Remove internal experiment terminology and clarify model usage

README.md

@@ -39,8 +39,8 @@ Supply context, a question, and the allowed answers. The model scores those answ
 Use it for workflow routing, tool-call validation, evidence checks, rubric scoring, and action selection.
 Questions and answer meanings are supplied at request time, rather than fixed to a single classification task.
 
-The model was fine-tuned with PyTorch and evaluated using [Eider](https://github.com/rdaum/eider), optimized for DGX Spark.
-The GGUF also works with llama.cpp's native decision endpoint. Use the [Dory support PR](https://github.com/ggml-org/llama.cpp/pull/30306) branch or an upstream version that includes it once merged.
+The model was fine-tuned with PyTorch. It has been evaluated with [Eider](https://github.com/rdaum/eider), an inference server optimized for DGX Spark, and llama.cpp.
+The GGUF supports llama.cpp's native decision endpoint in builds that include [Dory support](https://github.com/ggml-org/llama.cpp/pull/30306).
 
 ## Model and formats
 
@@ -50,7 +50,7 @@ The GGUF also works with llama.cpp's native decision endpoint. Use the [Dory sup
 | Adaptation | Rank-8 LoRA, merged into the supplied weights |
 | Recurrent computation | Two internal loops per inference |
 | Outputs | Binary probability, named-choice distribution, or ordered-score distribution |
-| Probability calibration | No fitted calibration profile |
+| Probability calibration | Uncalibrated answer probabilities |
 | Licence | [OpenMDW-1.1](LICENSE) |
 
 | Format | Files | Use |
@@ -61,8 +61,8 @@ The GGUF also works with llama.cpp's native decision endpoint. Use the [Dory sup
 | LoRA adapter | [`adapter/`](adapter/README.md) | Standalone adaptation for the original NVIDIA BF16 checkpoint. |
 
 All formats contain the same fine-tuned model. NVFP4 and Q8_0 are quantized from the merged BF16 weights.
-The adapter uses the supplied Python loader's tensor format, not PEFT format.
-Apply it only to the original NVIDIA checkpoint; the release weights already include it.
+The standalone adapter requires the supplied Python loader; it is not in PEFT format.
+Apply it only to the original NVIDIA checkpoint. The BF16, NVFP4, and GGUF files already include the adaptation.
 
 ## How typed decisions work
 
@@ -100,10 +100,10 @@ Source datasets are not redistributed here.
 
 | Area | Source | Training task |
 |---|---|---|
-| General decisions | [Open-Jev](https://huggingface.co/datasets/ZefanCai/Open-Jev), `release-v2-redistributable` | Control, workflow, reasoning, and game questions with supplied target distributions. |
+| General decisions | [Open-Jev](https://huggingface.co/datasets/ZefanCai/Open-Jev) | Control, workflow, reasoning, and game questions with supplied target distributions. |
 | General decisions | [Jev Distill Corpus v3](https://huggingface.co/datasets/SargeDev/jev-distill-corpus-v3), `yuri_v3` subset | Diverse typed questions with supplied probability targets. |
 | General decisions | [Procedural Typed Decisions](https://huggingface.co/datasets/tasksource/procedural-typed-decisions) | Rule-based retrieval, arithmetic, routing, state tracking, and planning. |
-| Inference and controls | [Open-Jev-v1.1](https://huggingface.co/datasets/ZefanCai/Open-Jev-v1.1), `community-hard-mix-v2-redistributable` | WANLI-derived inference and controlled tool, document, support, and rubric scenarios. |
+| Inference and controls | [Open-Jev-v1.1](https://huggingface.co/datasets/ZefanCai/Open-Jev-v1.1) | WANLI-derived inference and controlled tool, document, support, and rubric scenarios. |
 | Tools | [ToolACE](https://huggingface.co/datasets/Team-ACE/ToolACE) | Assess reference tool calls and variants with invalid arguments. |
 | Evidence grounding | [RAGTruth](https://github.com/ParticleMedia/RAGTruth) | Judge response support using documents and human span annotations. |
 | Reading | [RACE](https://huggingface.co/datasets/ehovy/race) | Select answers to reading-comprehension questions. |
@@ -114,10 +114,10 @@ Source datasets are not redistributed here.
 | Board games | [Competitive Game Dataset](https://huggingface.co/datasets/yyyyyyjjjjzzz/competitive-game-dataset), Connect Four | Select solver-labelled moves, preserving ties between optimal actions. |
 | Interactive tasks | [ScienceWorld gold paths](https://github.com/allenai/ScienceWorld/tree/e8216d6044e8e39be9fcb185e3b2dfb602584b52/goldpaths) | Select among 17 operation types from the goal, room, inventory, and past results. |
 | State tracking | [SCONE](https://nlp.stanford.edu/projects/scone/), Alchemy and Tangrams | Infer final world states from instruction sequences. |
-| Planning | Independent Eider generators motivated by [PlanBench](https://github.com/karthikv792/LLMs-Planning/tree/main/plan-bench) | Assess legality and choose optimal moves in block stacking and package delivery. |
+| Planning | Synthetic planning tasks inspired by [PlanBench](https://github.com/karthikv792/LLMs-Planning/tree/main/plan-bench) | Assess legality and choose optimal moves in block stacking and package delivery. |
 | Human judgements | [NVIDIA HelpSteer2](https://huggingface.co/datasets/nvidia/HelpSteer2) | Learn correctness, helpfulness, coherence, and preference distributions from human ratings. |
 
