ExFusion: Efficient Transformer Training via Multi-Experts Fusion

ExFusion: Efficient Transformer Training via Multi-Experts Fusion | Signal Canvas | ScienceToStartup

Page Freshness

Signal Canvas proof surface

Canonical route: /signal-canvas/exfusion-efficient-transformer-training-via-multi-experts-fusion

stale

Proof freshness: stale
Proof status: unverified
Display score: 4/10
Last proof check: 2026-03-31
Score updated: 2026-04-02
Score fresh until: 2026-05-02
References: 65
Source count: 3
Coverage: 50%

This page is showing the last landed evidence receipt and score bundle because the latest proof data is outside the freshness window.

Agent Handoff

Canonical ID exfusion-efficient-transformer-training-via-multi-experts-fusion | Route /signal-canvas/exfusion-efficient-transformer-training-via-multi-experts-fusion

REST example

curl https://sciencetostartup.com/api/v1/agent-handoff/signal-canvas/exfusion-efficient-transformer-training-via-multi-experts-fusion

MCP example

{
  "tool": "search_signal_canvas",
  "arguments": {
    "mode": "paper",
    "paper_ref": "exfusion-efficient-transformer-training-via-multi-experts-fusion",
    "query_text": "Summarize ExFusion: Efficient Transformer Training via Multi-Experts Fusion"
  }
}

source_context

{
  "surface": "signal_canvas",
  "mode": "paper",
  "query": "ExFusion: Efficient Transformer Training via Multi-Experts Fusion",
  "normalized_query": "2603.27965",
  "route": "/signal-canvas/exfusion-efficient-transformer-training-via-multi-experts-fusion",
  "paper_ref": "exfusion-efficient-transformer-training-via-multi-experts-fusion",
  "topic_slug": null,
  "benchmark_ref": null,
  "dataset_ref": null
}

Evidence Receipt

Route status: building

Claims: 7

References: 65

Proof: Verification pending

Freshness state: computing

Source paper: ExFusion: Efficient Transformer Training via Multi-Experts Fusion

PDF: https://arxiv.org/pdf/2603.27965v1

Source count: 3

Coverage: 50%

Last proof check: 2026-03-31T20:25:07.388Z

Signal Canvas receipt window

Not build-ready: ExFusion: Efficient Transformer Training via Multi-Experts Fusion

/buildability/exfusion-efficient-transformer-training-via-multi-experts-fusion

Ignoreblocked

Subject: ExFusion: Efficient Transformer Training via Multi-Experts Fusion

Verdict

Ignore

Verdict is Ignore because current viability and proof state do not clear the buildability gate.

Preparing verified analysis

GitHub Code Pulse

No public code linked for this paper yet.

Claim map

Strong 7Mixed 0Weak 0

Evidencepartial
ExFusion introduces multi-expert characteristics into the training process while incurring only marginal computational cost compared to standard dense training.
Implicationpartial
Directly stated in the abstract with clear methodology description
Verificationpartial
partial
Evidencepartial
After training, the learned weights are used to integrate multi-experts into a single unified expert, thereby eliminating additional overhead in storage and deployment.
Implicationpartial
Explicitly stated in the abstract with clear mechanism description
Verificationpartial
partial
Evidencepartial
For example, SwinV2-MoE-S (128 experts) [15], trained on a large-scale dataset (e.g., ImageNet-22k [16]), surpasses SwinV2-S by roughly 1% on the ImageNet-1k benchmark [17]. However, this improvement requires nearly a 30× increase in parameters
Implicationpartial
Directly stated with specific example comparing SwinV2-MoE-S to SwinV2-S
Verificationpartial
partial
Evidencepartial
Specifically, during the initialization phase, ExFusion upcycles the feed-forward network (FFN) of the Transformer into a multi-expert configuration
Implicationpartial
Directly stated in the abstract with clear technical description
Verificationpartial
partial
Evidencepartial
Extensive experiments on a variety of computer vision and natural language processing tasks demonstrate the effectiveness of the proposed method.
Implicationpartial
Directly stated in abstract with supporting experimental sections
Verificationpartial
partial
Evidencepartial
During training, these weights allow multiple experts to be fused into a single unified expert equivalent to the original FFN, which is subsequently used for forward computation.
Implicationpartial
Directly stated in the abstract with clear operational mechanism
Verificationpartial
partial
Evidencepartial
Could we harness the benefits of MoE during training yet retain a parameter count and computational complexity comparable to dense models, all without a significant rise in training expenses?
Implicationpartial
Directly stated in the methodology section as the core research question
Verificationpartial
partial

Author intelligence and commercialization panels stay hidden until the proof receipt is verified, cites at least 3 references, includes at least 2 sources, and clears 50% coverage. The paper narrative and citation surfaces remain public while verification is pending.

ExFusion: Efficient Transformer Training via Multi-Experts Fusion

Use Signal Canvas as the narrative proof surface

Use this Signal Canvas via API or MCP

Signal Canvas proof surface