Files with the highest combination of change frequency and waste ratio. These are candidates for refactoring or closer review.
Cumulative contribution over time. Watch developers race as positions shift month by month.
Commit activity distribution by hour and day of week across all contributors in this repository.
Performance has many faces. Navigara breaks down the effort to visualize what parts of codebase has been changed and where energy flowed. Our Architect AI can break the performance even further into particular components and patterns.
Breakdown of file changes over time. Play the timeline to see how change types evolved across periods.
Monthly overview of bugs introduced and fixed, based on symbol-level commit analysis. Fixes show whether the original author fixed their own bug (self-fix) or someone else did (cross-fix).
Bug attribution uses symbol-level matching from commit history. For each fix commit, we look at the changed symbols (functions, classes, methods) and trace backwards to find who last modified that symbol in a non-fix commit. This person is the probable bug introducer. The algorithm only works when commits have symbol-level data from the Navigara analysis engine โ the coverage rate shows what percentage of fix commits had this data available.
The current metrics model has a semantic inversion: when developer A creates a feature with a bug, they receive grow (positive). When developer B fixes that bug, they receive waste (negative). The bug creator is rewarded while the fixer is penalized. Bug attribution addresses this by explicitly tracking who introduced bugs and who fixed them, providing a more accurate picture of code quality contributions.
Currently computed client-side from commit data. Ideal server-side endpoint:
POST /v1/repositories/{repositoryId}/bug-attributions
Content-Type: application/json
Request:
{
"startTime": "2025-01-01T00:00:00Z",
"endTime": "2025-12-31T23:59:59Z"
}
Response:
{
"totalBugsAttributed": 42,
"selfFixRate": 35,
"coverageRate": 78,
"attributions": [
{
"filePath": "src/lib/auth.ts",
"symbol": "validateToken",
"introducer": { "name": "Alice", "email": "alice@co.com", "commitSha": "abc123" },
"fixer": { "name": "Bob", "email": "bob@co.com", "commitSha": "def456" },
"fixedAt": "2025-06-15T10:30:00Z",
"isSelfFix": false
}
]
}Reclassifies engineering effort based on bug attribution. Commits that introduced bugs are retrospectively counted as poor investments.
Investment Quality reclassifies engineering effort based on bug attribution data. Commits identified as buggy origins (those that introduced bugs later fixed by someone) have their grow and maintenance time moved into the Wasted Time category. Their waste (fix commits) remains counted as productive. All other commits retain their standard classification: grow is productive, maintenance is maintenance, and waste (fixes) is productive.
The standard model classifies commits as Growth, Maintenance, or Fixes. Investment Quality adds a quality lens: a commit that introduced a bug is retrospectively counted as a poor investment โ the engineering time spent on it was wasted because it ultimately required additional fix work. Fix commits (Fixes in the standard model) are reframed as productive, because fixing bugs is valuable work.
Currently computed client-side from commit and bug attribution data. Ideal server-side endpoint:
POST /v1/organizations/{orgId}/investment-quality
Content-Type: application/json
Request:
{
"startTime": "2025-01-01T00:00:00Z",
"endTime": "2025-12-31T23:59:59Z",
"bucketSize": "BUCKET_SIZE_MONTH",
"groupBy": ["repository_id" | "deliverer_email"]
}
Response:
{
"productivePct": 74,
"maintenancePct": 18,
"wastedPct": 8,
"buckets": [
{
"bucketStart": "2025-01-01T00:00:00Z",
"productive": 4.2,
"maintenance": 1.8,
"wasted": 0.6
}
]
}Latest analyzed commits in this repository.
| Hash | Message | Author | Effort |
|---|---|---|---|
| 5dce124 | Zero3 defragment utility (#7940) | nathon | maint |
| 2f0924a | Fix process hang in process-group shutdown (#7941) | Alexander Grund | maint |
| 3bdebc0 | Fix/fix autotp universal checkpoint ci (#7937) | Masahiro Tanaka | waste |
| 89bf0d2 | Refactor consolidate transpose (#7934) | nathon | maint |
| 607b55f | Update version after latest release (v0.18.9) (#7936) | Logan Adams | maint |
| 8c93851 | Update version.txt for latest incoming release 0.18.9 (#7935) | Logan Adams | maint |
| 9486ab7 | Remove Microsoft Corporation copyright from AGENTS.md and CLAUDE.md (#7932) | Zhipeng Wang | โ |
| 36f0b0c | Add fallback to full test (#7933) | Masahiro Tanaka | grow |
| 5efb24a | Merging AutoSP into DeepSpeed (#7860) | Neel Dani | grow |
| 2bae360 | Update README with latest news from DeepSpeed (#7931) | Zhipeng Wang | maint |
Zero3 defragment utility (#7940)
Fix process hang in process-group shutdown (#7941)
Fix/fix autotp universal checkpoint ci (#7937)
Refactor consolidate transpose (#7934)
Update version after latest release (v0.18.9) (#7936)
Update version.txt for latest incoming release 0.18.9 (#7935)
Remove Microsoft Corporation copyright from AGENTS.md and CLAUDE.md (#7932)
Add fallback to full test (#7933)
Merging AutoSP into DeepSpeed (#7860)
Update README with latest news from DeepSpeed (#7931)
Average context complexity and engagement score of file changes over time. Higher complexity means more intricate changes; higher impact means broader effect on the codebase.
