Sync model card metadata

MODEL_CARD.md

@@ -1,228 +1,228 @@
----
-license: apache-2.0
-language:
-- en
-library_name: transformers
-pipeline_tag: text-classification
-base_model:
-- Qwen/Qwen3-1.7B
-- Hanno-Labs/bosun-v3.1-1.7b
-datasets:
-- LocalLLaMA/typed-decisions
-- SargeDev/jev-distill-corpus-v3
-tags:
-- decision-model
-- system-one
-- calibrated-decisions
-- typed-decisions
-- choice
-- noul
-- score
-- qwen3
-- bosun
-- safetensors
-- pytorch
-model-index:
-- name: Pluto-2.1-1.7B
-  results:
-  - task:
-      type: text-classification
-      name: Typed Decisions
-    dataset:
-      name: LocalLLaMA/typed-decisions
-      type: LocalLLaMA/typed-decisions
-    metrics:
-    - type: accuracy
-      name: Accuracy
-      value: 0.6695
-    - type: brier
-      name: Brier
-      value: 0.1261
-    - type: kl
-      name: KL
-      value: 0.2139
-  - task:
-      type: text-classification
-      name: DecisionBench matched subset
-    dataset:
-      name: Hanno-Labs/decision-bench
-      type: Hanno-Labs/decision-bench
-    metrics:
-    - type: accuracy
-      name: Accuracy
-      value: 0.7208
-    - type: brier
-      name: Brier
-      value: 0.3730
-    - type: nll
-      name: NLL
-      value: 0.7439
----
-
-# Pluto-2.1
-
-### A 1.7B decision model that beats d1-2B on both measured decision benchmarks.
-
-Pluto-2.1 reads a state and typed questions, then returns a calibrated probability
-distribution for every candidate. It does not generate an answer, so there is
-nothing to parse, nothing to retry, and nothing to hallucinate.
-
-Built solo on one **RTX 5060 Ti 16GB**.
-
-**Headline:** **0.6695** on Typed Decisions and **0.7208** on the matched
-DecisionBench comparison, versus **0.6325** and **0.6339** for d1-2B. One
-standalone 1.7B checkpoint, **2–255 candidates**, **choice / noul / score**,
-**Apache-2.0**.
-
----
-
-## The receipts
-
-Same rows and same scorer for both models. Higher is better.
-
-| Model | Params | Typed Decisions | DecisionBench matched | Brier | KL |
-|---|---:|---:|---:|---:|---:|
-| **Pluto-2.1** | **1.738B** | **0.6695** | **0.7208** | **0.1261** | **0.2139** |
-| d1-2B | 1.918B | 0.6325 | 0.6339 | 0.1440 | 0.2958 |
-
-**What this means**
-
-- **+3.7 points** on the full 2,000-decision Typed Decisions test.
-- **+8.7 points** on the matched DecisionBench comparison.
-- **9.4% fewer parameters** than d1-2B.
-- **Sharper distributions**: lower Brier and KL on Typed Decisions, and lower
-  NLL and Brier on DecisionBench.
-- **One forward pass.** No autoregressive decoding and no token bill.
-
-![Accuracy](assets/benchmark_accuracy.png)
-
-![Uplift](assets/uplift.png)
-
-![Calibration](assets/calibration.png)
-
-![Parameters](assets/parameters.png)
-
-### Comparison hygiene
-
-Pluto-2.1 is a standalone merged checkpoint. The full 2,000-decision Typed
-Decisions test and a 1,565-row DecisionBench matched subset are reported. The
-DecisionBench subset is the same ≤16-candidate, ≤2,048-token accessibility slice
-used for the d1-2B comparison; it is not the full 23,900-row suite. The full
-suite is not claimed here.
-
----
-
-## Architecture
-
-![Architecture](assets/architecture.png)
-
-- Qwen3-1.7B backbone with 256 stable decision-token slots.
-- The fine-tuned adapter is merged into the base; this repository is self-contained.
-- One forward pass scores every presented candidate.
-- Up to 255 runtime choices per question.
-- No d1 weights, no d1-generated training targets, no ensemble, and no runtime
-  dependency on another model service.
-
-**Lineage:** Qwen3-1.7B → Hanno-Labs Bosun 3.1 1.7B → Pluto-2.1 2,000-step
-fine-tune → merged standalone checkpoint. This is disclosed because it is true;
-the release is not presented as trained from random initialization.
-
----
-
-## Quickstart
-
-```bash
-git clone https://huggingface.co/omida-ai/pluto-2.1
-pip install -e "pluto-2.1/code"
-```
-
-```python
-from pluto21 import Pluto21
-
-model = Pluto21("pluto-2.1")
-out = model.predict(
-    state={"transaction": "Duplicate charge confirmed by the bank."},
-    instructions="Was the duplicate charge confirmed?",
-    candidates=[
-        {"id": "yes", "label": "Yes"},
-        {"id": "no", "label": "No"},
-    ],
-    decision_type="noul",
-    row_id="demo-1",
-)
-print(out["probabilities"])
-```
-
-```python
-out = model.predict(
-    state={"ticket": "Customer was charged twice and the parcel is late."},
-    instructions="Which team should handle this?",
-    candidates=[
-        {"id": "billing", "label": "Billing", "description": "Charges and refunds"},
-        {"id": "shipping", "label": "Shipping", "description": "Delivery and parcels"},
-        {"id": "technical", "label": "Technical", "description": "App or site faults"},
-    ],
-    decision_type="choice",
-    row_id="demo-2",
-)
-print(out["probabilities"])
-```
-
-The runtime returns candidate-aligned probabilities. It never generates text.
