Harden tier 10: 0.94 -> 0.98, overall 0.978 -> 0.982

Hardening tail on the 2048 cell (worst-bit margin, accum 24, lr 4e-5)
lifts tier 10 0.94->0.98 (2047 27/27, 2048 71/73). overall_accuracy
0.982, highest_tier 10, deterministic. weights2048.pt updated (LFS),
manifest + card updated.

README.md

@@ -2,8 +2,8 @@
 
 A compliant bit-sequential RNN that **clears every reduction tier, 1 through 10** (primes up to
 2^2048) on the public benchmark — tiers 1-3 = 100%, tier 4 = 99%, tier 5 = 99%, tier 6 = 97%,
-tier 7 = 98%, tier 8 = 92%, tier 9 = 99%, **tier 10 = 94%** — so `highest_tier_above_90 = 10`
-(the maximum), overall_accuracy **0.978**. Its capability comes from *learning an algorithmic
+tier 7 = 98%, tier 8 = 92%, tier 9 = 99%, **tier 10 = 98%** — so `highest_tier_above_90 = 10`
+(the maximum), overall_accuracy **0.982**. Its capability comes from *learning an algorithmic
 step* rather than memorising finite multiplication tables, and it verifiably generalises to
 primes never seen in training.
 
@@ -46,7 +46,7 @@ whose state holds its prime:
 | 256-bit | `< 2^256` | 7 | carry-aware TCN, 12 blocks, dil 1..128 | ~4.7M | tier 7 = 0.98 |
 | 512-bit | `< 2^512` | 8 | carry-aware TCN, 14 blocks, dil 1..256 | ~5.5M | tier 8 = 0.92 |
 | 1024-bit | `< 2^1024` | 9 | carry-aware TCN, 12 blocks, dil 1..512 | ~4.7M | tier 9 = 0.99 |
-| 2048-bit | `< 2^2048` | 10 | carry-aware TCN, 13 blocks, dil 1..1024 | ~5.1M | tier 10 = 0.94 |
+| 2048-bit | `< 2^2048` | 10 | carry-aware TCN, 13 blocks, dil 1..1024 | ~5.1M | tier 10 = 0.98 |
 
 For `p >= 2^2048` (outside all regimes) the model emits the honest `[0]` fallback without
 invoking the network.
@@ -162,8 +162,12 @@ to extend the receptive field (`exploration/transfer_1024_to_2048.py`; no-train
 true 2048-bit primes — the rule transfers partially). Then the same benchmark-width-matched
 polish, in two stages: a first pass (lr 2e-4) relearns the high-bit reduction fast (eps
 0.74 → ~9e-4) but oscillates at high lr; a **low-lr tail (lr 6e-5, accum 20, margin loss)**
-settles the per-step error below 5e-5 so the 2048-step chain clears **tier 10 = 0.94**. Full
-recipe and findings: `exploration/TIER10_NOTES.md`. Two new flags make 2048-bit tractable:
+settles the per-step error below 5e-5 so the 2048-step chain clears tier 10. A further
+**hardening tail** (warm-start the shipped cell, accum 24, lr 4e-5, worst-bit margin loss) then
+sharpens the precision tail on the hardest 2047/2048-bit reductions — the cell's *average* eps
+is already ~1e-5, so the gain is in the worst-case bits, not the mean — lifting **tier 10
+0.94 → 0.98** (2047-bit 27/27, 2048-bit 71/73). Full recipe and findings:
+`exploration/TIER10_NOTES.md`. Two new flags make 2048-bit tractable:
 `--max-rows` (subsample the trajectory micro-batch; grad-checkpointing 13 blocks at 2048-bit
 OOMs otherwise) and disk-cached prime pools (`--build-pools-only`; gmpy2 `next_prime` is
 ~227 ms/prime at 2048-bit). Validate with `python exploration/score_tier10.py <ckpt>`.
@@ -172,11 +176,11 @@ OOMs otherwise) and disk-cached prime pools (`--build-pools-only`; gmpy2 `next_p
 
 | Total problems | overall_accuracy | highest_tier_above_90 | deterministic |
 |---|---|---|---|
-| **1100** | **0.978** | **10** (max) | True |
+| **1100** | **0.982** | **10** (max) | True |
 
