vegaml 0.1.0__tar.gz

vegaml-0.1.0/LICENSE

@@ -0,0 +1,191 @@
+
+                                 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
+
+   Copyright 2026 Nandakishor M
+
+   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.

vegaml-0.1.0/PKG-INFO

@@ -0,0 +1,251 @@
+Metadata-Version: 2.4
+Name: vegaml
+Version: 0.1.0
+Summary: Typed decisions from a frozen language model and a small physics engine: calibrated probabilities, conformal sets, 73k context, images.
+Author: Nandakishor M
+License-Expression: Apache-2.0
+Project-URL: Homepage, https://github.com/NandhaKishorM/vegaml
+Project-URL: Issues, https://github.com/NandhaKishorM/vegaml/issues
+Keywords: typed-decisions,calibration,conformal-prediction,decision-model,long-context
+Classifier: Development Status :: 4 - Beta
+Classifier: Intended Audience :: Developers
+Classifier: Intended Audience :: Science/Research
+Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3.10
+Classifier: Programming Language :: Python :: 3.11
+Classifier: Programming Language :: Python :: 3.12
+Classifier: Programming Language :: Python :: 3.13
+Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
+Classifier: Typing :: Typed
+Requires-Python: >=3.10
+Description-Content-Type: text/markdown
+License-File: LICENSE
+Requires-Dist: torch>=2.4
+Requires-Dist: transformers>=5.17
+Requires-Dist: safetensors>=0.4
+Requires-Dist: huggingface_hub<2.0,>=0.30
+Requires-Dist: numpy>=1.24
+Provides-Extra: vision
+Requires-Dist: pillow>=10.0; extra == "vision"
+Provides-Extra: dev
+Requires-Dist: pytest>=8.0; extra == "dev"
+Requires-Dist: ruff>=0.6; extra == "dev"
+Requires-Dist: tomli>=2.0; python_version < "3.11" and extra == "dev"
+Dynamic: license-file
+
+# vegaml
+
+Typed decisions from a frozen language model and a small physics engine. You give it a **state** and a
+**question whose answer type is fixed in advance**; it returns a value your code can use directly — a
+choice, a 0–1 score, or the probability that a statement is true — each with a calibrated probability,
+a conformal answer set and an explicit abstain flag.
+
+**800M parameters · 73,728-token context · images · runs on your own hardware.** No text is generated
+anywhere in the path.
+
+![Where this model beats a hosted reference](assets/vega_wins.png)
+
+```bash
+pip install vegaml
+```
+
+<sub>Releases are cut by tagging `vX.Y.Z`: CI builds, checks the version matches the tag, and
+publishes through PyPI Trusted Publishing — no API token exists in this repository.</sub>
+
+## Quick start
+
+```python
+import vegaml
+
+v = vegaml.load("nandakishorm/vega-08b-public-intents")      # mode="engine" by default
+
+out = v.decide(
+    {"from": "billing@acme.com", "subject": "Invoice overdue", "body": "Third notice. Pay now."},
+    {"team":  {"type": "choice", "instructions": "Which team should handle this?",
+               "criteria": {"billing": "payments and invoices",
+                            "technical": "product faults",
+                            "sales": "new business"}},
+     "churn": {"type": "noul", "instructions": "Is this customer at risk of leaving?",
+               "criteria": {"true": "shows intent to cancel", "false": "no such signal"}}})
+
+print(out["answers"]["team"]["choice"])        # 'billing'
+print(out["answers"]["team"]["probs"])         # calibrated, sums to 1
+print(out["answers"]["churn"]["abstain"])      # False -> the engine stands behind it
+```
+
+Both questions are answered from **one** read of the state.
+
+### Two readouts, one set of features
+
+`mode` picks how the pooled features are turned into an answer.
+
+| mode | needs examples | carries | use it when |
+|---|---|---|---|
+| `"engine"` *(default)* | no, zero-shot | calibration, conformal sets, abstain | the normal case, and every published figure below |
+| `"ttt"` | yes, 6 minimum | nothing — it answers even when it should not | you have labels for exactly this question |
+| `"both"` | yes | both, side by side | deciding which to trust |
+
+```python
+examples = [({"body": "cancel my account"}, {"churn": "true"}),
+            ({"body": "how do I export?"},  {"churn": "false"})]            # ... 20 or so
+
+report = v.fit(examples, questions)
+print(report["churn"]["cv_accuracy"], report["churn"]["at_or_below_chance"])
+
+out = v.decide(state, questions, mode="ttt")
+```
+
+`fit` returns a cross-validated accuracy per question. **Read it.** A head at or below chance has
+learned nothing and its probabilities are noise; the engine is the honest answer there.
