decisio 0.1.1__tar.gz

decisio-0.1.1/LICENSE

@@ -0,0 +1,202 @@
+
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright [yyyy] [name of copyright owner]
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.

decisio-0.1.1/NOTICE

@@ -0,0 +1,2 @@
+decisio
+Copyright 2026 Amin Roudaki and the Decisio contributors

decisio-0.1.1/PKG-INFO

@@ -0,0 +1,190 @@
+Metadata-Version: 2.4
+Name: decisio
+Version: 0.1.1
+Summary: A prefill-only decision server on vLLM: typed questions about a state, a probability distribution per question
+Author: Amin Roudaki
+License-Expression: Apache-2.0
+Project-URL: Repository, https://github.com/aminry/decisio
+Requires-Python: <3.13,>=3.12
+Description-Content-Type: text/markdown
+License-File: LICENSE
+License-File: NOTICE
+Requires-Dist: fastapi>=0.133
+Requires-Dist: numpy>=2.2
+Requires-Dist: pillow>=11
+Requires-Dist: psutil>=7
+Requires-Dist: pydantic>=2.12
+Requires-Dist: python-multipart>=0.0.20
+Requires-Dist: safetensors>=0.6.2
+Requires-Dist: scipy>=1.15
+Requires-Dist: torch>=2.13
+Requires-Dist: torchvision>=0.28
+Requires-Dist: transformers>=5.10.4
+Requires-Dist: uvicorn>=0.34
+Provides-Extra: serve
+Requires-Dist: vllm==0.30.0; sys_platform == "linux" and extra == "serve"
+Provides-Extra: bench
+Requires-Dist: httpx>=0.28; extra == "bench"
+Requires-Dist: typesafe-sdk==0.7.0; extra == "bench"
+Provides-Extra: dev
+Requires-Dist: httpx>=0.28; extra == "dev"
+Requires-Dist: pytest==9.1.1; extra == "dev"
+Requires-Dist: ruff==0.14.0; extra == "dev"
+Dynamic: license-file
+
+<!-- SPDX-License-Identifier: Apache-2.0 -->
+<!-- SPDX-FileCopyrightText: Copyright contributors to the decisio project -->
+
+# Decisio
+
+**A decision server: typed questions about a piece of text, a probability for every option, one forward pass per question.**
+
+Decisio serves a frozen open model, Qwen3.6-35B-A3B, on vLLM and reads its answer to a closed question from the option letters' logits.
+No text is generated.
+A request carries a state (a ticket, a document, a record) and one or more questions of three types: yes/no, a choice among named options, or a score on an ordered scale.
+The answer to each question is a distribution over its options, and the request format is TypeSafe's System One wire format, so existing clients work unchanged.
+
+What ships here is code, recipes and evaluation records.
+No trained weights: every measured answer comes from the official checkpoint.
+`EVAL_CARD.md` states every number, what it was measured on, and what was fitted on what.
+
+## What it does
+
+| | |
+| --- | --- |
+| Question types | yes/no (`noul`), choice among up to 255 options with descriptions, score on an ordered scale |
+| Many questions per request | the state is prefilled once and shared through vLLM's prefix cache |
+| Probabilities | a distribution per question, with one fitted global temperature; per-task calibration from labelled examples |
+| Task registration | `POST /v1/tasks`: from labelled examples of one recurring question the server fits per-task calibration and, for option lists of 10 or more, a linear head on the model's hidden state; guide in `docs/tasks.md` |
+| Abstention | an opt-in per-task threshold on a declared "can't tell" option (`POST /v1/abstention/tasks`) |
+| Image input | photos in the state, served by a second engine on the same card (`--image-model`) |
+| Rendering rules | enumerated option keys are hidden, snake_case labels are shown as words, ties resolve by key so the answer never depends on the order keys arrive in |
+| Determinism | one request at a time, on the same card: bit-identical answers when the server's request history is the same, and repeats agreed to 1.2e-7 in probability; identical requests sent inside one batch are not the same forward pass and differed by up to 0.59 in a label probability (`EVAL_CARD.md` section 4); the serving gates in `tests/gpu` check the one-at-a-time case |
+| Context | up to 32,768 tokens of state |
+