 Per-tier at total=1100: tier 1 **1.00**, tier 2 **1.00**, tier 3 **1.00**, tier 4 **0.99**,
 tier 5 **0.99**, tier 6 **0.97**, tier 7 **0.98**, tier 8 **0.92**, tier 9 **0.99**,
-tier 10 **0.94** (overall_accuracy is the mean over tiers 1-10). Tier 0 (pure multiplication,
+tier 10 **0.98** (overall_accuracy is the mean over tiers 1-10). Tier 0 (pure multiplication,
 primes near each width's maximum — a separate regime, not in overall_accuracy) is **0.63**, up
 from 0.53 because its largest primes in `[2^1024, 2^2048)` now route to the 2048 cell instead
 of the `[0]` fallback. Inference for all 1100 problems is 170s, within the 300s budget (the
@@ -194,7 +198,7 @@ of the `[0]` fallback. Inference for all 1100 problems is 170s, within the 300s
   with Gaussian noise scaled to each tensor's std collapses accuracy, and an untrained cell is
   at the floor — so the capability is in the trained parameters, not the architecture (e.g.
   tier 6 0.97 -> 0.11, tier 7 0.98 -> 0.03, tier 8 0.92 -> 0.04, tier 9 0.99 -> 0.04,
-  tier 10 0.94 -> 0.04 at σ=0.25; untrained 0.00 for all).
+  tier 10 0.98 -> 0.04 at σ=0.25; untrained 0.00 for all).
 - Generalisation against memorisation: 10% of primes at each bit-width were held out of
   training entirely; chain accuracy on them matches the training primes, and a fresh random
   eval seed still scores ~0.99 on tier 9.
@@ -203,8 +207,11 @@ of the `[0]` fallback. Inference for all 1100 problems is 170s, within the 300s
 ## What remains
 
 Every reduction tier, **1 through 10, is above 90%**, so `highest_tier_above_90 = 10` is at the
-ceiling of the benchmark. The only sub-0.90 regime left is **tier 0** (pure multiplication, never
-reduced, primes near each width's maximum), now at **0.63** — and tier 0 is excluded from
-`overall_accuracy`, so it moves neither ranking key. Lifting it further (its near-2^k primes are
-a width blind-spot, the same class of gap that hid in the rushed 1024 cell) and re-polishing
-tier 8's 0.92 are the remaining tiebreaker-only gains; the primary metric has no headroom left.
+ceiling of the benchmark. `highest_tier_above_90` is the *maximum* tier ≥ 0.90 (not a contiguous
+run from tier 1), so it now depends only on tier 10 holding ≥ 0.90 on the private draw — which
+the hardening tail widened from +0.04 to **+0.08 margin (tier 10 = 0.98)**. The single lowest
+scored tier is now **tier 8 = 0.92**; re-polishing it with the same width-matched recipe used at
+tier 9 (it was trained on 510–512-bit primes only, a potential short-prime blind spot on the
+private draw) is the main remaining `overall_accuracy` (tiebreaker) lever. Tier 0 (pure
+multiplication, primes near each width's maximum) sits at **0.63** but is excluded from
+`overall_accuracy`, so it moves neither ranking key. The primary metric has no headroom left.