+
+### Images
+
+```python
+from PIL import Image
+
+v.decide_image(Image.open("invoice.png"),
+               {"kind": {"type": "choice", "instructions": "What kind of document is this?",
+                         "criteria": {"invoice": "a bill", "receipt": "proof of payment",
+                                      "form": "a document to be filled in", "other": "none of these"}}})
+```
+
+The backbone is multimodal and its vision encoder is frozen with the rest, so a picture takes the place
+of the state text in the same prompt and the same spans are pooled. The benchmark figures below are
+text; the image path is functional but not covered by them.
+
+### 73,728-token context
+
+`max_len` is 73,728 tokens, and the architecture is built for it rather than merely permitting it:
+
+* **Read-once prefix caching.** A state of at least 4,096 tokens is encoded **once**; every question
+  about that state continues from a copy of that prefix. Twelve questions about one 73k-token contract
+  cost roughly one read, not twelve.
+* **A memory reader.** Long states are cut into 16 pieces, mean-pooled at the deepest feature layer and
+  fed to the engine alongside the spans, so evidence late in a long document still reaches the decision.
+* **No silent truncation.** An input that will not fit is refused, never quietly shortened.
+
+## Benchmarks
+
+The figure at the top of this page is the table below. Item-paired against a live Jev 1.13.0 over
+identical items, no tuning on the evaluation data. These are the areas where this model leads; the
+reference leads on most others, particularly reranking and multi-step reasoning.
+
+| task | this model | Jev 1.13.0 |
+|---|---|---|
+| Phishing screening, 800 emails | **75.4** acc · **252/400** caught · recall **0.63** | 61.9 · 99/400 · 0.25 |
+| Dates and quantities (temporal_numeric) | **46.7** | 20.0 |
+| Spam detection (enron-spam) | **1.000** | 0.920 |
+| News topic (ag-news) | **0.955** | 0.806 |
+| Typed scores (typed-decisions) | **0.438** | 0.395 |
+| Calibration error, product relevance | **6.0** | 22.0 |
+| Calibration error, overall (5,096 paired) | 9.5 | 9.3 |
+| Median latency per decision | **267 ms** (T4 fp16) · 280 ms (Apple M, fp32) | 591 ms (hosted) |
+
+2.5× the phishing caught at McNemar p = 3e-11, and 2.1× faster with nothing leaving the machine.
+
+## How it works
+
+One prompt, one forward pass, then physics.
+
+**1 — One formatted prompt.** The state, the question and every candidate answer are written into a
+single sequence:
+
+```
+Observation: {state} Measurement ({type}): {instructions} Possible outcomes: * {option 1} * {option 2} Outcome:
+```
+
+The option text is genuinely part of the model's input, not metadata kept outside it. No system prompt,
+no chat template, no demonstrations, and the gold answer never appears — it exists only as a training
+label.
+
+**2 — One forward pass, five pooled vectors.** The frozen language model runs **once** over that
+sequence. Vega then pools different token spans from the *same* hidden states: the situation span, the
+question span, one span per answer option, and the final token. Three questions about one state mean
+three prompts and three passes, except for the long-input case above, where the prefix is shared.
+
+**3 — Pooled vectors become initial conditions.** The situation vector projects to a world latent and
+gets a bounded nonlinear nudge; the question and the final token project to a probe and an impulse.
+Together they place a particle at position `z₀` with momentum `p₀` in a 64-dimensional decision space.
+
+**4 — Every candidate answer becomes a valley.** Each option's own features produce a Gaussian well —
+a centre `c_k`, a depth `a_k` and a width `σ_k`:
+
+```
+U(z) = ½κ‖z‖² − Σ_k a_k · exp( −‖z − c_k‖² / 2σ_k² )
+```
+
+The quadratic term keeps the particle bounded; each well pulls it toward one answer. Score levels sit on
+a one-dimensional rail, so ordinal neighbours are physical neighbours.
+
+**5 — The particle rolls and settles.** Damped Hamiltonian dynamics — symplectic Euler with friction and
+a learned state-space thermostat — for a fixed, small step budget, with early exit once it has settled.
+A decision is a short simulation with constant cost, not a sampling loop, so it is deterministic.