+Latency and cost on one RTX PRO 6000, from `EVAL_CARD.md`: about 28 ms server time for one question, 2.7 to 5.4 ms per question when many questions share a state, and about $0.001 to $0.012 per 1,000 decisions at $1.50 per card-hour.
+
+## Quickstart
+
+Requirements: Linux, one NVIDIA card with 96 GB (measured on an RTX PRO 6000 Blackwell) and a driver that supports CUDA 13.0 (the runtime `uv.lock` pins), Python 3.12, [uv](https://docs.astral.sh/uv/).
+
+```
+git clone https://github.com/aminry/decisio
+cd decisio
+uv sync --extra serve --frozen
+uv run python -m decisio.serve.vllm_engine --model Qwen/Qwen3.6-35B-A3B-FP8 --model-class hidden-readout
+```
+
+The first start downloads the checkpoint (about 36 GB) and warms the engine.
+The decisio plugin registers its model classes with vLLM through an entry point, so vLLM 0.30.0 loads them without any patch.
+`patches/` holds an optional latency patch series, off by default; see `patches/README.md`.
+
+Ask a question:
+
+```
+curl http://127.0.0.1:8000/v1/systemone -H 'Content-Type: application/json' -d '{
+  "state": "Hi, since this morning none of our 40 staff can log in to the dashboard. We get \"session expired\" right after entering the password. Payroll is due today.",
+  "questions": {
+    "urgent":   {"type": "noul",   "instructions": "Does this ticket need a response within the hour?"},
+    "category": {"type": "choice", "instructions": "Which team should handle this ticket?",
+                 "criteria": {"billing": "Invoices, payments, refunds", "access": "Login, passwords, permissions", "bug": "The product behaves wrongly", "other": null}},
+    "impact":   {"type": "score",  "instructions": "Rate the business impact using only the reported facts.",
+                 "criteria": ["No function impaired", "One user impaired, with a workaround", "Many users blocked from a core function", "Data loss or legal exposure"]}
+  }
+}'
+```
+
+Response, abridged (the values the first GPU start returned, rounded):
+
+```
+{"answers": {
+  "urgent":   {"type": "noul",   "noul": 0.90},
+  "category": {"type": "choice", "choice": "access", "probabilities": {"billing": 0.03, "access": 0.79, "bug": 0.16, "other": 0.02}},
+  "impact":   {"type": "score",  "score": 2.0, "probabilities": {"0": 0.00, "1": 0.01, "2": 0.96, "3": 0.02}}}}
+```
+
+Your probabilities will differ, here by up to 0.08: the three questions of one request are scored in one batch, and on this stack a question's probabilities move with what else is in its batch and with what the server answered before (a fresh clone's first start gave `access` 0.87, `runs/2026-10-01_quickstart-public`; `EVAL_CARD.md` section 4). The choices do not change.
+
+## Teach it your question in ten examples
+
+The model is frozen, but the server can learn one recurring question from your own labelled examples: it fits a per-task calibration and, for 10 or more options, a small head on the model's hidden state, each kept only if cross-validation on your examples shows a gain.
+With 10 labelled examples per intent, accuracy on held-out test items rose from 0.740 to 0.847 on BANKING77 (77 intents) and from 0.820 to 0.893 on CLINC150 (150 intents), against the same model unregistered.
+
+```
+curl http://127.0.0.1:8000/v1/tasks -H 'Content-Type: application/json' -d @examples/tasks/examples.json
+```
+
+`docs/tasks.md` walks through it and states what each number was measured on; `examples/tasks/` runs it on a laptop against the CPU stand-in.
+
+## Run with Docker
+
+```
+uv build --wheel                   # the image installs this wheel
+docker compose up --build          # needs the NVIDIA Container Toolkit and a 96 GB card
+curl http://127.0.0.1:8000/health  # answers once the first start has fetched the checkpoint
+```
+The first start downloads the checkpoint (about 36 GB) into the `decisio-data` volume; it is never part of the image.
+The container runs as a non-root user, compose publishes the port on 127.0.0.1 only, and `Dockerfile` and `compose.yaml` explain the rest.
+Release images go to `ghcr.io/aminry/decisio`, with their digest in the release notes.