-
----
-
-## Model details
-
-| | |
-|---|---|
-| Parameters | **1,737,985,024** (1.738B) |
-| Backbone | Qwen3-1.7B, 28 layers, 40,960-token context |
-| Decision interface | 256 stable decision-token slots |
-| Candidates | 2–255 per question |
-| Outputs | `choice`, `noul`, `score` distributions |
-| Weights | BF16, 3.44GB |
-| License | Apache-2.0 |
-| Initialization | Bosun 3.1 1.7B (Apache-2.0) |
-| Base | Qwen3-1.7B |
-| Runtime | `pluto21`, PyTorch + Transformers |
-
-## Training
-
-- 2,000 LoRA steps on Pluto’s Typed Decisions and JEV decision data.
-- Selected checkpoint: 2,000 steps, the only tested checkpoint that beats d1-2B
-  on both measured benchmarks.
-- Trained and merged on one 16GB consumer GPU.
-- The adapter is merged into the base; no PEFT dependency at inference.
-- The Typed Decisions test split was never used for training or checkpoint
-  selection.
-
-## Scope
-
-Pluto-2.1 is a decision model, not a chat or reasoning model. The DecisionBench
-comparison is the matched accessibility subset; the full 23,900-row suite is not
-claimed. As with any benchmark, validate on your own data before high-stakes use.
-
-## License and attribution
-
-Pluto-2.1 weights and runtime are released under **Apache-2.0**. Upstream
-components remain under their original licenses:
-
-- Qwen3-1.7B, Apache-2.0
-- Hanno-Labs Bosun 3.1 1.7B, Apache-2.0
-
-See `LICENSE`, `NOTICE`, and `serving.json`. No d1 weights or outputs are
-included, and d1 is not required at inference.
-
-## Citation
-
-```bibtex
-@misc{omida2026pluto21,
-  title  = {Pluto-2.1: A 1.7B Standalone Decision Model},
-  author = {Omida AI},
-  year   = {2026},
-  url    = {https://huggingface.co/omida-ai/pluto-2.1}
-}
-```
-
-Built by **Omida AI** on one 16GB consumer GPU.
+---
+license: apache-2.0
+language:
+- en
+library_name: pytorch
+pipeline_tag: text-classification
+base_model:
+- Qwen/Qwen3-1.7B
+- Hanno-Labs/bosun-v3.1-1.7b
+datasets:
+- LocalLLaMA/typed-decisions
+- SargeDev/jev-distill-corpus-v3
+tags:
+- decision-model
+- system-one
+- calibrated-decisions
+- typed-decisions
+- choice
+- noul
+- score
+- qwen3
+- bosun
+- safetensors
+- pytorch
+model-index:
+- name: Pluto-2.1-1.7B
+  results:
+  - task:
+      type: text-classification
+      name: Typed Decisions
+    dataset:
+      name: LocalLLaMA/typed-decisions
+      type: LocalLLaMA/typed-decisions
+    metrics:
+    - type: accuracy
+      name: Accuracy
+      value: 0.6695
+    - type: brier
+      name: Brier
+      value: 0.1261
+    - type: kl
+      name: KL
+      value: 0.2139
+  - task:
+      type: text-classification
+      name: DecisionBench matched subset
+    dataset:
+      name: Hanno-Labs/decision-bench
+      type: Hanno-Labs/decision-bench
+    metrics:
+    - type: accuracy
+      name: Accuracy
+      value: 0.7208
+    - type: brier
+      name: Brier
+      value: 0.3730
+    - type: nll
+      name: NLL
+      value: 0.7439
+---
+
+# Pluto-2.1
+
+### A 1.7B decision model that beats d1-2B on both measured decision benchmarks.
+
+Pluto-2.1 reads a state and typed questions, then returns a calibrated probability
+distribution for every candidate. It does not generate an answer, so there is
+nothing to parse, nothing to retry, and nothing to hallucinate.
+
+Built solo on one **RTX 5060 Ti 16GB**.
+
+**Headline:** **0.6695** on Typed Decisions and **0.7208** on the matched
+DecisionBench comparison, versus **0.6325** and **0.6339** for d1-2B. One
+standalone 1.7B checkpoint, **2–255 candidates**, **choice / noul / score**,
+**Apache-2.0**.
+
+---
+
+## The receipts
+
+Same rows and same scorer for both models. Higher is better.
+
+| Model | Params | Typed Decisions | DecisionBench matched | Brier | KL |
+|---|---:|---:|---:|---:|---:|
+| **Pluto-2.1** | **1.738B** | **0.6695** | **0.7208** | **0.1261** | **0.2139** |
+| d1-2B | 1.918B | 0.6325 | 0.6339 | 0.1440 | 0.2958 |
+
+**What this means**
+
+- **+3.7 points** on the full 2,000-decision Typed Decisions test.