+
+**6 — Where it settles is the answer.**
+
+```
+E_k = ‖z_T − c_k‖² / 2σ_k² − log a_k            P = softmax(−E / τ)
+```
+
+### What is new here
+
+**The architecture.** The language model is frozen and never trained — it is perception only, read at
+two intermediate layers and cut off above the deepest one, so a decision never pays for the layers above
+it and the LM head never runs. Everything learned lives in a 55 MB engine whose state is a *physical*
+one: a position and a momentum, not a logit vector.
+
+**The training method.** The engine is trained on the settling behaviour, not on next-token likelihood:
+a counterfactual objective pairs items that share an answer space, and a one-step world-dynamics block
+is trained to imagine the next state from the current one, so the latent carries what happens next
+rather than only what was said. Two low-rank adapters (rank 32) attach to seven engine projections, and
+a per-question sigmoid gate decides **for each question independently** whether an adapter contributes.
+The gate value and the chosen adapter are returned with every answer, so routing is auditable rather
+than implicit.
+
+**The calibration method.** The readout temperature is not a constant. It is predicted per decision from
+the physical state the particle ended in:
+
+```
+log τ = b + w · [ log(1 + residual kinetic energy),
+                  log(1 + distance to the nearest well bottom),
+                  fraction of the step budget used,
+                  log (number of options) ]
+```
+
+A particle still in motion, or stopped far from every well, is an uncertain decision and gets a hotter
+temperature. Each adapter route carries its own calibration vector. On top of that, every answer has a
+split-conformal set at a chosen risk level, an **abstain** flag below a fitted confidence floor, and an
+**unbound** flag when the particle settled far from every well — the model's own way of saying the
+question is outside what it knows.
+
+## Security
+
+A checkpoint is data from a third party, so the inference code lives in this package and is reviewed
+with it. `vegaml.load()` downloads **only** `vega_config.json`, `engine.safetensors` and the adapter
+files; nothing fetched at runtime is imported or executed, and weights load through safetensors, never
+pickle. Tests enforce that allow-list, and CI fails on any `pickle`, `torch.load`, `eval`, `exec`,
+`subprocess` or `sys.path` insertion reaching the shipped package.
+
+## Repository settings worth knowing
+
+Two things this repository cannot enforce on a private repo without a paid plan, documented here so
+nobody mistakes a gap for a gate:
+
+* **Branch protection.** Both classic protection and rulesets return `403 Upgrade to GitHub Pro or
+  make this repository public`, so the server enforces nothing: a force-push to `main`, a deletion,
+  or a merge over a red check are all possible. The workflows still run on every push and pull
+  request; only the enforcement is missing.
+
+  The nearest available substitute is a local hook, `.githooks/pre-push`, which refuses a force-push
+  or a deletion of `main`. Enable it in each clone with `git config core.hooksPath .githooks`. It
+  stops the accident from a configured clone and **nothing else** — not another machine, not the web
+  UI, not `--no-verify`. Treat it as a seatbelt, not a lock.
+* **CodeQL.** Code scanning needs GitHub Advanced Security on a private repository
+  (`422 Advanced security has not been purchased`), so the CodeQL job reports why it skipped instead
+  of failing forever. Secret scanning, the dependency CVE audit and the supply-chain gate do run.
+
+Making the repository public, or upgrading, turns both on with no change to the workflows.
+
+## Licence
+
+Apache 2.0.

vegaml-0.1.0/README.md

@@ -0,0 +1,216 @@
+# vegaml
+
+Typed decisions from a frozen language model and a small physics engine. You give it a **state** and a
+**question whose answer type is fixed in advance**; it returns a value your code can use directly — a
+choice, a 0–1 score, or the probability that a statement is true — each with a calibrated probability,
+a conformal answer set and an explicit abstain flag.
+
+**800M parameters · 73,728-token context · images · runs on your own hardware.** No text is generated
+anywhere in the path.
+
+![Where this model beats a hosted reference](assets/vega_wins.png)
+
+```bash
+pip install vegaml
+```
+
+<sub>Releases are cut by tagging `vX.Y.Z`: CI builds, checks the version matches the tag, and
+publishes through PyPI Trusted Publishing — no API token exists in this repository.</sub>
+
+## Quick start
+
+```python
+import vegaml
+
+v = vegaml.load("nandakishorm/vega-08b-public-intents")      # mode="engine" by default
+
+out = v.decide(
+    {"from": "billing@acme.com", "subject": "Invoice overdue", "body": "Third notice. Pay now."},
+    {"team":  {"type": "choice", "instructions": "Which team should handle this?",
+               "criteria": {"billing": "payments and invoices",
+                            "technical": "product faults",
+                            "sales": "new business"}},
+     "churn": {"type": "noul", "instructions": "Is this customer at risk of leaving?",
+               "criteria": {"true": "shows intent to cancel", "false": "no such signal"}}})
+
+print(out["answers"]["team"]["choice"])        # 'billing'
+print(out["answers"]["team"]["probs"])         # calibrated, sums to 1
+print(out["answers"]["churn"]["abstain"])      # False -> the engine stands behind it
+```
+
+Both questions are answered from **one** read of the state.