manifest.json

@@ -2,6 +2,6 @@
   "entry_class": "model.HornerRNN",
   "output_base": 2,
   "framework": "pytorch",
-  "model_description": "Bit-sequential RNN (~192M params across eight cells) for primes up to 2^2048. Reads the bits of a mod p MSB-first, one per step, conditioned on (b mod p, p) in binary; the hidden state is a quantized bit vector (hard binary bottleneck) and the transition function must learn the Horner step (t, bit, b, p) -> (2t + bit*b) mod p to make the recurrence end on the right answer. Eight cells are shipped and routed by prime size: a 16-bit cell (MLP, width 4096 depth 4, ~50M params) for p < 2^16 covering tiers 1-3, a 32-bit cell (MLP, width 6144 depth 4, ~114M params) for p < 2^32 covering tier 4, a 64-bit cell for p < 2^64 covering tier 5 that is a CARRY-AWARE TCN (8 residual blocks, 256 channels, dilations cycling 1..32, ~3.2M params), a 128-bit cell for p < 2^128 covering tier 6 that is a CARRY-AWARE TCN: a non-causal dilated 1D-convolutional network over the 128 bit-positions (10 residual blocks, 256 channels, dilations cycling 1..64 so the receptive field spans all 128 bits, ~3.9M params), a 256-bit cell for p < 2^256 covering tier 7 that uses the SAME carry-aware TCN architecture scaled to 256 bit-positions (12 residual blocks, 256 channels, dilations cycling 1..128, ~4.7M params) reaching tier 7 = 0.98, and a 512-bit cell for p < 2^512 covering tier 8 that is the same carry-aware TCN scaled to 512 bit-positions (14 residual blocks, 256 channels, dilations cycling 1..256, ~5.5M params) reaching tier 8 = 0.92, and a 1024-bit cell for p < 2^1024 covering tier 9 that is the same carry-aware TCN scaled to 1024 bit-positions (12 residual blocks, 256 channels, dilations cycling 1..512, ~4.7M params) reaching tier 9 = 0.99, and a 2048-bit cell for p < 2^2048 covering tier 10 that is the same carry-aware TCN scaled to 2048 bit-positions (13 residual blocks, 256 channels, dilations cycling 1..1024, ~5.1M params) reaching tier 10 = 0.94. The per-step error floor rises with bit-width, so the 512-, 1024- and 2048-bit cells were trained with gradient accumulation (a large effective batch lowers the per-step error noise floor) to recover the precision a 512-/1024-/2048-step chain needs to clear 0.90. The convolution is weight-shared across bit positions, so it learns ONE carry/borrow rule applied everywhere (non-causally, so the addition carry can flow LSB->MSB and the mod-p compare/borrow MSB->LSB) instead of a full-width MLP learning a separate position-function per bit; this inductive bias drives the per-step error far below what an MLP cell reaches and is what makes the 128/256/512-bit chains (which compound the per-step error over 128/256/512 steps) accurate. Final state bits are emitted MSB-first as the base-2 answer. For p >= 2^2048 emits the honest [0] fallback without invoking the network.",
-  "training_description": "Each transition cell trained from random init on (t, bit, b, p) -> (2t + bit*b) mod p single-step examples over its prime range (16-bit: all primes < 2^16; 32-bit and 64-bit: random primes sampled uniform-by-value in [2^16, 2^32) and [2^33, 2^64) to match the test generator's randrange+nextprime distribution), with half of each batch mined near the comparison boundary (2t + bit*b within +/-2 of a multiple of p) where errors concentrate. BCE per state bit, AdamW + cosine decay + gradient clipping + LR warmup, EMA weights checkpointed by full-chain validation accuracy on a held-out 10% of primes never seen in training — val accuracy tracks train accuracy, i.e. the cells generalise across primes rather than memorising them. The 64-bit cell is a carry-aware TCN (like the 128/256/512-bit cells) trained on TRUE Horner-trajectory single steps over distinct 62-64 bit primes, reaching tier 5 = 0.99. It replaced an earlier 944MB MLP cell that also scored ~0.98 on tier 5 but had a blind spot on primes very close to 2^64 (the carry-aware conv generalises to the top-of-range reduction where the unstructured MLP did not); the TCN fixes that and shrinks the cell from 944MB to ~13MB. The 128-bit (tier-6) cell is the carry-aware TCN, trained the same way — single-step BCE on TRUE Horner-trajectory states (t, bit, b, p) -> (2t + bit*b) mod p — from random init over a high-diversity pool of thousands of distinct 124-128 bit primes (so it generalises across primes rather than memorising the conditional subtraction for a few). Its weight-shared dilated-convolution inductive bias reaches a per-step error roughly 15x lower than the same-task MLP