8000 Advance development to 0.11.0.dev0 and freeze rollout baseline by DaoyuanLi2816 · Pull Request #91 · DaoyuanLi2816/mini-verl · GitHub
[go: up one dir, main page]
Skip to content

Advance development to 0.11.0.dev0 and freeze rollout baseline #91

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
DaoyuanLi2816 merged 2 commits into from
Aug 29, 2026
Merged

Advance development to 0.11.0.dev0 and freeze rollout baseline #91

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
0d6e007
Preregister Rollout Runtime v2 baseline
DaoyuanLi2816 Aug 29, 2026
8e5ebda
Freeze the HF rollout reference baseline
DaoyuanLi2816 Aug 29, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
8000
Diff view
Diff view
9 changes: 9 additions & 0 deletions CHANGELOG.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6,6 +6,15 @@ All notable changes to miniVERL are recorded here. The format follows

## [Unreleased]

### Rollout Runtime v2 baseline

- Advanced the development line to `0.11.0.dev0` and preregistered the fixed
RTX 4080 rollout workload before changing generation behavior.
- Added an exact-source competitive landscape that separates upstream claims,
miniVERL's current state and measurements that have not yet been run.
- Froze the 24-cell pre-v0.11 `hf_reference` actor-rollout baseline on the WSL2
RTX 4080, with exact wheel, weight, policy and raw-measurement identity.

### Documentation narrative

- Rebuilt the English, Chinese and PyPI landing narratives around the local
Expand Down
8 changes: 4 additions & 4 deletions PROJECT_STATE.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4,14 +4,14 @@ Current maintainer handoff for **miniVERL** (`mini-verl` package, `miniverl`
CLI). `release-state.yaml` is the canonical version source; this page indexes
current product and evidence state rather than repeating release history.

Last updated: 2026-08-16.
Last updated: 2026-08-29.

Canonical release state: stable `v0.10.1` (`2364e9b8e8b550f44d1b66a77fc2d407b76b05b5`), development `0.10.2.dev0`.
Canonical release state: stable `v0.10.1` (`2364e9b8e8b550f44d1b66a77fc2d407b76b05b5`), development `0.11.0.dev0`.

## Release state

- Stable release: `v0.10.1` at `2364e9b8e8b550f44d1b66a77fc2d407b76b05b5`.
- Development release: `0.10.2.dev0`.
- Development release: `0.11.0.dev0`.
- Stable docs: <https://daoyuanli2816.github.io/mini-verl/>.
- Development docs: <https://daoyuanli2816.github.io/mini-verl/dev/>.
- Historical build log: [v0.1-v0.9 archive](docs/history/project-state-v0.1-v0.9.md).
Expand All @@ -31,7 +31,7 @@ Executable compatibility claims are mutation-tested and recorded in
unknown-size quantized roles require proof instead of receiving an executable
plan.

Development `0.10.2.dev0` has a closed typed profile registry and torch-free
Development `0.11.0.dev0` has a closed typed profile registry and torch-free
compatibility introspection. Profile-scoped plans, caches, checkpoints and
exports bind an independent identity. The direct-GKD and sampled-k1 vanilla
policy-loss profiles both have pinned conformance and measured RTX 4080 paths.
Expand Down
1 change: 1 addition & 0 deletions benchmarks/README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -80,6 +80,7 @@ Listed so you can check that this repository does not.

| File | Hardware | What it shows |
| --- | --- | --- |
| `results/rollout-runtime-v2-hf-reference.json` | WSL2 RTX 4080 16 GB | Pre-v0.11 actor-rollout systems baseline: all 24 preregistered prompt/response/n/sampling cells completed. It freezes throughput and memory for the future `hf_cached` comparison; teacher scoring, updates, full-cycle timing and typed grouped semantics are not measured here. [Method, table and boundaries.](../docs/benchmarks/rollout-runtime-v2.md) |
| `results/alignment-external-v1.json` | RTX 4080 16 GB | **Preregistered early stop, not a method comparison.** Two declared lineages and eight candidates all measured 0/64 retained JSONNav utility against an unchanged 20% floor. No checkpoint, teacher, continuation arm or final-test result exists. [Study, correction provenance and evaluator boundary.](../docs/alignment-external/alignment-external-v1.md) |
| `results/alignment-lab-v1.json` | RTX 4080 16 GB | Alignment Lab v1: six methods from one Qwen3-0.6B SFT checkpoint, three seeds and 48 paired policy tasks per arm. The SFT start saturated alignment and utility; no continuation method improved it, so the pilot recommends no online teacher querying. [Report, figures and caveats.](../docs/alignment-lab/alignment-lab-v1.md) |
| `results/consumer-runtime-v1.json` | RTX 4080 16 GB | Preregistered eight-cell systems matrix for sequential/batch-2/batch-4/auto updates under dual-model and shared-backbone ownership. Batch-4 reached 3.866 and 3.475 trajectories/s at 3.035 and 2.227 GiB reserved; all equivalence gates passed. [Figure, method and caveats.](../docs/consumer-runtime-v1.md) |
Expand Down
Loading
0