Add model card and Apache-2.0 license

Adds a model card with YAML metadata (license: apache-2.0, library_name: cua-s1). This checkpoint was trained from scratch, so the card declares no base_model. The license matches cua-ai/cua-s1-4b-0.2.

The card documents intended and out-of-scope use, the text and multimodal (frozen SigLIP) layout, a training summary, measured cua-bench-s1 results, the pinned revision (1f93fd0f), and safetensors loading with cua_s1.nano. It also notes that the jev-use chooser does not support this checkpoint. Sources: trycua/cua libs/cua-s1/MODEL_CARD.md and README.md (trycua/cua#4186).

Only README.md changes; the weights do not.

README.md

@@ -0,0 +1,174 @@
+---
+license: apache-2.0
+library_name: cua-s1
+tags:
+  - computer-use
+  - gui-agent
+  - option-attention
+  - cua-s1
+  - safetensors
+language:
+  - en
+---
+
+# cua-s1-nano-0.1
+
+`cua-s1-nano-0.1` is a small research classifier for closed-option
+computer-use decisions. It has about 855,000 trainable parameters and was
+trained from scratch. Given a screen state and a closed set of candidate
+(element, action) options, it scores every option in one parallel forward
+pass and selects the highest-scoring option for each element.
+
+This checkpoint belongs to the [Cua-S1](https://github.com/trycua/cua/tree/main/libs/cua-s1)
+research family. It is not a general-purpose assistant. Do not treat a result
+on one task family as evidence of reliability outside it.
+
+## Files
+
+| Path | Contents |
+| --- | --- |
+| `text/model.safetensors`, `text/config.json` | Text checkpoint: a byte-level context encoder over an accessibility-tree excerpt for each element |
+| `multimodal/model.safetensors`, `multimodal/config.json` | Multimodal checkpoint: a trainable projection over frozen `google/siglip-base-patch16-224` features from a screenshot crop for each element |
+
+Each checkpoint has 855,296 parameters. Tensors are stored only in
+safetensors, and the JSON sidecar holds the architecture config and a SHA-256
+signature over the tensors. This repository contains no pickle files and does
+not redistribute the SigLIP backbone.
+
+## Intended use
+
+- Research on closed-option computer-use GUI decisions (element and action
+  selection) with a very small, fast architecture.
+- Comparing a from-scratch specialist against larger checkpoints, such as
+  [`cua-s1-4b-0.2`](https://huggingface.co/cua-ai/cua-s1-4b-0.2), on
+  [`cua-bench-s1`](https://github.com/trycua/cua/tree/main/libs/cua-bench-s1).
+- Studying task-family generalization, including out-of-domain checks.
+
+### Out of scope
+
+- General-purpose or open-ended computer operation.
+- Unsupervised operation on production accounts or sensitive data.
+- Actions with financial, legal, medical, employment, safety, or other
+  high-impact consequences.
+- Bypassing access controls, consent, rate limits, or service policies.
+- Treating a selected option as proof that the action is correct, safe, or
+  succeeded.
+
+## How it works
+
+An option-attention head uses each candidate option as a query against the
+element's context tokens and produces one logit per option. Options are
+always encoded as text by a shared byte-level option encoder. The context
+depends on the modality:
+
+- **Text**: a small trainable byte-level transformer over a rendered
+  accessibility-tree excerpt for the element. It needs no extra dependencies.
+- **Multimodal**: a frozen vision backbone over a screenshot crop of the
+  element, followed by a small trainable projection. This checkpoint's config
+  selects `siglip` (`google/siglip-base-patch16-224`).
+
+Scoring is one parallel forward pass, not autoregressive generation. Latency
+is in the low milliseconds per task on a GPU and under 100 ms on a CPU.
+
+## Training summary
+
+The model was trained from scratch; it has no base model. It was trained on
+the `data/v1` split of the `cua-bench-s1` core GUI task families:
+`form_filling`, `login_auth`, `consent_checkbox`, `multi_step_submit`,
+`pagination`, and `search_filter`. `cua-bench-s1` builds these families with
+its synthetic generator and its converters for
+[AndroidControl](https://github.com/google-research/google-research/tree/master/android_control)
+(Apache-2.0) and GUI-360 (MIT). For exact provenance, see
