flowchart LR
operator["Operator"]
lm["LM Studio
OpenAI-compatible API
qwen/qwen3.5-9b normalization"]
base["Base Model
Local Llama3-OpenBioLLM-8B"]
subgraph repository["Medical PHI Model Repository"]
catalogs["Catalogs and Patterns
Text files
12 categories, variants, subjects,
positive/negative templates"]
api["Generator API
ASP.NET Core / .NET 10
Validates, plans, normalizes,
and persists ordered batches"]
sql[("Training Database
SQL Server
Categories, keywords,
generation runs, training items")]
jsonl[("Training Data
JSONL
Source, train, validation,
development")]
pipeline["Training Pipeline
Python 3.12 / PyTorch / TRL / PEFT
Validate, split, SFT, train,
evaluate, merge"]
artifacts[("Model Artifacts
LoRA adapters, metrics,
merged BF16 model, GGUF files")]
catalogs -->|Reads| api
api -->|Persists and checkpoints| sql
sql -->|PowerShell export| jsonl
jsonl -->|Reads and rewrites| pipeline
pipeline -->|Writes| artifacts
end
operator -->|Runs| api
operator -->|Operates and reviews| pipeline
api -->|Normalizes text| lm
base -->|Weights and tokenizer| pipeline
classDef person fill:#fff2cc,stroke:#d6b656,color:#000;
classDef container fill:#438dd5,stroke:#0b4884,color:#fff;
classDef external fill:#999,stroke:#666,color:#fff;
class operator person;
class catalogs,api,sql,jsonl,pipeline,artifacts container;
class lm,base external;