+- **+8.7 points** on the matched DecisionBench comparison.
+- **9.4% fewer parameters** than d1-2B.
+- **Sharper distributions**: lower Brier and KL on Typed Decisions, and lower
+  NLL and Brier on DecisionBench.
+- **One forward pass.** No autoregressive decoding and no token bill.
+
+![Accuracy](assets/benchmark_accuracy.png)
+
+![Uplift](assets/uplift.png)
+
+![Calibration](assets/calibration.png)
+
+![Parameters](assets/parameters.png)
+
+### Comparison hygiene
+
+Pluto-2.1 is a standalone merged checkpoint. The full 2,000-decision Typed
+Decisions test and a 1,565-row DecisionBench matched subset are reported. The
+DecisionBench subset is the same ≤16-candidate, ≤2,048-token accessibility slice
+used for the d1-2B comparison; it is not the full 23,900-row suite. The full
+suite is not claimed here.
+
+---
+
+## Architecture
+
+![Architecture](assets/architecture.png)
+
+- Qwen3-1.7B backbone with 256 stable decision-token slots.
+- The fine-tuned adapter is merged into the base; this repository is self-contained.
+- One forward pass scores every presented candidate.
+- Up to 255 runtime choices per question.
+- No d1 weights, no d1-generated training targets, no ensemble, and no runtime
+  dependency on another model service.
+
+**Lineage:** Qwen3-1.7B → Hanno-Labs Bosun 3.1 1.7B → Pluto-2.1 2,000-step
+fine-tune → merged standalone checkpoint. This is disclosed because it is true;
+the release is not presented as trained from random initialization.
+
+---
+
+## Quickstart
+
+```bash
+git clone https://huggingface.co/omida-ai/pluto-2.1
+pip install -e "pluto-2.1/code"
+```
+
+```python
+from pluto21 import Pluto21
+
+model = Pluto21("pluto-2.1")
+out = model.predict(
+    state={"transaction": "Duplicate charge confirmed by the bank."},
+    instructions="Was the duplicate charge confirmed?",
+    candidates=[
+        {"id": "yes", "label": "Yes"},
+        {"id": "no", "label": "No"},
+    ],
+    decision_type="noul",
+    row_id="demo-1",
+)
+print(out["probabilities"])
+```
+
+```python
+out = model.predict(
+    state={"ticket": "Customer was charged twice and the parcel is late."},
+    instructions="Which team should handle this?",
+    candidates=[
+        {"id": "billing", "label": "Billing", "description": "Charges and refunds"},
+        {"id": "shipping", "label": "Shipping", "description": "Delivery and parcels"},
+        {"id": "technical", "label": "Technical", "description": "App or site faults"},
+    ],
+    decision_type="choice",
+    row_id="demo-2",
+)
+print(out["probabilities"])
+```
+
+The runtime returns candidate-aligned probabilities. It never generates text.
+
+---
+
+## Model details
+
+| | |
+|---|---|
+| Parameters | **1,737,985,024** (1.738B) |
+| Backbone | Qwen3-1.7B, 28 layers, 40,960-token context |
+| Decision interface | 256 stable decision-token slots |
+| Candidates | 2–255 per question |
+| Outputs | `choice`, `noul`, `score` distributions |
+| Weights | BF16, 3.44GB |
+| License | Apache-2.0 |
+| Initialization | Bosun 3.1 1.7B (Apache-2.0) |
+| Base | Qwen3-1.7B |
+| Runtime | `pluto21`, PyTorch + Transformers |
+
+## Training
+
+- 2,000 LoRA steps on Pluto’s Typed Decisions and JEV decision data.
+- Selected checkpoint: 2,000 steps, the only tested checkpoint that beats d1-2B
+  on both measured benchmarks.
+- Trained and merged on one 16GB consumer GPU.
+- The adapter is merged into the base; no PEFT dependency at inference.
+- The Typed Decisions test split was never used for training or checkpoint
+  selection.
+
+## Scope
+
+Pluto-2.1 is a decision model, not a chat or reasoning model. The DecisionBench
+comparison is the matched accessibility subset; the full 23,900-row suite is not
+claimed. As with any benchmark, validate on your own data before high-stakes use.
+
+## License and attribution
+
+Pluto-2.1 weights and runtime are released under **Apache-2.0**. Upstream
+components remain under their original licenses:
+
+- Qwen3-1.7B, Apache-2.0
+- Hanno-Labs Bosun 3.1 1.7B, Apache-2.0
+
+See `LICENSE`, `NOTICE`, and `serving.json`. No d1 weights or outputs are
+included, and d1 is not required at inference.
+
+## Citation
+
+```bibtex
+@misc{omida2026pluto21,
+  title  = {Pluto-2.1: A 1.7B Standalone Decision Model},
+  author = {Omida AI},
+  year   = {2026},
+  url    = {https://huggingface.co/omida-ai/pluto-2.1}
+}
+```
+
+Built by **Omida AI** on one 16GB consumer GPU.