sense-d

Fab7's decision model, which Weft runs on your machine. Nothing leaves the machine.

One microsoft/deberta-v3-xsmall answers every judgment Weft needs as a question with a probability. Each row is [CLS] text [SEP] question [SEP], read as one logit (yes minus no).

• Screen: the agent's terminal state, in 8 states (idle, generating, running a tool, after its Stop hook, waiting on a permission, a question, trust or sign-in, or stopped on an error).
• Transcript entry: the person's message, the agent's reply, a tool step that worked or failed, or internal text.
• Ask: its task, its effects (read, write, execute, beyond the machine) and its route (direct, plan, goal, review). The result follows from these.
• Practice rules: a convention decides seven from the task and effects, and each other rule is a question.
• Recall: whether a changed code window does a requirement.

The questions, calibration and text normalizing are data in manifest.json (schema fab7.sense.decision/1).

Files

File	What
sense.int8.onnx	The model, with int8 weights and fixed scales. Input input_ids and attention_mask (int64[B, L], L ≤ 256); output logit (float32[B]). Run it with ONNX Runtime and session.disable_quant_qdq = 1, so each row's answer is its own.
tokenizer.json	DeBERTa-v3's tokenizer
manifest.json	Every head: its questions, classes and calibration; the normalizer; file digests
golden_vectors.jsonl	Normalized texts, token ids, logits and answers to check a runtime against

Training

• The base: microsoft/deberta-v3-xsmall, first taught entailment on the human-labelled, commercially usable pairs of MoritzLaurer/dataset_train_nli.
• The fine-tuning: Fab7's own labels:
  • screens and transcripts from sandboxed harness sessions;
  • Asks, rules and windows labelled by a person or from templates;
  • published per-hunk code labels.
• Nothing was distilled from another model.

Performance (0.1.0, test split, texts training never saw)

Head	Score
Screen	accuracy 0.981 (820 screens)
Transcript entry	accuracy 0.991 (342)
Ask task	macro F1 0.791 (308 Asks)
Ask effects	macro F1 0.886; risky-effect recall 0.955
Ask route	accuracy 0.886
Practice rules	Jaccard 0.690
Recall	flags 98 % of windows that miss their requirement, at 89 % precision

On a CPU, a screen read (8 rows) takes about 190 ms on an Apple Silicon Mac; on a GPU, a few milliseconds.