+First GPU start (2026-10-01, one RTX PRO 6000 Blackwell, image built on the machine from the `Dockerfile`): healthy in 651 s including the checkpoint download and in 206 s from the cached volume; the example request below and the conformance gates C2 to C4 pass against the container (`runs/2026-10-01_docker-first-gpu-start`); the image published with v0.1.0 repeated the start from its digest (healthy in 223 s with the checkpoint cached, example and conformance pass, `pushed_image_0.1.0` in that run).
+
+## Without a GPU
+
+The unit tests and the CPU stand-in engine run on any machine:
+
+```
+uv sync --extra dev --frozen
+uv run pytest
+```
+
+`--backend hf` serves a small Hugging Face model on the CPU for development of the routes; it is not the measured system.
+
+A laptop version of the same model, packaged for [Ollama](https://ollama.com/aminroudaki/decisio)'s decision route, exists as `ollama pull aminroudaki/decisio`.
+Ollama builds its own prompt and applies no calibration, so its numbers are the ones on that page, not the ones here.
+
+## How it performs
+
+All numbers are on the served default described in `EVAL_CARD.md`, measured privately with the public harnesses; nothing here is a leaderboard score.
+
+| Measure | Result |
+| --- | --- |
+| JevBench, 231 published items, accuracy by tier | easy 1.000, standard 0.972, hard 0.694 |
+| Decision Index 0.2.1, four benchmarks | BANKING77 macro-F1 0.731, CLINC150+OOS 0.814, GPQA Diamond 0.454, MMLU-Pro 0.609 |
+| Intent heads from 10 labelled examples per intent | BANKING77 0.847, CLINC150 0.893 |
+| Image input, ImajevBench v2.0-lite | 0.791 on the 230 answerable items |
+| Latency, one question | about 28 ms server time |
+
+Where it stands: on the public harnesses this frozen model is the most accurate open one-pass system we know of on the hard tier, and it is behind TypeSafe's Jev on hard knowledge questions by several points and on intent taxonomies without labelled examples.
+With 10 labelled examples per intent, registered heads reach accuracy 0.847 on BANKING77 (150 held-out items) and 0.893 on CLINC150 (100 held-out items), means of three draws.
+They use labelled examples, so those figures are not comparable with zero-shot systems.
+Calibration trails too: hard-tier ECE on JevBench is 0.059, above the 0.05 we aimed for.
+`EVAL_CARD.md` has the full tables, the calibration figures, and the three disclosures about what was fitted on what.
+
+## Repository layout
+
+```
+src/decisio/serve/      the server: /v1/systemone, tasks, abstention, temperature, image route, conformance
+src/decisio/readout/    the letters readout, calibration and the intent head (reference implementations)
+src/decisio/vllm_plugin/  the vLLM entry point and the two model classes
+src/decisio/bench/      benchmark scoring (JevBench v1.5 open-set reading, Decision Index reports)
+benchmarks/             scripts that run the public harnesses against a server
+patches/                optional vLLM patch series, off by default
+Dockerfile  compose.yaml  docker/   the server image (vLLM 0.30.0 release image plus the wheel)
+tests/unit/  tests/gpu/ CPU tests run on every pull request; GPU tests run nightly (pytest -m gpu)
+runs/                   evaluation records: a manifest, per-item results and hashes per run
+docs/                   the task-registration guide and specification, design notes
+examples/tasks/         a runnable task-registration walk-through (CPU stand-in)
+EVAL_CARD.md            what was measured, on what, and what was fitted on what
+```
+
+## Contributing
+
+See `CONTRIBUTING.md`.
+Pull requests need a DCO sign-off (`git commit -s`), pass the CPU tests and lint, and are reviewed by a maintainer before merge.
+Security issues go through GitHub's private vulnerability reporting, as described in `SECURITY.md`.
+
+## Licence
+
+Apache-2.0 (`LICENSE`, `NOTICE`).
+The model weights are Alibaba's Qwen3.6-35B-A3B under Apache-2.0 and are downloaded, not redistributed.
+`THIRD-PARTY.md` lists everything else this project builds on.

