Skip to content

isrlab/selective-layer-audit

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

1 Commit
 
 
 
 
 
 
 
 
 
 

Repository files navigation

selective-layer-audit

An exact instrument for measuring how a trained selective state-space model (Mamba-1, Falcon-Mamba, Mamba-2) uses its per-channel state, and the input-driven mode migration it reveals.

Because a selective SSM layer has a diagonal, fixed state matrix, each channel's output is an exact sum of per-mode contributions. Accumulating the outer products of those contributions over a window gives a per-(layer, channel, window) Gram tensor, from which the exact output error of pruning any subset of modes follows in closed form, offline, at any budget. This repository is the code behind that instrument: the audit, the mechanism counterfactuals, the end-to-end pruning experiments, and the baselines.

Layout

audit/
  audit_run.py          exact per-mode decomposition (modal_scan) + validation gate
  gram_utils.py         Gram packing and nested tail sums
  prune_forward.py      deployed pruned forward path + instrument cross-check + e2e
  cross_sweep.py        swept deployment cross-check (independent reference side)
  e2e_baselines.py      static / modal-HSV / LAST baselines (empirical Gramians)
  bootstrap_ci.py       paired and block bootstrap confidence intervals
  mamba2_check.py       depth-localized migration in Mamba-2 (SSD)
  mamba2_prune.py       Mamba-2 end-to-end static vs scheduled pruning
  mamba2_mechanism.py   Mamba-2 frozen-signal mechanism
  normmatch_freezeB.py  energy-matched freeze-B control
  churn_gap.py          churn-vs-gap relation
  churn_gap_theory.py   energy-weighted churn vs migration gap
  timescales.py         pole-timescale structure of migration
  ltv_gramians.py       time-varying Hankel singular values sigma_i(t)
  probe.py, scheduler.py, learned_scheduler.py, half_transfer.py
  switched_signal.py    switched-system control with ground-truth regimes
  downstream.py         downstream tasks under mode pruning (lm-eval-harness)
  tests/                synthetic-mixer unit tests (no downloads)
  ghost_comparison/     native comparison against the released GHOST code

Install

pip install -r requirements.txt

The audit runs the reference (fp32, non-fused) SSM path so the exactness gate is meaningful; on machines with the fused Mamba/causal-conv1d kernels installed, the scripts disable the fast path where required.

Reproduce

The pipeline consumes public released checkpoints (Mamba-1 130M-2.8B, Falcon-Mamba-7B, Mamba-2 130M/780M). A typical run:

python audit/audit_run.py --model 130m           # per-mode Gram audit + gate
python audit/prune_forward.py --cross-check       # instrument == deployment
python audit/cross_sweep.py  --model 130m         # swept validation
python audit/e2e_baselines.py --model 130m --v2   # baselines at equal budget

Each script prints its validation gate first and writes summary JSON; the tables and figures in the paper are produced from those summaries. The ghost_comparison/ directory documents how to run the released GHOST code head-to-head, including the environment pins it requires.

Tests

pytest audit/tests

Citation

If you use this code, please cite the accompanying paper (see the repository description for the current reference).

About

An exact instrument for state usage in selective state-space models (Mamba), and the input-driven migration it reveals.

Resources

Stars

Watchers

Forks

Releases

Packages

Contributors

Languages