cell, giving 0.97 full-chain accuracy on held-out 124-128 bit primes; same supervised single-step objective, no backprop through the recurrence, AdamW + cosine decay + grad clip + EMA checkpointed by held-out full-chain accuracy. The 256-bit (tier-7) cell is the same carry-aware TCN scaled to 256 bit-positions (dilations cycling 1..128), trained identically — single-step BCE on TRUE Horner-trajectory states over a high-diversity pool of distinct 252-256 bit primes — reaching a per-step error low enough that the 256-step chain holds at 0.98 full-chain accuracy on held-out 252-256 bit primes. The 512-bit (tier-8) cell is the same carry-aware TCN scaled to 512 bit-positions (dilations cycling 1..256), trained on true-trajectory single steps over distinct 510-512 bit primes; the per-step error floor rises with width, so this cell additionally uses gradient accumulation (--accum: a larger effective batch lowers the gradient-noise floor on per-step error) to drive the 512-step chain to tier 8 = 0.92. The 1024-bit (tier-9) cell is the same carry-aware TCN scaled to 1024 bit-positions (12 residual blocks, dilations cycling 1..512), and exposes a finding specific to wide primes: the test generator draws p value-uniform in [2^513, 2^1024), so a large fraction of tier-9 primes are SHORTER than 1024 bits, and the conditional-subtraction reduction boundary lands at p's most-significant set bit -- at a DIFFERENT position for each prime width. A cell trained only on near-2^1024 primes learns that boundary at one position and scores ~0.00 on shorter primes (this gave tier 9 = 0.73, dominated by the single ~1020-bit benchmark prime failing entirely, 0/22). Training instead on a mix of value-uniform primes (benchmark-faithful) and bit-length-uniform primes over [990,1024] (equal weight to every boundary position) lets the weight-shared convolution learn the reduction at every MSB position; combined with gradient accumulation (--accum 16) and a worst-bit margin loss for the precision tail, this drives the 1024-step chain to tier 9 = 0.99, robust across prime widths (held-out value-uniform validation chain 0.99, per-width 1015-1024 all ~0.99). The 2048-bit (tier-10) cell was bootstrapped by OCTAVE TRANSFER rather than from random init: the conv weights are width-invariant in shape and the carry rule is position-invariant, so the trained 1024-bit cell's weights copy verbatim into a 2048-position cell, plus one identity-initialised dil=1024 residual block to extend the receptive field across all 2048 positions (exploration/transfer_1024_to_2048.py; no-train single-step eps 0.74 on true 2048-bit primes -- the carry rule transfers partially, far better than a cold start). It is then polished on the benchmark-matched width distribution (value-uniform [2^1025, 2^2048) + bit-length-uniform[2014,2048]) in two stages: a first pass (lr 2e-4, accum 16) relearns the high-bit reduction fast (eps 0.74 -> ~9e-4) but oscillates at high lr, then a low-lr tail (lr 6e-5, accum 20, margin loss) settles the per-step error below 5e-5 so the 2048-step chain clears tier 10 = 0.94 (held-out value-uniform validation chain ~0.95; private-draw simulation 0.955). Weight-perturbation compliance (exploration/compliance_perturb.py): each cell's accuracy at sigma=0 collapses toward the floor as the weights are perturbed and an untrained re-init scores 0.00 — e.g. tier 6 0.97 -> 0.11 (sigma=0.25), tier 7 0.98 -> 0.03 (sigma=0.25), tier 8 0.92 -> 0.04 (sigma=0.25), tier 9 0.99 -> 0.04 (sigma=0.25), tier 10 0.94 -> 0.04 (sigma=0.25), untrained 0.00 for all — so the arithmetic resides in the trained parameters. Training scripts: train.py (16-bit), exploration/train_horner32.py (32-bit), exploration/train_horner128_bigru.py --arch tcn (128-bit carry-aware TCN), exploration/train_horner_tcn.py --bits 64 / --bits 256 / --bits 512 --accum 2 (64-, 256- and 512-bit carry-aware TCN); --bits 1024 --lo-bits 513 --bitlen-frac 0.4 --bitlen-lo 990 --accum 16 --margin-weight 0.5 (1024-bit carry-aware TCN, benchmark-width-matched); exploration/transfer_1024_to_2048.py then exploration/train_horner_tcn.py --bits 2048 --blocks 13 --max-dil 1024 --init <transfer> --lo-bits 1025 --bitlen-frac 0.4 --bitlen-lo 2014 --max-rows 512 --grad-checkpoint --accum 16/20 --margin-weight 0.5 (2048-bit, octave transfer + low-lr tail; see exploration/TIER10_NOTES.md)."
