Skip to content
Navigation Menu
Sign in
Products
Platform
GitHub
Repository hosting, code review, and collaboration
Copilot
AI coding agent for tasks, projects, and automated work
Build & Ship
Actions
CI/CD and automation
Code Quality
Quality analysis, fixes, and enforcement
Codespaces
Cloud development environments
Secure
Code Security
Vulnerability detection and remediation
Secret Protection
Secret scanning and leak prevention
All Features
Developers
Learn
Documentation
Training
Certifications
Build
Build with Copilot
Customize Copilot
Integrations
Explore
Trending repos
GitHub Sponsors
Developer articles
Changelog
Changelog
Your source for the latest features, improvements, and fixes across GitHub products, all in one place.
Businesses
By Stage
Enterprises
Team
Startups
Nonprofits
By Industry
Healthcare
Financial services
Manufacturing
Government
See all industries
Evaluate GitHub
Why GitHub
Customer Stories
Business insights
Trust Center
Add-ons
GitHub Advanced Security
Enterprise-grade code security
Copilot for business
Enterprise-grade AI features
Premium Support
Enterprise-grade 24/7 support
Resources
Learn & Connect
Documentation
Resource center
Blog
Events & webinars
Community Forum
Customer Stories
Integrations
Support
Customer support
Training
GitHub Status
Company
About
Partners
Changelog
GitHub Universe
GitHub Universe
The event for developers, by developers — talks, workshops, and what’s next for GitHub.
Pricing
Search
/
Sign in
Sign up
Appearance settings
You signed in with another tab or window.
Reload
to refresh your session.
You signed out in another tab or window.
Reload
to refresh your session.
You switched accounts on another tab or window.
Reload
to refresh your session.
Dismiss alert
{{ message }}
neuralnet-cpp-dev
neuralnet.cpp
Repository navigation
Code
Issues
2
(2)
Pull requests
Discussions
Actions
Projects
Security and quality
Insights
More
items
融合：精度签名降级为运行期参数 + nn_enable_gpu_fusion（叠在 #14 之上）
- #16
#16
Merged
EthanPeng-2048
merged 30 commits into
main
neuralnet-cpp-dev/neuralnet.cpp:main
from
feat/fused-runtime-prec-dispatch
neuralnet-cpp-dev/neuralnet.cpp:feat/fused-runtime-prec-dispatch
Copy head branch name to clipboard
Oct 1, 2026
Conversation
Commits
30
(30)
Checks
Files changed
Merged
融合：精度签名降级为运行期参数 + nn_enable_gpu_fusion（叠在 #14 之上）
#16
EthanPeng-2048
merged 30 commits into
main
neuralnet-cpp-dev/neuralnet.cpp:main
from
feat/fused-runtime-prec-dispatch
neuralnet-cpp-dev/neuralnet.cpp:feat/fused-runtime-prec-dispatch
Copy head branch name to clipboard
Commits
Commits on Oct 1, 2026
融合 AOT 步骤①：常量池值退出 expr_spec_key（key 只描述结构）
Show description for 3cf591c
EthanPeng-2048
committed
3cf591c
View commit details
Copy full SHA for 3cf591c
Browse repository at this point
融合 AOT 步骤②：matmul 转置退出 expr_spec_key（运行期 operand layout）
Show description for b960450
EthanPeng-2048
committed
b960450
View commit details
Copy full SHA for b960450
Browse repository at this point
融合 AOT 步骤③：分组归约 R 退出 expr_spec_key（运行期循环）
Show description for 148d1d7
EthanPeng-2048
committed
148d1d7
View commit details
Copy full SHA for 148d1d7
Browse repository at this point
融合 AOT 步骤⑤前置：符号张量不再触发 CPU 视图缓存断言
Show description for 31051a3
EthanPeng-2048
committed
31051a3
View commit details
Copy full SHA for 31051a3
Browse repository at this point
融合 AOT 步骤⑤：dsl::compute 自登记锚点（结构覆盖不再依赖手写清单）
Show description for 41b4374
EthanPeng-2048
committed
41b4374
View commit details
Copy full SHA for 41b4374
Browse repository at this point
融合 AOT 步骤⑥（修订）：锚点覆盖台账 + 签名缺口报告
Show description for 63b7d72
EthanPeng-2048
committed
63b7d72
View commit details
Copy full SHA for 63b7d72
Browse repository at this point
融合 AOT 步骤⑥·补：模型级签名 pass（run-only 签名 7/7 → 2/7）
Show description for 03f9383
EthanPeng-2048
committed
03f9383
View commit details
Copy full SHA for 03f9383
Browse repository at this point
融合 AOT 步骤⑥·补2：模型 pass 配置矩阵 + 分项来源台账
Show description for 484ae35
EthanPeng-2048
committed
484ae35
View commit details
Copy full SHA for 484ae35
Browse repository at this point
