Skip to content

Pull requests: paiml/aprender

New pull request New
32 Open 1,921 Closed
32 Open 1,921 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Reviews
Filter by reviews
No reviews Review required Approved review Changes requested
Assignee
Filter by who’s assigned
Assigned to nobody Loading
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

fix(tests): test_qa_016 asserts the honest score contract instead of the /100 wording #2394 removed
#3053 opened Sep 8, 2026 by noahgift Contributor Loading…
1
1
perf(B5): pr-review-receipt moves off pull_request to ad-hoc dispatch — 39.4% of the merge path for ONE verdict in 30 days bse-001 BSE-001 build-system enhancement / unblocker
#3052 opened Sep 8, 2026 by noahgift Contributor Loading…
2
feat(F-1): the model you named is the model that answers — apr chat never substitutes the demo model, and a sharded SafeTensors index loads F-1 PP-066 (0.66) DAG row inst:A PP-066 instance claim (I14): inst:A pp-066 PP-066 (0.66) DAG row
#3050 opened Sep 8, 2026 by noahgift Contributor Loading…
2
1
fix(B4): the roadmap remedy had zero callers, and the hook that would run it exits 2 on its own line 4 bse-001 BSE-001 build-system enhancement / unblocker
#3048 opened Sep 8, 2026 by noahgift Contributor Loading…
1
2
perf(ci): pr-review-receipt is dispatch-only — 96 h of fleet runner-time for a job that gates nothing (PMAT-1078)
#3046 opened Sep 8, 2026 by noahgift Contributor Loading…
7
feat(R-0b): backend resolution reads the registry — a forced accelerator never downgrades to cpu inst:A PP-066 instance claim (I14): inst:A pp-066 PP-066 (0.66) DAG row R-0b PP-066 row R-0b
#3041 opened Sep 7, 2026 by noahgift Contributor Loading…
1
feat(parity): L0-1b — apr parity --per-op names the first diverging op (1.5B: post_ffn_residual layer 26); the CPU Q8_K reference is the inaccurate side inst:A PP-066 instance claim (I14): inst:A L0-1b PP-066 DAG row L0-1b (#2971) pp-066 PP-066 (0.66) DAG row
#3032 opened Sep 6, 2026 by noahgift Contributor Loading…
1
fix(C0-2): pin sovereign-ci.yml by sha; CB-2100 reachability finding filed (#3029) — PMAT-976 C0-2 PP-066 DAG row C0-2 inst:C PP-066 instance claim (I14): inst:C pp-066 PP-066 (0.66) DAG row
#3031 opened Sep 6, 2026 by noahgift Contributor Loading…
2
roadmap: dedupe twelve duplicated ids; PMAT-1060 README CPU-only claim vs wgpu default, for 0.65.3 (Refs PMAT-1060)
#3027 opened Sep 6, 2026 by noahgift Contributor Loading…
2
feat(parity): L0-1a — every supported model computes the same function on GPU as on CPU or the GPU refuses it (manifest, C14, REG-15 admission, effective-config parity) inst:A PP-066 instance claim (I14): inst:A L0-1a PP-066 DAG row L0-1a (#2971) pp-066 PP-066 (0.66) DAG row
#3026 opened Sep 6, 2026 by noahgift Contributor Loading…
2
docs(PP-066): session 2026-09-06 — SPEC-2.0 (v2.0, 18-row scope, claims ratchet, executable criteria), DAG rows/edges/amendments, decision receipts, roadmap mints, rescope quorum record inst:A PP-066 instance claim (I14): inst:A pp-066 PP-066 (0.66) DAG row
#3025 opened Sep 6, 2026 by noahgift Contributor Loading…
2
guard(G-10b): check_pmat_pinned.sh — every analyser reference resolves through scripts/pmat_bin.sh, counted and shrink-only against a measured baseline (PMAT-1063)
#3021 opened Sep 6, 2026 by noahgift Contributor Loading…