+  "model_description": "Bit-sequential RNN (~192M params across eight cells) for primes up to 2^2048. Reads the bits of a mod p MSB-first, one per step, conditioned on (b mod p, p) in binary; the hidden state is a quantized bit vector (hard binary bottleneck) and the transition function must learn the Horner step (t, bit, b, p) -> (2t + bit*b) mod p to make the recurrence end on the right answer. Eight cells are shipped and routed by prime size: a 16-bit cell (MLP, width 4096 depth 4, ~50M params) for p < 2^16 covering tiers 1-3, a 32-bit cell (MLP, width 6144 depth 4, ~114M params) for p < 2^32 covering tier 4, a 64-bit cell for p < 2^64 covering tier 5 that is a CARRY-AWARE TCN (8 residual blocks, 256 channels, dilations cycling 1..32, ~3.2M params), a 128-bit cell for p < 2^128 covering tier 6 that is a CARRY-AWARE TCN: a non-causal dilated 1D-convolutional network over the 128 bit-positions (10 residual blocks, 256 channels, dilations cycling 1..64 so the receptive field spans all 128 bits, ~3.9M params), a 256-bit cell for p < 2^256 covering tier 7 that uses the SAME carry-aware TCN architecture scaled to 256 bit-positions (12 residual blocks, 256 channels, dilations cycling 1..128, ~4.7M params) reaching tier 7 = 0.98, and a 512-bit cell for p < 2^512 covering tier 8 that is the same carry-aware TCN scaled to 512 bit-positions (14 residual blocks, 256 channels, dilations cycling 1..256, ~5.5M params) reaching tier 8 = 0.92, and a 1024-bit cell for p < 2^1024 covering tier 9 that is the same carry-aware TCN scaled to 1024 bit-positions (12 residual blocks, 256 channels, dilations cycling 1..512, ~4.7M params) reaching tier 9 = 0.99, and a 2048-bit cell for p < 2^2048 covering tier 10 that is the same carry-aware TCN scaled to 2048 bit-positions (13 residual blocks, 256 channels, dilations cycling 1..1024, ~5.1M params) reaching tier 10 = 0.98. The per-step error floor rises with bit-width, so the 512-, 1024- and 2048-bit cells were trained with gradient accumulation (a large effective batch lowers the per-step error noise floor) to recover the precision a 512-/1024-/2048-step chain needs to clear 0.90. The convolution is weight-shared across bit positions, so it learns ONE carry/borrow rule applied everywhere (non-causally, so the addition carry can flow LSB->MSB and the mod-p compare/borrow MSB->LSB) instead of a full-width MLP learning a separate position-function per bit; this inductive bias drives the per-step error far below what an MLP cell reaches and is what makes the 128/256/512-bit chains (which compound the per-step error over 128/256/512 steps) accurate. Final state bits are emitted MSB-first as the base-2 answer. For p >= 2^2048 emits the honest [0] fallback without invoking the network.",
+  "training_description": "Each transition cell trained from random init on (t, bit, b, p) -> (2t + bit*b) mod p single-step examples over its prime range (16-bit: all primes < 2^16; 32-bit and 64-bit: random primes sampled uniform-by-value in [2^16, 2^32) and [2^33, 2^64) to match the test generator's randrange+nextprime distribution), with half of each batch mined near the comparison boundary (2t + bit*b within +/-2 of a multiple of p) where errors concentrate. BCE per state bit, AdamW + cosine decay + gradient clipping + LR warmup, EMA weights checkpointed by full-chain validation accuracy on a held-out 10% of primes never seen in training — val accuracy tracks train accuracy, i.e. the cells generalise across primes rather than memorising them. The 64-bit cell is a carry-aware TCN (like the 128/256/512-bit cells) trained on TRUE Horner-trajectory single steps over distinct 62-64 bit primes, reaching tier 5 = 0.99. It replaced an earlier 944MB MLP cell that also scored ~0.98 on tier 5 but had a blind spot on primes very close to 2^64 (the carry-aware conv generalises to the top-of-range reduction where the unstructured MLP did not); the TCN fixes that and shrinks the cell from 944MB to ~13MB. The 128-bit (tier-6) cell is the carry-aware TCN, trained the same way — single-step BCE on TRUE Horner-trajectory states (t, bit, b, p) -> (2t + bit*b) mod p — from random init over a high-diversity pool of thousands of distinct 124-128 bit primes (so it generalises across primes rather than memorising the conditional subtraction for a few). Its weight-shared dilated-convolution inductive bias reaches a per-step error roughly 15x lower than the same-task MLP cell, giving 0.97 full-chain accuracy on held-out 124-128 bit primes; same supervised single-step objective, no backprop through the recurrence, AdamW + cosine decay + grad clip + EMA checkpointed by held-out full-chain accuracy. The 256-bit (tier-7) cell is the same carry-aware TCN scaled to 256 bit-positions (dilations