Repository
DeepSpeed
DeepSpeed is a deep learning optimization library that makes distributed training and inference easy, efficient, and effective.
Average Developer Performance (ETV)
Year-by-year Trend:+330%Contributors ranked by total performance (ETV) from analyzed commits.
| # | |||||
|---|---|---|---|---|---|
| 1 | Masahiro Tanaka71 commits | 8.8 | 2.6 | 4.6 | 1.7 |
| 2 |  Stas Bekman52 commits | 3.6 | 1.1 | 2 | 0.5 |
| 3 | M mingzhi.liu9 commits | 1.9 | 1.2 | 0.7 | 0 |
| 4 | C cyyever1 commits | 1.9 | 0 | 0.1 | 1.8 |
| 5 | O olruwase10 commits | 1.8 | 0.1 | 1.6 | 0.2 |
| 6 |  Ma, Guokai18 commits | 1.8 | 0.3 | 1.2 | 0.3 |
| 7 | E erc8gx2 commits | 1.6 | 1.2 | 0.4 | 0 |
| 8 | J junjie.mao14 commits | 1.6 | 0.6 | 0.5 | 0.5 |
| 9 | Logan Adams70 commits | 1.4 | 0 | 1.2 | 0.2 |
| 10 | T tjruwase5 commits | 1.3 | 0.5 | 0.8 | 0.1 |
| 11 | L lian78 commits | 1.2 | 0.6 | 0.1 | 0.5 |
| 12 | N nsonnenschein4 commits | 1.1 | 0 | 0 | 1 |
| 13 | N neeldani981 commits | 1.1 | 0.6 | 0.4 | 0 |
| 14 | L leejianwoo9 commits | 1 | 0.3 | 0.6 | 0.1 |
| 15 | 1 132897193+joshwoo20031 commits | 0.8 | 0.5 | 0.3 | 0 |
| 16 | 1 167664334+haosenwang10181 commits | 0.8 | 0 | 0 | 0.8 |
| 17 | H hollowman11 commits | 0.8 | 0 | 0.1 | 0.7 |
| 18 | T tunji.ruwase15 commits | 0.7 | 0.4 | 0.2 | 0.1 |
| 19 | S siqi2023111 commits | 0.7 | 0.7 | 0 | 0 |
| 20 | 1 142979319+oelayan75 commits | 0.7 | 0.4 | 0.2 | 0.1 |
| 21 | Z zhengchenyu165 commits | 0.6 | 0 | 0.2 | 0.4 |
| 22 | S sisodiarakshit4565 commits | 0.6 | 0 | 0.1 | 0.5 |
| 23 | Z zhipeng.rainbowserie10 commits | 0.6 | 0.4 | 0.2 | 0 |
| 24 | S sdvillal2 commits | 0.6 | 0.1 | 0 | 0.5 |
| 25 | 3 33092912+hwchen20175 commits | 0.5 | 0.4 | 0.1 | 0 |
| 26 | H he.yujiao1 commits | 0.4 | 0.3 | 0.1 | 0 |
| 27 | B bruno.magalhaes1 commits | 0.4 | 0.3 | 0 | 0 |
| 28 | F fastrunner100901 commits | 0.4 | 0 | 0.1 | 0.3 |
| 29 | T tanggf1 commits | 0.3 | 0 | 0 | 0.3 |
| 30 | M mailofphalani1 commits | 0.3 | 0.2 | 0.1 | 0 |
| 31 | M mkovalenko14 commits | 0.3 | 0.1 | 0.1 | 0.1 |
| 32 | M mauryaavinash951 commits | 0.3 | 0.2 | 0.1 | 0 |
| 33 | T therealnaveenkamal4 commits | 0.3 | 0 | 0.1 | 0.2 |
| 34 | F flamefire4 commits | 0.2 | 0.2 | 0 | 0 |
| 35 | V vensenmu4 commits | 0.2 | 0 | 0.1 | 0.2 |