-Planning examples use independently generated problems and search-derived targets, not published PlanBench instances or generator code.
+Planning tasks use independently generated problems and search-derived targets. Published PlanBench instances and generator code are not used.
 Training and evaluation keep related source groups together, such as episodes, games, documents, conversations, and equivalent planning problems.
 Public test examples are excluded from training.
 
@@ -134,28 +134,28 @@ NVFP4 uses packed E2M1 weights with E4M3 block scales and an F32 tensor scale in
 Both comparisons use the same 512 questions, two recurrent loops, the same typed-decision prompts, and no probability calibration.
 Within each comparison, base Dory and the fine-tune use the same quantization and runtime.
 
-### Eider — NVFP4
-The 512-question test set was scored after checkpoint selection on a separate 1,024-question validation set.
+The model was selected using a separate 1,024-question validation set before the test questions were scored.
+
+### NVFP4 — evaluated with Eider
 
 Base Dory is the original NVIDIA BF16 model quantized with the same NVFP4 recipe used for this fine-tune.
-Both models use the same questions, typed-decision prompt, answer labels, two loops, and inference settings.
-This comparison measures the effect of fine-tuning within this decision interface, rather than Dory’s performance with other prompting methods.
+These results measure decisions with the supplied prompt format and allowed-answer scoring. They do not measure the base model with other prompting methods.
 [Comparison results and hashes](evaluation/base-comparison.json) record the matched setup.
 
-| Test task | Questions | Base Dory accuracy | Typed Decisions accuracy | Base Brier ↓ | Typed Decisions Brier ↓ |
+| Test task | Questions | Base Dory accuracy | Fine-tuned accuracy | Base Brier ↓ | Fine-tuned Brier ↓ |
 |---|---:|---:|---:|---:|---:|
 | SCONE state tracking | 192 | 29.69% | 75.52% | 1.0003 | 0.3779 |
 | Block-stacking and delivery planning | 192 | 42.71% | 92.19% | 0.6045 | 0.1909 |
 | HelpSteer2 human judgements | 128 | 47.66% | 50.78% | 0.3385 | 0.2290 |
 | **Combined** | 512 | 39.06% | 75.59% | 0.6865 | 0.2706 |
 
-### llama.cpp — Q8_0
+### Q8_0 GGUF — evaluated with llama.cpp
 
 The native `/v1/systemone` endpoint scored both models with Q8_0 weights.
 All 512 questions use identical prompt tokens and answer labels across both runtimes.
 This measures fine-tuning under llama.cpp; differences from the NVFP4 table can reflect both quantization and runtime arithmetic.
 
-| Test task | Questions | Base Dory accuracy | Typed Decisions accuracy | Base Brier ↓ | Typed Decisions Brier ↓ |
+| Test task | Questions | Base Dory accuracy | Fine-tuned accuracy | Base Brier ↓ | Fine-tuned Brier ↓ |
 |---|---:|---:|---:|---:|---:|
 | SCONE state tracking | 192 | 31.77% | 76.04% | 0.8954 | 0.3967 |
 | Block-stacking and delivery planning | 192 | 38.02% | 96.88% | 0.6506 | 0.1405 |
@@ -168,10 +168,7 @@ Accuracy counts any answer tied for the highest target probability as correct.
 Brier error compares the complete predicted and target distributions.
 These are adapted decision tasks, not scores under the source datasets' original benchmark protocols.
 
-Additional reused development sets scored 79.35% on 2,048 questions and 65.17% on 89 general questions.
-They are not independent final tests. [Full evaluation records](evaluation/nvfp4-comparison.json) contain the per-source results and checkpoint comparisons.
-The merged BF16 weights have not received a separate full accuracy evaluation.
-
+The BF16 weights have not received the same accuracy evaluation.
 