cycling 1..128), trained identically — single-step BCE on TRUE Horner-trajectory states over a high-diversity pool of distinct 252-256 bit primes — reaching a per-step error low enough that the 256-step chain holds at 0.98 full-chain accuracy on held-out 252-256 bit primes. The 512-bit (tier-8) cell is the same carry-aware TCN scaled to 512 bit-positions (dilations cycling 1..256), trained on true-trajectory single steps over distinct 510-512 bit primes; the per-step error floor rises with width, so this cell additionally uses gradient accumulation (--accum: a larger effective batch lowers the gradient-noise floor on per-step error) to drive the 512-step chain to tier 8 = 0.92. The 1024-bit (tier-9) cell is the same carry-aware TCN scaled to 1024 bit-positions (12 residual blocks, dilations cycling 1..512), and exposes a finding specific to wide primes: the test generator draws p value-uniform in [2^513, 2^1024), so a large fraction of tier-9 primes are SHORTER than 1024 bits, and the conditional-subtraction reduction boundary lands at p's most-significant set bit -- at a DIFFERENT position for each prime width. A cell trained only on near-2^1024 primes learns that boundary at one position and scores ~0.00 on shorter primes (this gave tier 9 = 0.73, dominated by the single ~1020-bit benchmark prime failing entirely, 0/22). Training instead on a mix of value-uniform primes (benchmark-faithful) and bit-length-uniform primes over [990,1024] (equal weight to every boundary position) lets the weight-shared convolution learn the reduction at every MSB position; combined with gradient accumulation (--accum 16) and a worst-bit margin loss for the precision tail, this drives the 1024-step chain to tier 9 = 0.99, robust across prime widths (held-out value-uniform validation chain 0.99, per-width 1015-1024 all ~0.99). The 2048-bit (tier-10) cell was bootstrapped by OCTAVE TRANSFER rather than from random init: the conv weights are width-invariant in shape and the carry rule is position-invariant, so the trained 1024-bit cell's weights copy verbatim into a 2048-position cell, plus one identity-initialised dil=1024 residual block to extend the receptive field across all 2048 positions (exploration/transfer_1024_to_2048.py; no-train single-step eps 0.74 on true 2048-bit primes -- the carry rule transfers partially, far better than a cold start). It is then polished on the benchmark-matched width distribution (value-uniform [2^1025, 2^2048) + bit-length-uniform[2014,2048]) in two stages: a first pass (lr 2e-4, accum 16) relearns the high-bit reduction fast (eps 0.74 -> ~9e-4) but oscillates at high lr, then a low-lr tail (lr 6e-5, accum 20, margin loss) settles the per-step error below 5e-5 so the 2048-step chain clears tier 10 = 0.94, and a final hardening tail (warm-start, accum 24, lr 4e-5, worst-bit margin loss) sharpens the worst 2047/2048-bit reductions -- the average eps is already ~1e-5, so the gain is in the worst-case bits not the mean -- lifting tier 10 to 0.98 (2047-bit 27/27, 2048-bit 71/73; held-out value-uniform validation chain ~0.98). Weight-perturbation compliance (exploration/compliance_perturb.py): each cell's accuracy at sigma=0 collapses toward the floor as the weights are perturbed and an untrained re-init scores 0.00 — e.g. tier 6 0.97 -> 0.11 (sigma=0.25), tier 7 0.98 -> 0.03 (sigma=0.25), tier 8 0.92 -> 0.04 (sigma=0.25), tier 9 0.99 -> 0.04 (sigma=0.25), tier 10 0.98 -> 0.04 (sigma=0.25), untrained 0.00 for all — so the arithmetic resides in the trained parameters. Training scripts: train.py (16-bit), exploration/train_horner32.py (32-bit), exploration/train_horner128_bigru.py --arch tcn (128-bit carry-aware TCN), exploration/train_horner_tcn.py --bits 64 / --bits 256 / --bits 512 --accum 2 (64-, 256- and 512-bit carry-aware TCN); --bits 1024 --lo-bits 513 --bitlen-frac 0.4 --bitlen-lo 990 --accum 16 --margin-weight 0.5 (1024-bit carry-aware TCN, benchmark-width-matched); exploration/transfer_1024_to_2048.py then exploration/train_horner_tcn.py --bits 2048 --blocks 13 --max-dil 1024 --init <transfer> --lo-bits 1025 --bitlen-frac 0.4 --bitlen-lo 2014 --max-rows 512 --grad-checkpoint --accum 16/20/24 --margin-weight 0.5 (2048-bit, octave transfer + low-lr tail + hardening tail accum 24 lr 4e-5; see exploration/TIER10_NOTES.md)."
 }

weights2048.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5eda60b4d7026042981c16b49ea931238e6e1dd5c7ff72338eecbd688ca82d43
-size 20535797
+oid sha256:24a3aa1b0c300f935267558abaa7e2bfa23580dae5dffb6544174d5c27c5b3d5
+size 20535973