decisio-0.1.1/README.md

@@ -0,0 +1,156 @@
+<!-- SPDX-License-Identifier: Apache-2.0 -->
+<!-- SPDX-FileCopyrightText: Copyright contributors to the decisio project -->
+
+# Decisio
+
+**A decision server: typed questions about a piece of text, a probability for every option, one forward pass per question.**
+
+Decisio serves a frozen open model, Qwen3.6-35B-A3B, on vLLM and reads its answer to a closed question from the option letters' logits.
+No text is generated.
+A request carries a state (a ticket, a document, a record) and one or more questions of three types: yes/no, a choice among named options, or a score on an ordered scale.
+The answer to each question is a distribution over its options, and the request format is TypeSafe's System One wire format, so existing clients work unchanged.
+
+What ships here is code, recipes and evaluation records.
+No trained weights: every measured answer comes from the official checkpoint.
+`EVAL_CARD.md` states every number, what it was measured on, and what was fitted on what.
+
+## What it does
+
+| | |
+| --- | --- |
+| Question types | yes/no (`noul`), choice among up to 255 options with descriptions, score on an ordered scale |
+| Many questions per request | the state is prefilled once and shared through vLLM's prefix cache |
+| Probabilities | a distribution per question, with one fitted global temperature; per-task calibration from labelled examples |
+| Task registration | `POST /v1/tasks`: from labelled examples of one recurring question the server fits per-task calibration and, for option lists of 10 or more, a linear head on the model's hidden state; guide in `docs/tasks.md` |
+| Abstention | an opt-in per-task threshold on a declared "can't tell" option (`POST /v1/abstention/tasks`) |
+| Image input | photos in the state, served by a second engine on the same card (`--image-model`) |
+| Rendering rules | enumerated option keys are hidden, snake_case labels are shown as words, ties resolve by key so the answer never depends on the order keys arrive in |
+| Determinism | one request at a time, on the same card: bit-identical answers when the server's request history is the same, and repeats agreed to 1.2e-7 in probability; identical requests sent inside one batch are not the same forward pass and differed by up to 0.59 in a label probability (`EVAL_CARD.md` section 4); the serving gates in `tests/gpu` check the one-at-a-time case |
+| Context | up to 32,768 tokens of state |
+
+Latency and cost on one RTX PRO 6000, from `EVAL_CARD.md`: about 28 ms server time for one question, 2.7 to 5.4 ms per question when many questions share a state, and about $0.001 to $0.012 per 1,000 decisions at $1.50 per card-hour.
+
+## Quickstart
+
+Requirements: Linux, one NVIDIA card with 96 GB (measured on an RTX PRO 6000 Blackwell) and a driver that supports CUDA 13.0 (the runtime `uv.lock` pins), Python 3.12, [uv](https://docs.astral.sh/uv/).
+
+```
+git clone https://github.com/aminry/decisio
+cd decisio
+uv sync --extra serve --frozen
+uv run python -m decisio.serve.vllm_engine --model Qwen/Qwen3.6-35B-A3B-FP8 --model-class hidden-readout
+```
+
+The first start downloads the checkpoint (about 36 GB) and warms the engine.
+The decisio plugin registers its model classes with vLLM through an entry point, so vLLM 0.30.0 loads them without any patch.
+`patches/` holds an optional latency patch series, off by default; see `patches/README.md`.
+
+Ask a question:
+
+```
+curl http://127.0.0.1:8000/v1/systemone -H 'Content-Type: application/json' -d '{
