Review economics
deepspeedai/DeepSpeed
Read from the 100 most recently merged pull requests ·
Pull requests with attributed agent authorship took 1.2× the reviews submitted of the rest.
≥9%
Attributed
a floor
8
Attributed PRs
92
Other PRs
100
Read
Attributed — the latest 5
#8203 Unmanaged gradient accumulation: ZeRO stage 2 support
4 reviews · 3.7d to merge
#8212 Share DeepNVMe pinned-tensor manager and route swap buffers through I/O handles
6 reviews · 22h to merge
#8195 MoE dispatch: fix silent Tutel + TP corruption, speed up native and Tutel paths
7 reviews · 4.1d to merge
#8184 Add managed_gradient_accumulation for ZeRO stage 0/1
2 rounds · 10 reviews · 6.1d to merge
#8200 Skip CUDA transformer forward tests when TransformerBuilder is incompatible
2 reviews · 2h to merge
The rest — the latest 5
#8199 Stop trim_mean sorting the caller's list in place
2 reviews · 4.9d to merge
#8178 Make PipelineModule.set_checkpoint_interval actually change the interval
1 round · 3 reviews · 10.1d to merge
#8209 Reduce redundant work in AutoEP token routing
1 reviews · 9h to merge
#8208 Fix MLU_Accelerator conformance to the DeepSpeedAccelerator ABC
1 reviews · 12h to merge
#8206 Consolidate LR scheduler tensor-LR tests into a contract suite
2 reviews · 22h to merge
How this was measured
9% of merged commits carry agent attribution — a floor, not the share; tools that only complete code inline leave no commit trail, so the unattributed side includes AI-assisted work; reads 100 of 3140 merged PRs — the rest have not been analyzed yet.
Detected: Cursor agent, GitHub Copilot coding agent, Claude Code.
“Attributed” means a commit carried an agent’s signature — a co-author trailer, an agent commit identity, or an agent bot account. Tools that only complete code inline leave no such mark, so the other column is “rest”, not “human-written”. Full method and its limits
The badge reads this repo’s current report, so it follows the number.
Tell me when this moves
We re-read deepspeedai/DeepSpeed weekly and email only when the number changes materially.
Read your own repos
This one is public. For a private repo, run the same read locally through your own GitHub credentials — nothing is installed and nothing is sent to us.
npx @ambera/review-taxWant this continuously on deepspeedai/DeepSpeed — each pull request paired to the task it came from, and the work graded from its review loop rather than its diff size? Claim this repo in Forge
Browse every repo people have read · Read a different repository