Replace personal project examples with generic SolarFlux/FluxNet

OSPREY/DeepONet/NeuralHamilton → SolarFlux/FluxNet/WavePredict
across README, docs, and skill files.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Author: Axect

.claude/skills/pytorch-train/SKILL.md

@@ -36,9 +36,9 @@ Before creating any files, confirm these with the user:
 
 | Item | Example | Notes |
 |------|---------|-------|
-| Project name | `OSPREY`, `NeuralHamilton` | Used in `project:` field and directory names |
-| Version | `v0.10`, `v1.32` | Determines config subdirectory |
-| Model name | `deeponet`, `mambonet`, `mlp` | File prefix and `net:` path |
+| Project name | `SolarFlux`, `WavePredict` | Used in `project:` field and directory names |
+| Version | `v0.3`, `v1.32` | Determines config subdirectory |
+| Model name | `fluxnet`, `wavenet`, `mlp` | File prefix and `net:` path |
 | Task type | regression / classification | Determines criterion, metric direction |
 | Model module path | `model.MLP`, `recipes.regression.model.MLP` | Importlib path for `net:` field |
 | net_config | `{nodes: 64, layers: 4}` | Architecture hyperparameters |
@@ -86,11 +86,11 @@ configs/<MAIN_CONTRIBUTION>_v<VERSION>/<MODEL_NAME>_{run,opt,best}.yaml
 ```
 
 Examples:
-- `configs/OSPREY_v0.10/deeponet_run.yaml`
-- `configs/OSPREY_v0.10/deeponet_opt.yaml`
-- `configs/OSPREY_v0.10/deeponet_best.yaml`
+- `configs/SolarFlux_v0.3/fluxnet_run.yaml`
+- `configs/SolarFlux_v0.3/fluxnet_opt.yaml`
+- `configs/SolarFlux_v0.3/fluxnet_best.yaml`
 
-The `project:` field follows: `<PROJECT>_v<VERSION>_<MODEL>` (e.g., `OSPREY_v0.10_DeepONet`).
+The `project:` field follows: `<PROJECT>_v<VERSION>_<MODEL>` (e.g., `SolarFlux_v0.3_FluxNet`).
 
 ### 2a: run.yaml (HPO Base Config)
 

.claude/skills/pytorch-train/references/config_templates.md

@@ -8,7 +8,7 @@ Full annotated YAML templates for each config type. Copy and modify as needed.
 
 ```yaml
 # ── Project Identification ──
-project: <PROJECT>_v<VERSION>_<MODEL>    # e.g., OSPREY_v0.10_DeepONet
+project: <PROJECT>_v<VERSION>_<MODEL>    # e.g., SolarFlux_v0.3_FluxNet
 device: cuda:0                            # cuda:N or cpu
 
 # ── Model ──
@@ -62,7 +62,7 @@ checkpoint_config:
 
 ```yaml
 # ── Study ──
-study_name: <MODEL>_TPE                   # e.g., DeepONet_TPE
+study_name: <MODEL>_TPE                   # e.g., FluxNet_TPE
 trials: 50                                # 30-100 depending on search space
 seed: 42
 metric: val_loss
@@ -112,7 +112,7 @@ search_space:
 
 ```yaml
 # ── Project Identification ──
-project: <PROJECT>_v<VERSION>_<MODEL>     # Remove _Opt suffix
+project: <PROJECT>_v<VERSION>_<MODEL>     # Remove _Opt suffix (e.g., SolarFlux_v0.3_FluxNet)
 device: cuda:0
 
 # ── Model ──

README.md

@@ -88,14 +88,14 @@ python -m cli analyze        ← validate results, generate plots
 This template ships with a built-in [Claude Code](https://claude.ai/claude-code) skill that guides you through the entire experiment lifecycle:
 
 ```
-You: "Set up HPO for my DeepONet model, version 0.10"
+You: "Set up HPO for my FluxNet model, version 0.3"
 
-Agent: Creates configs/OSPREY_v0.10/deeponet_run.yaml
-       Creates configs/OSPREY_v0.10/deeponet_opt.yaml
+Agent: Creates configs/SolarFlux_v0.3/fluxnet_run.yaml
+       Creates configs/SolarFlux_v0.3/fluxnet_opt.yaml
        Runs preflight to catch any config issues
        Launches HPO with SPlus + ExpHyperbolicLR defaults
        Runs hpo-report to analyze results
-       Extracts best params → deeponet_best.yaml
+       Extracts best params → fluxnet_best.yaml
        Launches final multi-seed training
 ```
 

