From SWE-ZERO to SWE-HERO: Execution-free to Execution-based Fine-tuning for Software Engineering Agents

From SWE-ZERO to SWE-HERO: Execution-free to Execution-based Fine-tuning for Software Engineering Agents | Signal Canvas | ScienceToStartup

Page Freshness

Signal Canvas proof surface

Canonical route: /signal-canvas/from-swe-zero-to-swe-hero-execution-free-to-execution-based-fine-tuning-for-software-engineering-agents

stale

Proof freshness: stale
Proof status: unverified
Display score: 8/10
Last proof check: 2026-04-03
Score updated: 2026-04-03
Score fresh until: 2026-05-03
References: 0
Source count: 0
Coverage: 33%

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 from-swe-zero-to-swe-hero-execution-free-to-execution-based-fine-tuning-for-software-engineering-agents | Route /signal-canvas/from-swe-zero-to-swe-hero-execution-free-to-execution-based-fine-tuning-for-software-engineering-agents

REST example

curl https://sciencetostartup.com/api/v1/agent-handoff/signal-canvas/from-swe-zero-to-swe-hero-execution-free-to-execution-based-fine-tuning-for-software-engineering-agents

MCP example

{
  "tool": "search_signal_canvas",
  "arguments": {
    "mode": "paper",
    "paper_ref": "from-swe-zero-to-swe-hero-execution-free-to-execution-based-fine-tuning-for-software-engineering-agents",
    "query_text": "Summarize From SWE-ZERO to SWE-HERO: Execution-free to Execution-based Fine-tuning for Software Engineering Agents"
  }
}

source_context

{
  "surface": "signal_canvas",
  "mode": "paper",
  "query": "From SWE-ZERO to SWE-HERO: Execution-free to Execution-based Fine-tuning for Software Engineering Agents",
  "normalized_query": "2604.01496",
  "route": "/signal-canvas/from-swe-zero-to-swe-hero-execution-free-to-execution-based-fine-tuning-for-software-engineering-agents",
  "paper_ref": "from-swe-zero-to-swe-hero-execution-free-to-execution-based-fine-tuning-for-software-engineering-agents",
  "topic_slug": null,
  "benchmark_ref": null,
  "dataset_ref": null
}

Evidence Receipt

Route status: building

Claims: 8

References: Pending verification

Proof: Verification pending

Freshness state: computing

Source paper: From SWE-ZERO to SWE-HERO: Execution-free to Execution-based Fine-tuning for Software Engineering Agents

PDF: https://arxiv.org/pdf/2604.01496v1

Source count: Pending verification

Coverage: 33%

Last proof check: 2026-04-03T20:50:41.059Z

Signal Canvas receipt window

Watch and verify: From SWE-ZERO to SWE-HERO: Execution-free to Execution-based Fine-tuning for Software Engineering Agents

/buildability/from-swe-zero-to-swe-hero-execution-free-to-execution-based-fine-tuning-for-software-engineering-agents

Watchwatch

Subject: From SWE-ZERO to SWE-HERO: Execution-free to Execution-based Fine-tuning for Software Engineering Agents

Verdict

Watch

Verdict is Watch because viability or proof quality is intermediate and should be re-evaluated before execution.

Preparing verified analysis

GitHub Code Pulse

No public code linked for this paper yet.

Claim map

Strong 8Mixed 0Weak 0

Evidencepartial
We introduce SWE-ZERO to SWE-HERO, a two-stage SFT recipe that achieves state-of-the-art results on SWE-bench by distilling open-weight frontier LLMs.
Implicationpartial
Explicitly stated in the abstract as the main achievement of the paper
Verificationpartial
partial
Evidencepartial
Notably, SWE-HERO-32B achieves a 62.2% resolution rate on SWE-bench Verified.
Implicationpartial
Specific numeric result directly stated in the abstract
Verificationpartial
partial
Evidencepartial
Our pipeline replaces resource-heavy dependencies with an evolutionary refinement strategy
Implicationpartial
Directly stated in the abstract as a key methodological innovation
Verificationpartial
partial
Evidencepartial
SWE-ZERO utilizes large-scale, execution-free trajectories to master code semantics and repository-level reasoning
Implicationpartial
Explicitly described as the first stage of the method in the abstract
Verificationpartial
partial
Evidencepartial
SWE-HERO applies targeted, execution-backed refinement to transition these semantic intuitions into rigorous engineering workflows.
Implicationpartial
Explicitly described as the second stage of the method in the abstract
Verificationpartial
partial
Evidencepartial
Furthermore, despite being trained exclusively on Python, our agents demonstrate robust zero-shot transferability on SWE-bench Multilingual, reaching 44.1% and confirming the paradigm's generalizability across diverse languages.
Implicationpartial
Specific numeric result and transferability claim directly stated in the abstract
Verificationpartial
partial
Evidencepartial
We release a dataset of 300k SWE-ZERO and 13k SWE-HERO trajectories distilled from Qwen3-Coder-480B
Implicationpartial
Specific dataset statistics directly stated in the abstract
Verificationpartial
partial
Evidencepartial
Our empirical results set a new benchmark for open-source models of comparable size.
Implicationpartial
Directly stated in the abstract as a key contribution
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.

From SWE-ZERO to SWE-HERO: Execution-free to Execution-based Fine-tuning for Software Engineering Agents

Use Signal Canvas as the narrative proof surface

Use this Signal Canvas via API or MCP

Signal Canvas proof surface