diff --git a/persistent-memory.d/2026-09-14-fv-seat-reorg-and-orca-blocker.md b/persistent-memory.d/2026-09-14-fv-seat-reorg-and-orca-blocker.md index 7a82b59..62f4555 100644 --- a/persistent-memory.d/2026-09-14-fv-seat-reorg-and-orca-blocker.md +++ b/persistent-memory.d/2026-09-14-fv-seat-reorg-and-orca-blocker.md @@ -101,7 +101,14 @@ cost decode speed, which was the standing risk of giving up FP4 tensor-core comp - Quality A/B orca vs dealignai (the actual reason for the swap — the W4A4 long-context degradation axis). Needs a controlled harness + noise floor. - Deep-prefill probe at 262K against THIS checkpoint. Startup is not a depth test. -- Rollback is two `.env` keys; `.env.bak-preorca-20260914-023408` on the host. +- ⚠⚠ **NO LOCAL ROLLBACK.** dealignai weights DELETED 2026-09-14 on operator instruction + (125 GiB reclaimed). `.env.bak-preorca-20260914-023408` still names the old paths but they + no longer exist — it is a record, not a revert. Reverting = 126 GiB re-download. + The pristine 170 GiB `qwen38-flash-next-orcarouter-nvfp4` IS retained (redo the conversion + from it; do not delete it without a reason). +- Disk: ~75 GiB of non-PLE shards are duplicated between the pristine and converted orca dirs + (the convert's hardlinks hit EXDEV inside the container). Now that both sit directly on + /tank, relinking them would reclaim it — not urgent at 3.8T free. - Disk: the convert **copied** ~75 GiB of unchanged shards because hardlinks hit `EXDEV` (separate bind mounts of the same fs). Harmless; reclaimable by relinking if /tank tightens. diff --git a/persistent-memory.md b/persistent-memory.md index b56d8ca..e123746 100644 --- a/persistent-memory.md +++ b/persistent-memory.md @@ -162,8 +162,11 @@ KV 344,155 tokens @ 262,144 ctx, 1.31x concurrency. ⚠ The reorg note's dealign (~121 tok/s) came from a DIFFERENT harness — do not treat this as a measured win over dealignai; a controlled A/B needs dealignai back on this same harness plus a noise floor. -⚠ Not yet done: quality A/B vs dealignai, and a deep-prefill probe at 262K on this -checkpoint. `.env` rollback is two keys (`.env.bak-preorca-20260914-023408`). +⚠⚠ **NO LOCAL ROLLBACK — dealignai weights DELETED 2026-09-14** on operator instruction +(125 GiB reclaimed, /tank 57% used). Reverting = re-downloading 126 GiB. The `.env` backup +names paths that no longer exist. ⚠ Not yet done, and now unbacked: quality A/B vs dealignai +(its reference arm is gone) and a 262K deep-prefill probe. The pristine 170 GiB orcarouter +download is RETAINED — it is what makes the PLE conversion redoable without re-fetching. → `persistent-memory.d/2026-09-14-fv-seat-reorg-and-orca-blocker.md` ### fv-ml1 seat topology now diff --git a/stacks/flash-next-seat/.env.example b/stacks/flash-next-seat/.env.example index 2196f84..38e9b0c 100644 --- a/stacks/flash-next-seat/.env.example +++ b/stacks/flash-next-seat/.env.example @@ -27,8 +27,11 @@ FN_CONTAINER_NAME=vllm-flash-next # in-house (2026-09-14) so vLLM's from_quant_config branch 1 selects the FP8 PLE method. # ⚠ compressed-tensors, NOT modelopt_fp4 — this checkpoint is weight-only on BOTH axes # (W8 float attn / W4 float experts, input_activations: null), which is why it was chosen -# over the dealignai W4A4 build. Rollback: the dealignai pair below, both still on disk. -# FN_MODEL=/tank/aimodels/qwen38-flash-next-abliterated-nvfp4 FN_QUANT=modelopt_fp4 +# over the dealignai W4A4 build. +# ⚠⚠ NO LOCAL ROLLBACK: the dealignai weights were DELETED 2026-09-14 (operator instruction). +# Reverting means re-downloading 126 GiB of dealignai/Qwen3.8-Flash-Next-ABLITERATED-NVFP4. +# The pristine 170 GiB orcarouter source is kept at qwen38-flash-next-orcarouter-nvfp4 so the +# PLE bf16→FP8 conversion can be redone without re-fetching. FN_MODEL=/tank/aimodels/qwen38-flash-next-orcarouter-nvfp4-plefp8 FN_QUANT=compressed-tensors FN_SERVED_NAME=qwen3.8-flash-next-uncensored diff --git a/stacks/flash-next-seat/README.md b/stacks/flash-next-seat/README.md index 8d6d0ff..3e35127 100644 --- a/stacks/flash-next-seat/README.md +++ b/stacks/flash-next-seat/README.md @@ -99,9 +99,16 @@ zero saturation — and the same FP8-PLE treatment dealignai already shipped, so regression against the seat it replaced. `weight_scale` is written BF16 [1] to match the published format. MTP head (31 tensors) and the vision tower carry through untouched. -⚠ Still unmeasured: a controlled quality A/B against dealignai — which is the entire reason -for the swap — and a deep-prefill probe at 262K on this checkpoint. Rollback is two `.env` -keys; the dealignai checkpoint is still on disk. +⚠⚠ **THERE IS NO LOCAL ROLLBACK.** The dealignai checkpoint was deleted on operator +instruction 2026-09-14 (125 GiB reclaimed). Reverting this seat now means **re-downloading +126 GiB** from `dealignai/Qwen3.8-Flash-Next-ABLITERATED-NVFP4`, not flipping two `.env` +keys. The `.env` backup (`.env.bak-preorca-20260914-023408`) still names the old paths, but +those paths no longer exist — treat it as a record of the old settings, not a working revert. + +⚠ Still unmeasured, and now unbacked: a controlled quality A/B against dealignai — which was +the entire reason for the swap, and whose reference arm is gone — and a deep-prefill probe at +262K on this checkpoint. The 170 GiB pristine `qwen38-flash-next-orcarouter-nvfp4` download is +retained; it is what lets the PLE conversion be redone without re-fetching. Rejected alternatives, for the record: `nvidia/…-NVFP4` is the cleanest ModelOpt build but is not abliterated; `lovedheart/…-Pruned-RTXPRO-6000` prunes to 448 of 512 experts;