1
1
audit: post-publish dogfood aprender 0.65.2 — NO-GO: README says CPU-only, plain install compiles the wgpu backend (Refs PMAT-246)
#3010 opened Sep 6, 2026 by noahgift Contributor Loading…
1
fix(I-26): spec_conformance.sh — a §12 expiry is the Expires marker, never the first date in the cell; rows past expiry are RED (the §4 andon); derived_expiries.json regenerated (row 1 → 2026-09-19) (PMAT-974, #2889)
#3009 opened Sep 6, 2026 by noahgift Contributor Loading…
1
fix(I-25): --workload is bound to the prompt corpus — receipt.workload only with corpus_sha256 of a prompts file whose _meta.corpus equals the label; --workload W1 --profile short is refused with one line (PMAT-973, #2888, #2756)
#3008 opened Sep 6, 2026 by noahgift Contributor Loading…
1
fix(C0-4): perf_gate.sh — an Arm A that measured nothing on a c=1-only receipt says so (REPORT ArmA scaling: c=1 only, no scaling measured) and the gate never turns an arm's silence into VERDICT PASS; selftest rows both polarities (PMAT-978, #2893, #2830)
#3007 opened Sep 6, 2026 by noahgift Contributor Loading…
3
fix(I-24): parity_block.py — a zero or empty COMPARATOR band is a named refusal, never a ZeroDivisionError traceback; case table both polarities wired into guard-runner-labels (PMAT-972, #2887, #2735)
#3006 opened Sep 6, 2026 by noahgift Contributor Loading…
1
fix(T-2): apr finetune --max-seq-len is honoured on the wgpu instruct pipeline or refused with the documented code — never clamped to 512; one effective_max_seq_len() for every path, printed (PMAT-1009, #2924)
#3005 opened Sep 6, 2026 by noahgift Contributor Loading…
1
feat(registry): R-0a — BackendRegistry probes and enumerates cpu/cuda/wgpu at startup and apr devices prints every kind as a line (PMAT-989, #2904) inst:B PP-066 instance claim (I14): inst:B pp-066 PP-066 (0.66) DAG row R-0 PP-066 DAG row R-0 (BackendRegistry)
#3004 opened Sep 6, 2026 by noahgift Contributor Loading…
2
fix(R-3): the cuBLAS-backward training banner is an event (launch counter), and --gpu-backend cuda on a cpu-only build is FeatureDisabled exit 9 — never a silent CPU run (PMAT-991, #2906) inst:C PP-066 instance claim (I14): inst:C pp-066 PP-066 (0.66) DAG row R-3 PP-066 DAG row R-3
#3001 opened Sep 6, 2026 by noahgift Contributor Loading…
3
docs(audits): PR triage into 0.66 — 29 dispositions pp-066 PP-066 (0.66) DAG row
#2998 opened Sep 5, 2026 by noahgift Contributor • Draft
1
fix(cgp): a binary on PATH is not a device on the bus — Coverage Nightly was red on an environment fact pp-066/merge triage 0.66 — MERGE disposition
#2848 opened Sep 2, 2026 by noahgift Contributor Loading…
1
fix(beat): the ollama decode floor was calibrated on sm_89 and is now asserted on sm_121 (#2835) pp-066/carry-0.67 triage 0.66 — carried to 0.67 row
#2838 opened Sep 1, 2026 by noahgift Contributor • Draft
2
fix(security): wasmtime 43 -> 47.0.4 -- the real fix for RUSTSEC-2026-0269, and it needed ZERO source changes pp-066/merge triage 0.66 — MERGE disposition
#2807 opened Aug 31, 2026 by noahgift Contributor Loading…
1
3
measure(perf): PERF-062 band ladder at HEAD — ordered streams remove the corruption (#2789) pp-066/merge triage 0.66 — MERGE disposition
#2802 opened Aug 31, 2026 by noahgift Contributor Loading…
1
ProTip! Filter pull requests by the default branch with base:main.