decoding for LLM Inference Acceleration

Reference 55

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation a7fa7d87-6cf9-444e-a70e-feef499a50ce · outbound

This paper cites PyramidDrop: Accelerating Your Large Vision-Language Models via Pyramid Visual Redundancy Reduction.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning PyramidDrop: Accelerating Your Large Vision-Language Models via Pyramid Visual Redundancy Reduction

Reference 56

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 43b4f0ac-d0b9-43aa-b95b-3accc145fc80 · outbound

This paper cites Specee: Accelerating large language model inference with speculative early exiting.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Specee: Accelerating large language model inference with speculative early exiting

Reference 57

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 90f6e4e4-9d6a-40ca-ac97-2974fca5bac3 · outbound

This paper cites Longspec: Long-context speculative decoding with efficient drafting and verification.arXiv e-prints, pages arXiv–2502, 2025.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Longspec: Long-context speculative decoding with efficient drafting and verification.arXiv e-prints, pages arXiv–2502, 2025

Reference 58

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 3cd65eb9-9a7b-40c5-950c-0cbc922ea680 · outbound

This paper cites Visionzip: Longer is better but not necessary in vision language models.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Visionzip: Longer is better but not necessary in vision language models

Reference 59

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation a4f9cf90-e7b9-4db2-909b-475ca8403595 · outbound

This paper cites Deep but reliable: Advancing multi-turn reasoning for thinking with images, 2026.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Deep but reliable: Advancing multi-turn reasoning for thinking with images, 2026

Reference 60

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 1bf4a10b-c5a0-4f54-a40f-2973e24afd36 · outbound

This paper cites React: Synergizing reasoning and acting in language models.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning React: Synergizing reasoning and acting in language models

Reference 61

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 11d17b4a-0b54-47a3-95ab-0573ef314f30 · outbound

This paper cites Recode: Unify plan and action for universal granularity control.arXiv preprint arXiv:2510.23564, 2025.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Recode: Unify plan and action for universal granularity control.arXiv preprint arXiv:2510.23564, 2025

Reference 62

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 9fcbd1c0-cbd5-4a0b-b67d-522c33f97ba2 · outbound

This paper cites Draft& verify: Lossless large language model acceleration via self-speculative decoding.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Draft& verify: Lossless large language model acceleration via self-speculative decoding

Reference 63

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation ffaec66f-9e07-42a3-a149-5aa13e857523 · outbound

This paper cites Thyme: Think Beyond Images.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Thyme: Think Beyond Images

Reference 64

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 7d601fa2-3a86-448c-abc2-b676d0b0ec21 · outbound

This paper cites Skywork-r1v4: Toward agentic multimodal intelligence through interleaved thinking with images and deepresearch.arXiv preprint arXiv:2512.02395, 2025.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Skywork-r1v4: Toward agentic multimodal intelligence through interleaved thinking with images and deepresearch.arXiv preprint arXiv:2512.02395, 2025

Reference 65

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation b43cb2e3-f411-45ce-8550-35ee74138446 · outbound

This paper cites Pyvision-rl: Forging open agentic vision models via rl.arXiv preprint arXiv:2602.20739, 2026.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Pyvision-rl: Forging open agentic vision models via rl.arXiv preprint arXiv:2602.20739, 2026

Reference 66

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 3fcb76a2-6f1f-473a-9f8e-f4d68862fa0d · outbound

This paper cites A stitch in time saves nine: Small vlm is a precise guidance for accelerating large vlms.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning A stitch in time saves nine: Small vlm is a precise guidance for accelerating large vlms

Reference 67

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 52934f59-9efa-4e84-a9a9-b67ee85806c3 · outbound

This paper cites DeepEyes: Incentivizing "Thinking with Images" via Reinforcement Learning.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning DeepEyes: Incentivizing "Thinking with Images" via Reinforcement Learning

Reference 68

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 99bea7b0-4e6a-41ce-8900-29b07aa04b48 · outbound

This paper cites Hierarchical Skip Decoding for Efficient Autoregressive Text Generation.

SpecEyes: Accelerating Agentic Multimodal LLMs via Speculative Perception and Planning Hierarchical Skip Decoding for Efficient Autoregressive Text Generation

Reference 69

Resolution: unresolved; no resolver link, observed 2026-07-13T19:34:58.789459Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Coverage vector

Reference resolution

Outbound references

Pith citing papers