Skip to content

[Feat] Score browser decision backends step by step on myjev-browser-decisions (offline, any text backend) #111

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

Draft
Birfy wants to merge 1 commit into
base: main
Choose a base branch
from
Draft

[Feat] Score browser decision backends step by step on myjev-browser-decisions (offline, any text backend) #111

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
11 changes: 11 additions & 0 deletions CHANGELOG.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4,6 +4,17 @@ The format follows [Keep a Changelog](https://keepachangelog.com/en/1.1.0/); ver

## Unreleased

### Added

- `python -m benchmarks.legs.browser_steps`: offline step-level scoring of browser decision backends on
[myjev-browser-decisions](https://huggingface.co/datasets/Mlchemist/myjev-browser-decisions) (pinned revision,
hash-checked at download, never vendored; CC BY-NC-SA 4.0). `predict` sends each step through `DecisionModel`
(any text backend), one request at a time, no retries, failures scored wrong; `score` reports operation and
element accuracy, step success, Brier/ECE and latency with denominators, on the full split and the Jev-answered
subset; `thresholds` turns a calibration run into a deferral threshold table. `--shape dataset` re-sends
jev-ultrafast's body unchanged, `--shape ours` swaps in our operation text and rules. Only a random floor is
reported so far. See [docs/results/browser-steps](docs/results/browser-steps/README.md) (#109).

## 0.3.0 - 2026-10-30

### Added
Expand Down
4 changes: 4 additions & 0 deletions benchmarks/README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -11,6 +11,10 @@ The [accuracy leaderboard](accuracy-leaderboard.md) ranks that frozen run and
defines how to add comparable results. Request latency has a separate
[System1-Omni leaderboard](https://github.com/ThinkFlowLab/system1-omni/pull/131).

Browser decision backends are scored step by step, offline, on myjev-browser-decisions
with `python -m benchmarks.legs.browser_steps`; see
[docs/results/browser-steps](../docs/results/browser-steps/README.md).

## Run

Use the repository's development environment and a running service. Clone
Expand Down
Loading
Loading