+  "state": "Hi, since this morning none of our 40 staff can log in to the dashboard. We get \"session expired\" right after entering the password. Payroll is due today.",
+  "questions": {
+    "urgent":   {"type": "noul",   "instructions": "Does this ticket need a response within the hour?"},
+    "category": {"type": "choice", "instructions": "Which team should handle this ticket?",
+                 "criteria": {"billing": "Invoices, payments, refunds", "access": "Login, passwords, permissions", "bug": "The product behaves wrongly", "other": null}},
+    "impact":   {"type": "score",  "instructions": "Rate the business impact using only the reported facts.",
+                 "criteria": ["No function impaired", "One user impaired, with a workaround", "Many users blocked from a core function", "Data loss or legal exposure"]}
+  }
+}'
+```
+
+Response, abridged (the values the first GPU start returned, rounded):
+
+```
+{"answers": {
+  "urgent":   {"type": "noul",   "noul": 0.90},
+  "category": {"type": "choice", "choice": "access", "probabilities": {"billing": 0.03, "access": 0.79, "bug": 0.16, "other": 0.02}},
+  "impact":   {"type": "score",  "score": 2.0, "probabilities": {"0": 0.00, "1": 0.01, "2": 0.96, "3": 0.02}}}}
+```
+
+Your probabilities will differ, here by up to 0.08: the three questions of one request are scored in one batch, and on this stack a question's probabilities move with what else is in its batch and with what the server answered before (a fresh clone's first start gave `access` 0.87, `runs/2026-10-01_quickstart-public`; `EVAL_CARD.md` section 4). The choices do not change.
+
+## Teach it your question in ten examples
+
+The model is frozen, but the server can learn one recurring question from your own labelled examples: it fits a per-task calibration and, for 10 or more options, a small head on the model's hidden state, each kept only if cross-validation on your examples shows a gain.
+With 10 labelled examples per intent, accuracy on held-out test items rose from 0.740 to 0.847 on BANKING77 (77 intents) and from 0.820 to 0.893 on CLINC150 (150 intents), against the same model unregistered.
+
+```
+curl http://127.0.0.1:8000/v1/tasks -H 'Content-Type: application/json' -d @examples/tasks/examples.json
+```
+
+`docs/tasks.md` walks through it and states what each number was measured on; `examples/tasks/` runs it on a laptop against the CPU stand-in.
+
+## Run with Docker
+
+```
+uv build --wheel                   # the image installs this wheel
+docker compose up --build          # needs the NVIDIA Container Toolkit and a 96 GB card
+curl http://127.0.0.1:8000/health  # answers once the first start has fetched the checkpoint
+```
+The first start downloads the checkpoint (about 36 GB) into the `decisio-data` volume; it is never part of the image.
+The container runs as a non-root user, compose publishes the port on 127.0.0.1 only, and `Dockerfile` and `compose.yaml` explain the rest.
+Release images go to `ghcr.io/aminry/decisio`, with their digest in the release notes.
+First GPU start (2026-10-01, one RTX PRO 6000 Blackwell, image built on the machine from the `Dockerfile`): healthy in 651 s including the checkpoint download and in 206 s from the cached volume; the example request below and the conformance gates C2 to C4 pass against the container (`runs/2026-10-01_docker-first-gpu-start`); the image published with v0.1.0 repeated the start from its digest (healthy in 223 s with the checkpoint cached, example and conformance pass, `pushed_image_0.1.0` in that run).