README_KR.md

@@ -88,14 +88,14 @@ python -m cli analyze        ← 결과 검증 및 플롯 생성
 이 템플릿에는 전체 실험 생애주기를 안내하는 내장 [Claude Code](https://claude.ai/claude-code) skill이 포함되어 있습니다:
 
 ```
-User: "DeepONet 모델 버전 0.10에 HPO 설정해줘"
+User: "FluxNet 모델 버전 0.3에 HPO 설정해줘"
 
-Agent: configs/OSPREY_v0.10/deeponet_run.yaml 생성
-       configs/OSPREY_v0.10/deeponet_opt.yaml 생성
+Agent: configs/SolarFlux_v0.3/fluxnet_run.yaml 생성
+       configs/SolarFlux_v0.3/fluxnet_opt.yaml 생성
        preflight를 실행해 설정 오류 확인
        최적 SPlus + ExpHyperbolicLR 기본값으로 HPO 실행
        hpo-report를 실행해 결과 분석
-       최적 파라미터 추출 → deeponet_best.yaml
+       최적 파라미터 추출 → fluxnet_best.yaml
        최종 다중 시드 학습 시작
 ```
 

docs/01_pipeline.md

@@ -17,7 +17,7 @@ Here is a complete annotated `run.yaml` for a regression task with SPlus + ExpHy
 
 ```yaml
 # ── Project Identification ──
-project: OSPREY_v0.10_DeepONet   # Convention: <NAME>_v<VERSION>_<MODEL>
+project: SolarFlux_v0.3_FluxNet   # Convention: <NAME>_v<VERSION>_<MODEL>
 device: cuda:0                    # cuda:N or cpu
 
 # ── Model ──
@@ -73,10 +73,10 @@ Change datasets by changing the `data:` field — one line. The function must re
 Config files live under `configs/<MAIN_CONTRIBUTION>_v<VERSION>/`:
 
 ```
-configs/OSPREY_v0.10/
-├── deeponet_run.yaml    # HPO base config
-├── deeponet_opt.yaml    # HPO search config
-└── deeponet_best.yaml   # Final training config (created after HPO)
+configs/SolarFlux_v0.3/
+├── fluxnet_run.yaml    # HPO base config
+├── fluxnet_opt.yaml    # HPO search config
+└── fluxnet_best.yaml   # Final training config (created after HPO)
 ```
 
 See **[Chapter 2: Configuration Deep Dive](02_config.html)** for every field explained.
@@ -88,7 +88,7 @@ See **[Chapter 2: Configuration Deep Dive](02_config.html)** for every field exp
 A single shape mismatch can waste hours of GPU time. Preflight catches it in seconds by running one batch forward and backward through the full stack — data loading, model instantiation, optimizer step, gradient check.
 
 ```bash
-python -m cli preflight configs/OSPREY_v0.10/deeponet_run.yaml --device cuda:0
+python -m cli preflight configs/SolarFlux_v0.3/fluxnet_run.yaml --device cuda:0
 ```
 
 The output is a table of checks:
@@ -114,16 +114,16 @@ Fix every FAIL and investigate every WARN before proceeding. Use `--json` for ma
 Also available before preflight:
 
 ```bash
-python -m cli validate configs/OSPREY_v0.10/deeponet_run.yaml  # Schema + semantic checks only
-python -m cli preview configs/OSPREY_v0.10/deeponet_run.yaml   # Print model architecture
+python -m cli validate configs/SolarFlux_v0.3/fluxnet_run.yaml  # Schema + semantic checks only
+python -m cli preview configs/SolarFlux_v0.3/fluxnet_run.yaml   # Print model architecture
 ```
 
 ---
 
 ## Phase 3: Training
 
 ```bash
-python -m cli train configs/OSPREY_v0.10/deeponet_run.yaml --device cuda:0
+python -m cli train configs/SolarFlux_v0.3/fluxnet_run.yaml --device cuda:0
 ```
 
 What happens when you run this:
@@ -145,11 +145,11 @@ Two diagnostic callbacks run automatically during every training session:
 For long runs, queue with pueue so they survive session termination:
 
 ```bash
-pueue group add OSPREY
-pueue add -g OSPREY -- bash -c \
+pueue group add SolarFlux
+pueue add -g SolarFlux -- bash -c \
   "cd /path/to/project && .venv/bin/python -m cli train \
-   configs/OSPREY_v0.10/deeponet_run.yaml --device cuda:0"
-pueue status -g OSPREY
+   configs/SolarFlux_v0.3/fluxnet_run.yaml --device cuda:0"
+pueue status -g SolarFlux
 ```
 
 ---
@@ -159,7 +159,7 @@ pueue status -g OSPREY
 HPO finds the best hyperparameters by running many short training trials and using the results to guide the search. The optimizer config (`opt.yaml`) defines the search space:
 