+
+### Two readouts, one set of features
+
+`mode` picks how the pooled features are turned into an answer.
+
+| mode | needs examples | carries | use it when |
+|---|---|---|---|
+| `"engine"` *(default)* | no, zero-shot | calibration, conformal sets, abstain | the normal case, and every published figure below |
+| `"ttt"` | yes, 6 minimum | nothing — it answers even when it should not | you have labels for exactly this question |
+| `"both"` | yes | both, side by side | deciding which to trust |
+
+```python
+examples = [({"body": "cancel my account"}, {"churn": "true"}),
+            ({"body": "how do I export?"},  {"churn": "false"})]            # ... 20 or so
+
+report = v.fit(examples, questions)
+print(report["churn"]["cv_accuracy"], report["churn"]["at_or_below_chance"])
+
+out = v.decide(state, questions, mode="ttt")
+```
+
+`fit` returns a cross-validated accuracy per question. **Read it.** A head at or below chance has
+learned nothing and its probabilities are noise; the engine is the honest answer there.
+
+### Images
+
+```python
+from PIL import Image
+
+v.decide_image(Image.open("invoice.png"),
+               {"kind": {"type": "choice", "instructions": "What kind of document is this?",
+                         "criteria": {"invoice": "a bill", "receipt": "proof of payment",
+                                      "form": "a document to be filled in", "other": "none of these"}}})
+```
+
+The backbone is multimodal and its vision encoder is frozen with the rest, so a picture takes the place
+of the state text in the same prompt and the same spans are pooled. The benchmark figures below are
+text; the image path is functional but not covered by them.
+
+### 73,728-token context
+
+`max_len` is 73,728 tokens, and the architecture is built for it rather than merely permitting it:
+
+* **Read-once prefix caching.** A state of at least 4,096 tokens is encoded **once**; every question
+  about that state continues from a copy of that prefix. Twelve questions about one 73k-token contract
+  cost roughly one read, not twelve.
+* **A memory reader.** Long states are cut into 16 pieces, mean-pooled at the deepest feature layer and
+  fed to the engine alongside the spans, so evidence late in a long document still reaches the decision.
+* **No silent truncation.** An input that will not fit is refused, never quietly shortened.
+
+## Benchmarks
+
+The figure at the top of this page is the table below. Item-paired against a live Jev 1.13.0 over
+identical items, no tuning on the evaluation data. These are the areas where this model leads; the
+reference leads on most others, particularly reranking and multi-step reasoning.
+
+| task | this model | Jev 1.13.0 |
+|---|---|---|
+| Phishing screening, 800 emails | **75.4** acc · **252/400** caught · recall **0.63** | 61.9 · 99/400 · 0.25 |
+| Dates and quantities (temporal_numeric) | **46.7** | 20.0 |
+| Spam detection (enron-spam) | **1.000** | 0.920 |
+| News topic (ag-news) | **0.955** | 0.806 |
+| Typed scores (typed-decisions) | **0.438** | 0.395 |
+| Calibration error, product relevance | **6.0** | 22.0 |
+| Calibration error, overall (5,096 paired) | 9.5 | 9.3 |
+| Median latency per decision | **267 ms** (T4 fp16) · 280 ms (Apple M, fp32) | 591 ms (hosted) |
+
+2.5× the phishing caught at McNemar p = 3e-11, and 2.1× faster with nothing leaving the machine.
+
+## How it works
+
+One prompt, one forward pass, then physics.
+
+**1 — One formatted prompt.** The state, the question and every candidate answer are written into a
+single sequence:
+
+```
+Observation: {state} Measurement ({type}): {instructions} Possible outcomes: * {option 1} * {option 2} Outcome:
+```
+
+The option text is genuinely part of the model's input, not metadata kept outside it. No system prompt,
+no chat template, no demonstrations, and the gold answer never appears — it exists only as a training
+label.
+
+**2 — One forward pass, five pooled vectors.** The frozen language model runs **once** over that
+sequence. Vega then pools different token spans from the *same* hidden states: the situation span, the
+question span, one span per answer option, and the final token. Three questions about one state mean
+three prompts and three passes, except for the long-input case above, where the prefix is shared.