+
+## Without a GPU
+
+The unit tests and the CPU stand-in engine run on any machine:
+
+```
+uv sync --extra dev --frozen
+uv run pytest
+```
+
+`--backend hf` serves a small Hugging Face model on the CPU for development of the routes; it is not the measured system.
+
+A laptop version of the same model, packaged for [Ollama](https://ollama.com/aminroudaki/decisio)'s decision route, exists as `ollama pull aminroudaki/decisio`.
+Ollama builds its own prompt and applies no calibration, so its numbers are the ones on that page, not the ones here.
+
+## How it performs
+
+All numbers are on the served default described in `EVAL_CARD.md`, measured privately with the public harnesses; nothing here is a leaderboard score.
+
+| Measure | Result |
+| --- | --- |
+| JevBench, 231 published items, accuracy by tier | easy 1.000, standard 0.972, hard 0.694 |
+| Decision Index 0.2.1, four benchmarks | BANKING77 macro-F1 0.731, CLINC150+OOS 0.814, GPQA Diamond 0.454, MMLU-Pro 0.609 |
+| Intent heads from 10 labelled examples per intent | BANKING77 0.847, CLINC150 0.893 |
+| Image input, ImajevBench v2.0-lite | 0.791 on the 230 answerable items |
+| Latency, one question | about 28 ms server time |
+
+Where it stands: on the public harnesses this frozen model is the most accurate open one-pass system we know of on the hard tier, and it is behind TypeSafe's Jev on hard knowledge questions by several points and on intent taxonomies without labelled examples.
+With 10 labelled examples per intent, registered heads reach accuracy 0.847 on BANKING77 (150 held-out items) and 0.893 on CLINC150 (100 held-out items), means of three draws.
+They use labelled examples, so those figures are not comparable with zero-shot systems.
+Calibration trails too: hard-tier ECE on JevBench is 0.059, above the 0.05 we aimed for.
+`EVAL_CARD.md` has the full tables, the calibration figures, and the three disclosures about what was fitted on what.
+
+## Repository layout
+
+```
+src/decisio/serve/      the server: /v1/systemone, tasks, abstention, temperature, image route, conformance
+src/decisio/readout/    the letters readout, calibration and the intent head (reference implementations)
+src/decisio/vllm_plugin/  the vLLM entry point and the two model classes
+src/decisio/bench/      benchmark scoring (JevBench v1.5 open-set reading, Decision Index reports)
+benchmarks/             scripts that run the public harnesses against a server
+patches/                optional vLLM patch series, off by default
+Dockerfile  compose.yaml  docker/   the server image (vLLM 0.30.0 release image plus the wheel)
+tests/unit/  tests/gpu/ CPU tests run on every pull request; GPU tests run nightly (pytest -m gpu)
+runs/                   evaluation records: a manifest, per-item results and hashes per run
+docs/                   the task-registration guide and specification, design notes
+examples/tasks/         a runnable task-registration walk-through (CPU stand-in)
+EVAL_CARD.md            what was measured, on what, and what was fitted on what
+```
+
+## Contributing
+
+See `CONTRIBUTING.md`.
+Pull requests need a DCO sign-off (`git commit -s`), pass the CPU tests and lint, and are reviewed by a maintainer before merge.
+Security issues go through GitHub's private vulnerability reporting, as described in `SECURITY.md`.
+
+## Licence
+
+Apache-2.0 (`LICENSE`, `NOTICE`).
+The model weights are Alibaba's Qwen3.6-35B-A3B under Apache-2.0 and are downloaded, not redistributed.
+`THIRD-PARTY.md` lists everything else this project builds on.