+[Data sources and provenance](https://github.com/trycua/cua/tree/main/libs/cua-bench-s1#data-sources-and-provenance).
+The two-stage training recipe (base training, then an optional fine-tune) is
+[`libs/cua-s1/training/train_nano.py`](https://github.com/trycua/cua/blob/main/libs/cua-s1/training/train_nano.py).
+No dataset is distributed with this checkpoint.
+
+## Evaluation
+
+These are measured results from the `cua-bench-s1`
+[Results section](https://github.com/trycua/cua/tree/main/libs/cua-bench-s1#results):
+
+- Held-out cross-dataset text split: task-level accuracy of 0.000 to 0.286,
+  depending on family. The per-element text context carries no goal string.
+- Same-distribution multimodal split (the split it was trained on): 1.000.
+- `chess`, on the same 15-position subset used for every model: 0.000 task
+  accuracy and 0.227 element accuracy in both modalities.
+- `game_control`: 0.000 task accuracy and 0.333 element accuracy. Neither
+  result survives chance correction.
+- `safety_gate`: 0.000 zero-shot, and 1.000 after fine-tuning on that
+  family's own split.
+- External `general_decision` benchmark: no result, because the model cannot
+  produce that benchmark's text-only decision shape.
+
+The strong same-distribution result and weak cross-dataset result mean that
+this checkpoint mostly reflects its training distribution. It is not a
+general GUI decision model.
+
+## How to run
+
+Pin the download to the revision the Cua-S1 documentation was verified
+against. The weights are unchanged at later revisions of this repository
+that change only documentation.
+
+| Artifact | Revision |
+| --- | --- |
+| `cua-ai/cua-s1-nano-0.1` (weights) | `1f93fd0fdcbe33740334948f967dff9f6c8e9f34` |
+
+From a checkout of [`trycua/cua`](https://github.com/trycua/cua):
+
+```bash
+uv sync --project libs/cua-s1/python                          # text modality
+uv sync --project libs/cua-s1/python --extra nano-vision     # adds the multimodal backbone
+libs/cua-s1/python/.venv/bin/hf download cua-ai/cua-s1-nano-0.1 \
+  --revision 1f93fd0fdcbe33740334948f967dff9f6c8e9f34 \
+  --local-dir "$HOME/cua-s1-models/cua-s1-nano-0.1"
+```
+
+```python
+from pathlib import Path
+from cua_s1.nano import load_nano_checkpoint, select_device
+
+root = Path.home() / "cua-s1-models" / "cua-s1-nano-0.1"
+# Validates the format, version, and SHA-256 tensor signature before returning.
+model, collator, config = load_nano_checkpoint(root / "text", select_device("auto"))
+```
+
+The multimodal checkpoint loads the same way from `root / "multimodal"`. On
+first use, it downloads the frozen `google/siglip-base-patch16-224` backbone
+through Transformers.
+
+### jev-use chooser
+
+The [jev-use closed-candidate chooser](https://github.com/trycua/cua/blob/main/libs/cua-driver/examples/jev-use/decision-models.md)
+supports only the Cua-S1-4B adapters. It fails setup for `cua-s1-nano-0.1`,
+and no Cua Driver integration runs this checkpoint yet. To drive the jev-use
+flow with a Cua-S1 model, use
+[`cua-s1-4b-0.2`](https://huggingface.co/cua-ai/cua-s1-4b-0.2) and follow
+[Get the weights and run inference](https://github.com/trycua/cua/tree/main/libs/cua-s1#get-the-weights-and-run-inference).
+Use this checkpoint through `cua_s1.nano` directly, or through
+`cua-bench-s1`'s model adapters.
+
+## Limitations
+
+- It needs a closed, pre-enumerated option set for each element and cannot
+  propose an action outside that set.
+- It scores each element independently, so it cannot compare which element
+  to act on next across a screen.
+- Its small size trades capacity for speed.
+- Multimodal behavior inherits the frozen backbone's limitations and depends
+  on screenshot-crop quality.
+- `game_control` and `chess` are out-of-domain checks only.
+- No weights-backed CI or canonical Cua Driver desktop E2E covers this
+  checkpoint.
+
+Run it in an isolated environment with least-privilege credentials and
+bounded actions. Verify outcomes independently, and require human
+confirmation before consequential or irreversible actions.
+
+## License
+
+The weights in this repository are licensed under Apache-2.0. The frozen
+`google/siglip-base-patch16-224` backbone used by the multimodal checkpoint
+is not redistributed here: its own license governs it. The Cua-S1 source code
+is MIT-licensed in [`trycua/cua`](https://github.com/trycua/cua).