dogfood nld_classifier_v3

[evelyn-with-warp]

Description

• add nld_classifier_v3, bonded with bert_tiny_v3.onnx
• let local/dev build subscribe to this feature

for metrics comparison with V2, please checkout the PR https://github.com/warpdotdev/ml_models/pull/10

Linked Issue

• The linked issue is labeled ready-to-spec or ready-to-implement.
• Where appropriate, screenshots or a short video of the implementation are included below (especially for user-visible or UI changes).

Testing

• I have manually tested my changes locally with ./script/run

Screenshots / Videos

https://www.loom.com/share/c03060910bf54dc3bfc825948222fe32
improved user experience on this report https://warpdev.slack.com/archives/C08QY6ZBK2A/p1779810034759499;
however, for this one, since whoops is a valid cli command, the classifier will still flicker as user typing going on

Agent Mode

• Warp Agent Mode - This PR was created via Warp's AI Agent Mode

[oz-for-oss]

@evelyn-with-warp

I'm starting a first review of this pull request.

You can view the conversation on Warp.

I completed the review and no human review was requested for this pull request.

Comment /oz-review on this pull request to retrigger a review (up to 3 times on the same pull request).

Powered by Oz

[oz-for-oss]

Overview

This PR adds the nld_classifier_v3 Cargo feature, wires the new BERT Tiny v3 ONNX model through the input classifier crate and evaluation binary, and updates local/dev bundle scripts to ship v3 while keeping preview on v2 and stable/other channels on v1.

Concerns

• No blocking correctness, security, or spec-alignment issues found in the annotated diff.

Verdict

Found: 0 critical, 0 important, 0 suggestions

Approve

Comment /oz-review on this pull request to retrigger a review (up to 3 times on the same pull request).

Powered by Oz