| 36 | 7 70597027+lymdlut1 commits | 0.2 | 0 | 0.2 | 0 |
| 37 | 4 44502768+rezayazdaniaminabadi1 commits | 0.2 | 0 | 0 | 0.2 |
| 38 | A artem.kuzmitckii5 commits | 0.2 | 0.1 | 0.1 | 0.1 |
| 39 | 8 83907395+bias921 commits | 0.2 | 0.2 | 0 | 0 |
| 40 | 6 62723901+rraminen5 commits | 0.2 | 0 | 0.1 | 0.1 |
| 41 | 1 118277438+fabiosanger1 commits | 0.2 | 0 | 0.1 | 0.1 |
| 42 | Y yejing.lai7 commits | 0.1 | 0.1 | 0 | 0 |
| 43 | Q qganthony1 commits | 0.1 | 0 | 0 | 0.1 |
| 44 | 5 56556451+alexk1012 commits | 0.1 | 0.1 | 0 | 0 |
| 45 | H harikp20021 commits | 0.1 | 0 | 0 | 0.1 |
| 46 | M mtanaka1 commits | 0.1 | 0 | 0.1 | 0 |
| 47 | 1 113481193+lekurile2 commits | 0.1 | 0 | 0 | 0.1 |
| 48 | 4 45323446+u-rara1 commits | 0.1 | 0 | 0 | 0.1 |
| 49 | 3 35650601+saurabhkoshatwar1 commits | 0.1 | 0 | 0.1 | 0 |
| 50 | 3 39188949+ksugama1 commits | 0.1 | 0 | 0.1 | 0 |
| 51 | T tafflann1 commits | 0.1 | 0 | 0.1 | 0 |
| 52 | Y yi.sheng1 commits | 0.1 | 0 | 0.1 | 0 |
| 53 | K krishnabkc151 commits | 0.1 | 0 | 0 | 0.1 |
| 54 | 5 55744355+huanyuqu1 commits | 0.1 | 0 | 0 | 0.1 |
| 55 | S srsikander4 commits | 0.1 | 0 | 0.1 | 0 |
| 56 | 1 165311345+loscrossos2 commits | 0.1 | 0 | 0 | 0.1 |
| 57 | 4 45557362+qgallouedec2 commits | 0.1 | 0 | 0.1 | 0 |
| 58 | A ayushtanwar17291 commits | 0.1 | 0 | 0 | 0.1 |
| 59 | Y yifan06101 commits | 0.1 | 0 | 0 | 0.1 |
| 60 | E emmanuelferdman2 commits | 0.1 | 0 | 0.1 | 0 |
| 61 | L limjcst1 commits | 0.1 | 0.1 | 0 | 0 |
| 62 | F fabiendupont1 commits | 0.1 | 0 | 0.1 | 0 |
| 63 | J jerasley1 commits | 0.1 | 0 | 0 | 0.1 |
| 64 | A aeeeeeep2 commits | 0.1 | 0 | 0 | 0 |
| 65 | D digger-yu1 commits | 0.1 | 0 | 0 | 0.1 |
| 66 | Q qibin05061 commits | 0.1 | 0 | 0 | 0.1 |
| 67 | X xiongjyu1 commits | 0.1 | 0 | 0 | 0 |
| 68 | H himanshuwindows8.11 commits | 0.1 | 0 | 0 | 0.1 |
| 69 | A alexwgh3331 commits | 0.1 | 0 | 0.1 | 0 |
| 70 | F ffheyy00171 commits | 0.1 | 0 | 0 | 0.1 |
| 71 | M metarufolds1 commits | 0.1 | 0 | 0 | 0.1 |
| 72 | 1 17906905141 commits | 0.1 | 0 | 0 | 0.1 |
| 73 | C cl57435909212 commits | 0.1 | 0 | 0 | 0 |
| 74 | 4 45830328+michaelroyzen1 commits | 0.1 | 0 | 0 | 0.1 |
| 75 | R romaactor1 commits | 0.1 | 0 | 0 | 0.1 |
| 76 | W woctordho1 commits | 0.1 | 0 | 0 | 0.1 |
| 77 | G giulio97.leone1 commits | 0.1 | 0 | 0.1 | 0 |