decisio-0.1.1/pyproject.toml

@@ -0,0 +1,73 @@
+# SPDX-License-Identifier: Apache-2.0
+# SPDX-FileCopyrightText: Copyright contributors to the decisio project
+
+[build-system]
+requires = ["setuptools>=77"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "decisio"
+version = "0.1.1"
+description = "A prefill-only decision server on vLLM: typed questions about a state, a probability distribution per question"
+license = "Apache-2.0"
+license-files = ["LICENSE", "NOTICE"]
+readme = "README.md"
+authors = [{ name = "Amin Roudaki" }]
+requires-python = ">=3.12,<3.13"
+# The server front, the readout, the tasks and the CPU stand-in; vLLM is the serve extra (Linux, a CUDA card).
+# Lower bounds only: uv.lock pins every version, resolved together with the serve extra, so the CPU tests run the
+# versions vLLM 0.30.0 runs with (it pins torch 2.13.0, torchvision 0.28.0 and fastapi below 0.137).
+# huggingface_hub is imported directly (decisio.serve.hidden_engine) and arrives through transformers; declaring it
+# re-resolves uv.lock beyond that one edge (tilelang loses a never-true setuptools marker), so it is left for a
+# lock refresh.
+dependencies = [
+    "fastapi>=0.133",
+    "numpy>=2.2",
+    "pillow>=11",
+    "psutil>=7",
+    "pydantic>=2.12",
+    "python-multipart>=0.0.20",
+    "safetensors>=0.6.2",
+    "scipy>=1.15",
+    "torch>=2.13",
+    "torchvision>=0.28",
+    "transformers>=5.10.4",
+    "uvicorn>=0.34",
+]
+
+[project.optional-dependencies]
+# the GPU serving stack; the decisio plugin registers its model classes on exactly this vLLM version
+serve = ["vllm==0.30.0; sys_platform == 'linux'"]
+# the benchmark stage (benchmarks/): the conformance client and the HTTP drivers; the JevBench harness and the
+# Decision Index kit are fetched at pinned commits by benchmarks/fetch_harnesses.sh
+bench = ["httpx>=0.28", "typesafe-sdk==0.7.0"]
+dev = ["httpx>=0.28", "pytest==9.1.1", "ruff==0.14.0"]
+
+[project.urls]
+Repository = "https://github.com/aminry/decisio"
+
+# vLLM loads this in every process it starts (docs/design/vllm-plugin.md); it registers decisio's model classes on
+# vllm==0.30.0 and nothing on any other version
+[project.entry-points."vllm.general_plugins"]
+decisio = "decisio.vllm_plugin:register"
+
+[tool.setuptools.packages.find]
+where = ["src"]
+
+[tool.setuptools.package-data]
+"decisio.bench" = ["*.json"]
+
+[tool.pytest.ini_options]
+testpaths = ["tests"]
+addopts = "--strict-markers -m 'not gpu and not slow'"
+markers = [
+    "gpu: needs a CUDA card, vLLM 0.30.0 and a local checkpoint (tests/gpu; run with -m gpu)",
+    "slow: loads a small Hugging Face model on the CPU",
+]
+
+[tool.ruff]
+line-length = 120
+target-version = "py312"
+
+[tool.ruff.lint]
+select = ["E", "F", "I", "W"]

decisio-0.1.1/setup.cfg

@@ -0,0 +1,4 @@
+[egg_info]
+tag_build = 
+tag_date = 0
+

decisio-0.1.1/src/decisio/__init__.py

@@ -0,0 +1,13 @@
+# SPDX-License-Identifier: Apache-2.0
+# SPDX-FileCopyrightText: Copyright contributors to the decisio project
+"""decisio: a prefill-only decision server on vLLM.
+
+  decisio.serve         the HTTP server (`python -m decisio.serve.vllm_engine`), its routes, tasks and engines
+  decisio.readout       the letters readout and the per-task corrections (calibration, intent head)
+  decisio.vllm_plugin   the model classes vLLM loads through the `vllm.general_plugins` entry point
+  decisio.bench         the benchmark stage: JevBench v1.5 open-set scoring, Decision Index reports, latency
+
+Importing `decisio` pulls in nothing beyond the standard library.
+"""
+
+__version__ = "0.1.1"

decisio-0.1.1/src/decisio/bench/__init__.py

@@ -0,0 +1,3 @@
+# SPDX-License-Identifier: Apache-2.0
+# SPDX-FileCopyrightText: Copyright contributors to the decisio project
+"""The benchmark stage: JevBench v1.5 open-set scoring, Decision Index reports and rows, the latency grid."""