Serve Pythia-Covariates v7 with private pinned weights and covariate/LLOQ inputs (part 2)

Serve Pythia-Covariates v7 with private pinned weights and covariate/LLOQ inputs

Fix assay-aware VPCs; retain existing public model releases

Support interactive synthetic cohorts, MLP resampling and configurable binned VPCs (part 2)

Support interactive synthetic cohorts, MLP resampling and configurable binned VPCs

Match dependency lock to Hugging Face Gradio build extras

Serve canonical v1 v6 v7 synthetic cohorts alongside Pythia inference

Align VPC intervals to model query times

Evaluate synchronized VPCs at exact times

Allow 100 Pythia generation draws

Guard custom cohort inference requests

Default inference and VPC seeds to 43

Use Pharmpy VPCs and seed 42

Serve selectable Pythia and Pythia-Dose models

Fix public solver to Heun 8

Cap public inference workload

Run public inference on CPU Basic

Document hosted PFF API contract

Declare ZeroGPU inference endpoint

Use native Gradio inference endpoints

Deploy private-model PFF-PK inference API

initial commit