融合 AOT 步骤⑥·补3：删除 13 个已被模型 pass 覆盖的 per-layer dry-run 块（−223 行）
Show description for 98f887a
EthanPeng-2048
committed
98f887a
View commit details
Copy full SHA for 98f887a
Browse repository at this point
run-only 签名回填：fold · 文档掩码 f16 变体（--f16 --doc-mask 的 miss）
Show description for 226b80f
EthanPeng-2048
committed
226b80f
View commit details
Copy full SHA for 226b80f
Browse repository at this point
run-only 签名回填：MNIST/CNN f16 训练的 8 条 miss（含 per-layer 块删除引入的 2 条）
Show description for 145a267
EthanPeng-2048
committed
145a267
View commit details
Copy full SHA for 145a267
Browse repository at this point
AOT 融合 key 契约测试：把"结构 = 表达式类型"锁进 CI
Show description for 0c68197
EthanPeng-2048
committed
0c68197
View commit details
Copy full SHA for 0c68197
Browse repository at this point
AOT 融合构建期收为单步：gen_fused 并入 scan_exprs，删除 bin 序列化
Show description for 0bdaa97
EthanPeng-2048
committed
0bdaa97
View commit details
Copy full SHA for 0bdaa97
Browse repository at this point
文档修正：去掉对已删除文件 gen_fused.cpp 的当前态引用
Show description for 2c29e02
EthanPeng-2048
committed
2c29e02
View commit details
Copy full SHA for 2c29e02
Browse repository at this point
run-only 签名回填：text_train（tinystories smoke）× f16 的 2 条 miss
Show description for b48dd5a
EthanPeng-2048
committed
b48dd5a
View commit details
Copy full SHA for b48dd5a
Browse repository at this point
锚点价值运行时验证 + 闭合世界报错带 key
Show description for 9300bf8
EthanPeng-2048
committed
9300bf8
View commit details
Copy full SHA for 9300bf8
Browse repository at this point
文档同步：AGENTS.md 跟进 AOT 融合现状 + 13 号清单归档入库
Show description for 06157d5
EthanPeng-2048
committed
06157d5
View commit details
Copy full SHA for 06157d5
Browse repository at this point
A1a: emitter 增加运行期精度分派（V1）发射模式
Show description for aba9ecb
EthanPeng-2048
committed
aba9ecb
View commit details
Copy full SHA for aba9ecb
Browse repository at this point
A1b-d: 运行期精度分派落地——签名从「身份」降级为「参数」
Show description for 2b43846
EthanPeng-2048
committed
2b43846
View commit details
Copy full SHA for 2b43846
Browse repository at this point
A1e: 删除签名回填工作流与 f16 扫两遍（扫描工作量减半）
Show description for 6c9e091
EthanPeng-2048
committed
6c9e091
View commit details
Copy full SHA for 6c9e091
Browse repository at this point
A3: 文档同步「签名从身份降级为参数」+ 历史归档
Show description for 94d5c91
EthanPeng-2048
committed
94d5c91
View commit details
Copy full SHA for 94d5c91
Browse repository at this point
A2: 库自带融合能力 nn_enable_gpu_fusion + 修复多 TU 链接
Show description for 476e057
EthanPeng-2048
committed
476e057
View commit details
Copy full SHA for 476e057
Browse repository at this point
A1f: 删除精度变体残留机制（注册表 variants / 死分支）
Show description for f334cd2
EthanPeng-2048
committed
f334cd2
View commit details
Copy full SHA for f334cd2
Browse repository at this point
A2b: 样例补 GPU 端到端验收（含 f16 分派路径）+ 修 CMake include 顺序
Show description for a2c90fb
EthanPeng-2048
committed
a2c90fb
View commit details
Copy full SHA for a2c90fb
Browse repository at this point
诊断：`NN_SCAN_DUMP_SOURCES=1` 逐条打印结构来源 + 锚点/dry-run 边界实测
Show description for 7c3af60
EthanPeng-2048
committed
7c3af60
View commit details
Copy full SHA for 7c3af60
Browse repository at this point
A2c: 扩展来源诊断（含指令序列）+ 实测第二种自定义算子形态
Show description for bf8d10e
EthanPeng-2048
committed
bf8d10e
View commit details
Copy full SHA for bf8d10e
Browse repository at this point
A4: 把 A2 端到端验收注册为常驻 ctest（ctest 19 → 20）
Show description for 371ffb9
EthanPeng-2048
committed
371ffb9
View commit details
Copy full SHA for 371ffb9
Browse repository at this point
A5: 接入孤儿测试 `fused_gpu_test`（ctest 20 → 21）
Show description for d0c9609
EthanPeng-2048
committed
d0c9609
View commit details
Copy full SHA for d0c9609
Browse repository at this point
A6: 修 `NN_PREC_TRACE` 对 V1 的盲区 + 实测 V1 覆盖（并更正上一轮的错误结论）
Show description for 17381ab
EthanPeng-2048
committed
17381ab
View commit details
Copy full SHA for 17381ab
Browse repository at this point
审核：修 9 处真缺陷 + 消除 4 类"两处同步"点（融合 AOT 分支收尾）
Show description for 14892f5
EthanPeng-2048
committed
14892f5
View commit details
Copy full SHA for 14892f5
Browse repository at this point
You can’t perform that action at this time.
