Commit History

  1. gemma4_e2b_int4_rawmetal raw-Metal int4 pack on the native QAT grid — tokenizer, generation config and stop ids inside; manifest carries sizes and sha256s

    Verified signature
    c818d13

  1. E2B raw-Metal int4 pack — our exact int4 recipe, bi dropped, softcapped head — plus the KV-sharing assistant pack; manifest carries sizes and sha256s

    Verified signature
    476e397

  1. E2B raw-Metal int4 pack — our exact int4 recipe, bi dropped, softcapped head — plus the KV-sharing assistant pack; manifest carries sizes and sha256s

    Verified signature
    33c005d

  1. e2b cap bundle: README with provenance, KV arithmetic, digests

    Verified signature
    26c0964
  2. E2B pf64 at cap 131,072 — the checkpoint's own max_position_embeddings

    Verified signature
    1ec5c73
  3. model card: standard structure, measured facts, verified license declarations

    Verified signature
    0d7b6cb

  1. README: ring3 per-layer-type SDPA bundles

    Verified signature
    15c2afa
  2. Add ring3 gemma4_e2b_qat_decode_int4lin_tbl_pf64_ring_c2048_gqafull

    Verified signature
    0d2065c
  3. Add ring3 gemma4_e2b_qat_decode_int4lin_tbl_pf64_ring_c4096_gqafull

    Verified signature
    f62c233
  4. Add ring3-smoke gemma4_e2b_qat_decode_int4lin_tbl_pf64_ring_c4096_gqafull_l5

    Verified signature
    f1485ac
  5. README: ring2 capacity-ladder bundles

    Verified signature
    023e7a0
  6. Add ring2 gemma4_e2b_qat_decode_int4lin_tbl_pf64_ring_c8192

    Verified signature
    9030ad4
  7. Add ring2 gemma4_e2b_qat_decode_int4lin_tbl_pf64_ring_c2048

    Verified signature
    37f9a33
  8. Add ring2-smoke gemma4_e2b_qat_decode_int4lin_tbl_pf64_ring_c2048_l5

    Verified signature
    2fedcea
  9. README: ring bundles

    Verified signature
    cb38966
  10. Add ring gemma4_e2b_qat_decode_int4lin_tbl_pf64_ring_c16384

    Verified signature
    9f5ec84
  11. README: ring-smoke bundles

    Verified signature
    beecbbf
  12. Add ring-smoke gemma4_e2b_qat_decode_int4lin_tbl_pf64_ring_c16384_l5

    Verified signature
    6d4e0b3
  13. docs: stable/ measured — memory fix confirmed, capacity-16384 decode cost makes it non-deployable as shipped

    Verified signature
    54f007f
  14. prune: remove superseded/non-functional artifacts

    Verified signature
    5005b4c
  15. prune: remove superseded/non-functional artifacts

    Verified signature
    73bef81
  16. docs: stable/ folders in Status, oracle verification line, known-issue points at available fix

    Verified signature
    98d2913
  17. README: stable bundles

    Verified signature
    50c2ae0
  18. Add stable gemma4_e2b_qat_decode_int4lin_tbl_pf64_stable_c16384

    Verified signature
    4591f92
  19. README: stable-smoke bundles

    Verified signature
    25b51a7
  20. Add stable-smoke gemma4_e2b_qat_decode_int4lin_tbl_pf64_stable_c16384_l5

    Verified signature
    bd55b68
  21. docs: correct known-issue mechanism (shape-signature specialization cache, guided affected equally at depth)

    Verified signature
    1d86775

  1. docs: refresh measured performance (local retest 2026-08-18)

    Verified signature
    81c2abf
  2. docs: descriptive model card

    Ben Graville
    Verified signature
    60bc67d
  3. README: w4a8 bundles

    Ben Graville
    Verified signature
    03102f0
  4. Add w4a8/gemma4_e2b_qat_decode_int4lin_a8_tbl_pf64

    Ben Graville
    Verified signature
    1563877
  5. add missing chat_template.jinja to gpu-pipelined/gemma4_e2b_qat_decode_int4lin_tbl_pf64/tokenizer (additive fix, bundle otherwise untouched)

    Ben Graville
    Verified signature
    b1b5044
  6. add missing chat_template.jinja to gpu-pipelined/gemma4_e2b_qat_decode_int4lin_tbl/tokenizer (additive fix, bundle otherwise untouched)

    Ben Graville
    Verified signature
    2a59909
  7. Add PLE gather_raw table sidecar

    Ben Graville
    Verified signature
    9cde3bf
  8. README: ctx16384 bundles

    Ben Graville
    Verified signature
    8a0128c
  9. Add ctx16384 gemma4_e2b_qat_decode_int4lin_tbl_pf64

    Ben Graville
    Verified signature
    38150d1
  10. README: ctx8192 bundles

    Ben Graville
    Verified signature
    8a3f3be
  11. Add ctx8192 gemma4_e2b_qat_decode_int4lin_tbl_pf64

    Ben Graville
    Verified signature
    c2524f8
  12. Add gemma4_e2b_qat_decode_int4lin_tbl_pf64

    Ben Graville
    Verified signature
    a791d48
  13. Add gemma4_e2b_qat_decode_int4lin_tbl

    Ben Graville
    Verified signature
    9ad2f9f
  14. Upload README.md with huggingface_hub

    Ben Graville
    Verified signature
    de7f65f
  15. initial commit

    Ben Graville
    Verified signature
    1a19324