 ```yaml
-study_name: DeepONet_TPE
+study_name: FluxNet_TPE
 trials: 50
 seed: 42
 metric: val_loss
@@ -199,8 +199,8 @@ search_space:
 Run HPO:
 
 ```bash
-python -m cli train configs/OSPREY_v0.10/deeponet_run.yaml \
-  --optimize-config configs/OSPREY_v0.10/deeponet_opt.yaml \
+python -m cli train configs/SolarFlux_v0.3/fluxnet_run.yaml \
+  --optimize-config configs/SolarFlux_v0.3/fluxnet_opt.yaml \
   --device cuda:0
 ```
 
@@ -221,12 +221,12 @@ After HPO completes, analyze the results before creating `best.yaml`:
 python -m cli hpo-report
 
 # Explicit — use when multiple studies exist
-python -m cli hpo-report --db OSPREY_v0.10_DeepONet_Opt.db --study-name DeepONet_TPE
+python -m cli hpo-report --db SolarFlux_v0.3_FluxNet_Opt.db --study-name FluxNet_TPE
 
 # With boundary warnings — recommended
 python -m cli hpo-report \
-  --db OSPREY_v0.10_DeepONet_Opt.db \
-  --opt-config configs/OSPREY_v0.10/deeponet_opt.yaml
+  --db SolarFlux_v0.3_FluxNet_Opt.db \
+  --opt-config configs/SolarFlux_v0.3/fluxnet_opt.yaml
 ```
 
 The report shows:
@@ -279,9 +279,9 @@ checkpoint_config:
 Then validate and run:
 
 ```bash
-python -m cli validate configs/OSPREY_v0.10/deeponet_best.yaml
-python -m cli preflight configs/OSPREY_v0.10/deeponet_best.yaml
-python -m cli train configs/OSPREY_v0.10/deeponet_best.yaml --device cuda:0
+python -m cli validate configs/SolarFlux_v0.3/fluxnet_best.yaml
+python -m cli preflight configs/SolarFlux_v0.3/fluxnet_best.yaml
+python -m cli train configs/SolarFlux_v0.3/fluxnet_best.yaml --device cuda:0
 ```
 
 With 5 seeds and 150 epochs, this is a long run. Use pueue.
@@ -293,7 +293,7 @@ With 5 seeds and 150 epochs, this is a long run. Use pueue.
 After training completes, check the run directories and analyze results:
 
 ```bash
-ls runs/OSPREY_v0.10_DeepONet/     # One subdirectory per group name
+ls runs/SolarFlux_v0.3_FluxNet/     # One subdirectory per group name
 python -m cli analyze               # Interactive model loading and evaluation
 ```
 
@@ -303,7 +303,7 @@ The `runs/` directory structure after training:
 
 ```
 runs/
-└── OSPREY_v0.10_DeepONet/
+└── SolarFlux_v0.3_FluxNet/
     └── MLP_n_64_l_5_SPlus_lr_3.42e-01.../
         ├── config.yaml              # Exact config used for this group
         ├── 58/

docs/02_config.md

@@ -189,20 +189,20 @@ Configs live under `configs/` and follow this convention:
 configs/<CONTRIBUTION>_v<VERSION>/<MODEL>_{run,opt,best}.yaml
 ```
 
-- `<CONTRIBUTION>` — the experiment name or project shorthand (e.g., `OSPREY`, `Neural_Hamilton`)
+- `<CONTRIBUTION>` — the experiment name or project shorthand (e.g., `SolarFlux`, `WavePredict`)
 - `<VERSION>` — integer version, incremented when the search space or architecture changes significantly
 - `<MODEL>` — the model or configuration variant being tested
 
 Examples:
 
 ```
-configs/OSPREY_v1/MLP_run.yaml       # base run config for HPO
-configs/OSPREY_v1/MLP_opt.yaml       # HPO search space
-configs/OSPREY_v1/MLP_best.yaml      # best config after HPO
+configs/SolarFlux_v1/MLP_run.yaml       # base run config for HPO
+configs/SolarFlux_v1/MLP_opt.yaml       # HPO search space
+configs/SolarFlux_v1/MLP_best.yaml      # best config after HPO
 
-configs/Neural_Hamilton_v2/HNN_run.yaml
-configs/Neural_Hamilton_v2/HNN_opt.yaml
-configs/Neural_Hamilton_v2/HNN_best.yaml
+configs/WavePredict_v2/HNN_run.yaml
+configs/WavePredict_v2/HNN_opt.yaml
+configs/WavePredict_v2/HNN_best.yaml
 ```
 
 This convention makes it immediately clear which phase of the workflow each file belongs to, and version numbers let you track search space evolution without losing old configs.