CoSiNE

New5-30 minutesCodeNeed >5GB

Evolve antibody sequences along realistic affinity-maturation trajectories

Readiness

Readiness: Need >5GB
Smoke: not_tested
Artifacts: not_applicable
Production: need_gt5gb_cuda_flashattn_esm_jax_checkpoint_pending
Storage: 5-20GB
Strategy: shared CUDA PyTorch/flash-attn/triton/ESM/JAX SIF + checkpoint overlay
Next: Build/reuse CUDA PyTorch/flash-attn/triton/ESM/JAX shared SIF or overlay, stage source/submodule and HF checkpoint, run unguided sampling through Slurm, then wire runner.

No obvious large model/database dependency identified; validate upstream docs before install. 2026-07-05 SOP batch recheck: backend inventory found no staged CoSiNE source/submodule/checkpoint/SIF/artifacts/runner. Upstream requires Python >=3.10 with heavy Torch/flash-attn/triton/fair-esm/JAX/transformers/iglm/pyhmmer stack plus HF checkpoint before sampling. Reclassified from Smoke Pending to Need >5GB.