Cua-S1 4B 0.2 text worker

This recipe runs the Cua-S1 4B 0.2 text adapter through Transformers and PEFT behind the Rust frontend. It is the correctness reference for native execution. The model is in src/models/cua_s1/text/, the HTTP worker in src/frontend/cua_s1_text.py, and src/models/cua_s1/README.md documents the contract. Only choice questions are supported.

Run the commands from the repository root, on Linux with an NVIDIA GPU and Python 3.12. The pinned versions match the upstream reference environment:

python3.12 -m venv .venv
.venv/bin/python -m pip install -r recipe/cua_s1/requirements-text.txt
.venv/bin/hf download Qwen/Qwen3.5-4B \
  --revision 851bf6e806efd8d0a36b00ddf55e13ccb7b8cd0a --local-dir weights/Qwen3.5-4B
.venv/bin/hf download cua-ai/cua-s1-4b-0.2 \
  --revision 16818868b0cc7813808aae4e87b417657046ab79 --local-dir weights/cua-s1-4b-0.2

Upstream's libs/cua-s1/ci/fetch_pinned_weights.py --dest weights --verify-only (in trycua/cua at 0e75660ce4c2edda519e0c795fa3ad98abf4e76f) checks every downloaded file against upstream's lock.

Start the worker, which runs one warmup decision before it listens, then the frontend:

PYTHONPATH=src .venv/bin/python -m frontend.cua_s1_text \
  --base weights/Qwen3.5-4B --adapter weights/cua-s1-4b-0.2/text --port 8000
cargo build --release --locked
OMNI_JEV_BIND=127.0.0.1:8080 OMNI_JEV_BACKEND_URL=http://127.0.0.1:8000 ./target/release/omni-jev

Requests run one at a time. Bodies over 4 MiB, more than 64 questions, or a prompt over 16,384 tokens get 413. The worker computes logits for every position, as upstream does, so memory grows with prompt length: the 15,446-token test input peaked at about 21.3 GiB in bfloat16.

curl http://127.0.0.1:8080/v1/systemone \
  -H 'Content-Type: application/json' \
  -d '{"model":"cua-s1-4b-0.2","state":"Dialog: Delete 3 files permanently? Buttons: Delete, Cancel","questions":{"pick":{"type":"choice","instructions":"Keep the files.","criteria":{"delete":"Click Delete","cancel":"Click Cancel"}}}}'

On an RTX 6000 Ada in bfloat16, the response is:

{"model":"cua-ai/cua-s1-4b-0.2@16818868b0cc7813808aae4e87b417657046ab79:text","answers":{"pick":{"type":"choice","choice":"cancel","probabilities":{"delete":0.0024726232513785362,"cancel":0.9975274205207825},"confidence":0.9750249565060322}},"usage":{"input_tokens":153,"output_tokens":0}}

The tests need neither weights nor a GPU; with CUA_S1_BASE=weights/Qwen3.5-4B they also check the tokenizer:

.venv/bin/python -m pip install pytest httpx
PYTHONPATH=src .venv/bin/python -m pytest tests/cua_s1