+
+**3 — Pooled vectors become initial conditions.** The situation vector projects to a world latent and
+gets a bounded nonlinear nudge; the question and the final token project to a probe and an impulse.
+Together they place a particle at position `z₀` with momentum `p₀` in a 64-dimensional decision space.
+
+**4 — Every candidate answer becomes a valley.** Each option's own features produce a Gaussian well —
+a centre `c_k`, a depth `a_k` and a width `σ_k`:
+
+```
+U(z) = ½κ‖z‖² − Σ_k a_k · exp( −‖z − c_k‖² / 2σ_k² )
+```
+
+The quadratic term keeps the particle bounded; each well pulls it toward one answer. Score levels sit on
+a one-dimensional rail, so ordinal neighbours are physical neighbours.
+
+**5 — The particle rolls and settles.** Damped Hamiltonian dynamics — symplectic Euler with friction and
+a learned state-space thermostat — for a fixed, small step budget, with early exit once it has settled.
+A decision is a short simulation with constant cost, not a sampling loop, so it is deterministic.
+
+**6 — Where it settles is the answer.**
+
+```
+E_k = ‖z_T − c_k‖² / 2σ_k² − log a_k            P = softmax(−E / τ)
+```
+
+### What is new here
+
+**The architecture.** The language model is frozen and never trained — it is perception only, read at
+two intermediate layers and cut off above the deepest one, so a decision never pays for the layers above
+it and the LM head never runs. Everything learned lives in a 55 MB engine whose state is a *physical*
+one: a position and a momentum, not a logit vector.
+
+**The training method.** The engine is trained on the settling behaviour, not on next-token likelihood:
+a counterfactual objective pairs items that share an answer space, and a one-step world-dynamics block
+is trained to imagine the next state from the current one, so the latent carries what happens next
+rather than only what was said. Two low-rank adapters (rank 32) attach to seven engine projections, and
+a per-question sigmoid gate decides **for each question independently** whether an adapter contributes.
+The gate value and the chosen adapter are returned with every answer, so routing is auditable rather
+than implicit.
+
+**The calibration method.** The readout temperature is not a constant. It is predicted per decision from
+the physical state the particle ended in:
+
+```
+log τ = b + w · [ log(1 + residual kinetic energy),
+                  log(1 + distance to the nearest well bottom),
+                  fraction of the step budget used,
+                  log (number of options) ]
+```
+
+A particle still in motion, or stopped far from every well, is an uncertain decision and gets a hotter
+temperature. Each adapter route carries its own calibration vector. On top of that, every answer has a
+split-conformal set at a chosen risk level, an **abstain** flag below a fitted confidence floor, and an
+**unbound** flag when the particle settled far from every well — the model's own way of saying the
+question is outside what it knows.
+
+## Security
+
+A checkpoint is data from a third party, so the inference code lives in this package and is reviewed
+with it. `vegaml.load()` downloads **only** `vega_config.json`, `engine.safetensors` and the adapter
+files; nothing fetched at runtime is imported or executed, and weights load through safetensors, never
+pickle. Tests enforce that allow-list, and CI fails on any `pickle`, `torch.load`, `eval`, `exec`,
+`subprocess` or `sys.path` insertion reaching the shipped package.
+
+## Repository settings worth knowing
+
+Two things this repository cannot enforce on a private repo without a paid plan, documented here so
+nobody mistakes a gap for a gate:
+
+* **Branch protection.** Both classic protection and rulesets return `403 Upgrade to GitHub Pro or
+  make this repository public`, so the server enforces nothing: a force-push to `main`, a deletion,
+  or a merge over a red check are all possible. The workflows still run on every push and pull
+  request; only the enforcement is missing.
+
+  The nearest available substitute is a local hook, `.githooks/pre-push`, which refuses a force-push
+  or a deletion of `main`. Enable it in each clone with `git config core.hooksPath .githooks`. It
+  stops the accident from a configured clone and **nothing else** — not another machine, not the web
+  UI, not `--no-verify`. Treat it as a seatbelt, not a lock.
+* **CodeQL.** Code scanning needs GitHub Advanced Security on a private repository
+  (`422 Advanced security has not been purchased`), so the CodeQL job reports why it skipped instead
+  of failing forever. Secret scanning, the dependency CVE audit and the supply-chain gate do run.
+
+Making the repository public, or upgrading, turns both on with no change to the workflows.
+
+## Licence
+
+Apache 2.0.