 ## Limitations
 
@@ -186,9 +183,8 @@ Changing the prompt, label mapping, loop count, or quantization can change the r
 
 ## Use the GGUF with llama.cpp
 
-The GGUF works with llama.cpp's native `/v1/systemone` decision endpoint. It can serve decisions directly, without an Eider server.
-**Dory architecture support is still in NVIDIA's unmerged [llama.cpp PR #30306](https://github.com/ggml-org/llama.cpp/pull/30306).**
-Use that branch until architecture support reaches upstream. This requirement also applies to ordinary generation.
+Use a llama.cpp build that includes Dory architecture support to serve the GGUF through `/v1/systemone`.
+The build instructions below use the [Dory support PR](https://github.com/ggml-org/llama.cpp/pull/30306) branch. Once merged, an upstream build containing that change also works.
 
 ### Download
 
@@ -228,7 +224,7 @@ cmake --build llama.cpp-dory/build --target llama-server -j 6
 ./llama.cpp-dory/build/bin/llama-server \
     --model ./dory-decisions/dory-decisions-Q8_0.gguf \
     --alias dory-decisions \
-    --host 127.0.0.1 --port 18091 \
+    --host 127.0.0.1 --port 8080 \
     --ctx-size 4096 --gpu-layers all
 ```
 
@@ -238,15 +234,13 @@ Omit that option for CPU-only use. Adjust `--ctx-size` for your workload and ava
 ### Ask a typed question
 
 ```sh
-curl http://127.0.0.1:18091/v1/systemone \
+curl http://127.0.0.1:8080/v1/systemone \
     -H 'Content-Type: application/json' \
     -d '{"state":"The key is red.","questions":{"colour":{"type":"choice","instructions":"What colour is the key?","criteria":{"red":"Red","blue":"Blue"}}}}'
 ```
 
 The endpoint returns answer probabilities with zero generated tokens.
-The GGUF provides a named `systemone` prompt template and selects llama.cpp's existing `granite-switch` decision convention.
-This metadata selects the answer-label convention; the model architecture remains Dory.
-The normal chat template is also retained.
+The GGUF includes the decision prompt template and answer-label mapping. It also retains the normal chat template.
 
 | Question type | Supported input |
 |---|---|
@@ -254,15 +248,12 @@ The normal chat template is also retained.
 | `noul` | True/false, with optional text descriptions; defaults are `Yes` and `No`. |
 | `score` | 2–10 ordered text descriptions. |
 
-This llama.cpp convention supports 10 score levels; Eider supports 64.
-The tested server rejects excess score levels with HTTP 400. Template validation failures can return HTTP 500.
 The template accepts text, JSON states, and plain text chat. Serialize tool-call history as state text.
 Image and audio inputs are unsupported.
 
 ## Attribution and provenance
 
 NVIDIA supplies the base model; the linked dataset authors supply the training sources.
-[Eider](https://github.com/rdaum/eider) supplies the evaluated inference implementation.
 This is an independent derivative under [OpenMDW-1.1](LICENSE); see [NOTICE](NOTICE).
 
 [BF16 provenance](provenance.json), [NVFP4 provenance](nvfp4/provenance.json), and [file hashes](formats.json) identify the source weights and adapter.

adapter/README.md

@@ -1,7 +1,7 @@
-# Eider Dory adapter
+# Dory Typed Decisions — LoRA adapter
 
-This adapter uses Eider tensor names. It is not a PEFT adapter.
-The complete model weights already include these changes. Do not apply the adapter twice.
+This rank-8 LoRA adapter uses the tensor format supported by the included `dory_lora.py` loader. It is not in PEFT format.
+The BF16, NVFP4, and GGUF model files already include the adaptation. Use this standalone adapter only with the original NVIDIA BF16 weights.
 
 To load the adapter with the supplied Python module, use the original BF16 snapshot:
 

nvfp4/adapter/README.md

@@ -1,7 +1,7 @@
-# Eider Dory adapter
+# Dory Typed Decisions — LoRA adapter
 
-This adapter uses Eider tensor names. It is not a PEFT adapter.
-The complete model weights already include these changes. Do not apply the adapter twice.
+This rank-8 LoRA adapter uses the tensor format supported by the included `dory_lora.py` loader. It is not in PEFT format.
+The BF16, NVFP4, and GGUF model files already include the adaptation. Use this standalone adapter only with the original NVIDIA BF16 weights.
 
 To load the adapter with the supplied Python module, use the original BF16 snapshot: