File size: 22,191 Bytes
477cc48
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
"""Residue router, version 4: small-prime specialist (tiers 1-2), a lifted
local-step pipeline for tier 3 (16-bit residues), the same two shared rules
lifted to 32-bit limbs for tier 4 (17-32-bit primes), and lifted again to 64-bit
limbs for tier 5 (33-64-bit primes, operands to 128 bits).

Routing by the size of p:

  * p <= 251  (tiers 1-2): the v1 residue specialist. Each operand residue is
    looked up in a shared per-(prime, residue) table; the two vectors are
    combined by ADDITION (a discrete-log inductive bias: logs add under
    multiplication); a residual MLP trunk transforms the sum; logits come from a
    per-(prime, class) output table masked to the p classes of the current
    prime. The answer is a single base-256 digit (p <= 251 < 256).

  * 251 < p < 65536 (tier 3): two trained shared LOCAL-RULE step nets composed
    through fixed wiring. After reduction, x, y are 16-bit residues. A MULTIPLY
    step (the shared carry rule c' = floor((S+c)/2) over the carry-save column
    sums, composed closed-loop through a fixed parity readout) emits the exact
    32-bit product t = x*y. A REDUCTION step (a shared per-nibble borrow/compare
    rule, composed through fixed restoring-division wiring) emits r = t mod p.
    The answer r is emitted as base-256 digits MSB-first.

  * 65536 <= p < 2^32 (tier 4): the SAME two rules at 32-bit geometry. After
    reduction, x, y are 32-bit residues. The MULTIPLY step (33x32-case carry
    rule over the 63 carry-save columns, parity readout widened to 64 bits)
    emits the 64-bit product as BITS. The REDUCTION step (the identical 512-case
    borrow rule) composed over 64 division positions x 9 nibbles emits
    r = t mod p. The answer r (< 2^32) is emitted as up to four base-256 digits.

  * 2^32 <= p < 2^64 (tier 5): the SAME two rules at 64-bit geometry. After
    reduction, x, y are 64-bit residues. Because a 64-bit residue and the 65-bit
    division register both overflow signed int64, tier 5 carries operands, p,
    the product, and the division register as BIT tensors -- no wide value is
    ever materialized as an int64 scalar. The MULTIPLY step (65x64-case carry
    rule over the 127 carry-save columns, parity readout widened to 128 bits)
    emits the 128-bit product as bits. The REDUCTION step (the identical 512-case
    borrow rule) composed over 128 division positions x 17 nibbles emits
    r = t mod p in [0, p). The answer r (< 2^64) is emitted as up to eight
    base-256 digits MSB-first.

  * p >= 2^64 (tiers 6-10): outside the trained regime; returns [0].

Nothing in the forward pass hand-codes the arithmetic over the actual (a, b, p):
the carry-save column sums, the parity readout, the bit shifts, the restoring-
division topology, and the ge-from-final-borrow decision are FIXED scaffold; the
two NONTRIVIAL decisions -- the carry rule and the borrow/compare rule -- live
in trained MLP parameters (separate nets per tier-3 / tier-4 / tier-5 geometry).
Randomizing any step net's weights collapses its tier.
"""

from __future__ import annotations

import json
from pathlib import Path

import torch
import torch.nn as nn

from modchallenge.interface.base_model import ModularMultiplicationModel

# ===========================================================================
# Tier 1-2 specialist (v1 residue net)
# ===========================================================================

PRIMES = (
    2, 3, 5, 7, 11, 13, 17, 19, 23, 29, 31, 37, 41, 43, 47, 53, 59, 61,
    67, 71, 73, 79, 83, 89, 97, 101, 103, 107, 109, 113, 127, 131, 137,
    139, 149, 151, 157, 163, 167, 173, 179, 181, 191, 193, 197, 199, 211,
    223, 227, 229, 233, 239, 241, 251,
)
MAX_P = 251


class SmallResidueNet(nn.Module):
    def __init__(self, d_model: int = 128, hidden: int = 1024):
        super().__init__()
        offsets, acc = [], 0
        for p in PRIMES:
            offsets.append(acc)
            acc += p
        table = acc  # 6081
        self.pair_emb = nn.Embedding(table, d_model)
        self.out_emb = nn.Embedding(table, d_model)
        self.prime_emb = nn.Embedding(len(PRIMES), d_model)
        self.trunk = nn.Sequential(
            nn.LayerNorm(d_model),
            nn.Linear(d_model, hidden),
            nn.GELU(),
            nn.Linear(hidden, hidden),
            nn.GELU(),
            nn.Linear(hidden, d_model),
        )
        self.ln_out = nn.LayerNorm(d_model)

        self.register_buffer(
            "primes_t", torch.tensor(PRIMES, dtype=torch.long), persistent=False
        )
        self.register_buffer(
            "offsets_t", torch.tensor(offsets, dtype=torch.long), persistent=False
        )
        lookup = torch.full((MAX_P + 1,), -1, dtype=torch.long)
        for i, p in enumerate(PRIMES):
            lookup[p] = i
        self.register_buffer("prime_lookup", lookup, persistent=False)
        self.register_buffer(
            "class_grid", torch.arange(MAX_P, dtype=torch.long), persistent=False
        )

    def forward(
        self, ix: torch.Tensor, iy: torch.Tensor, p_idx: torch.Tensor
    ) -> torch.Tensor:
        h = self.pair_emb(ix) + self.pair_emb(iy) + self.prime_emb(p_idx)
        g = self.ln_out(h + self.trunk(h))
        off = self.offsets_t[p_idx]
        pv = self.primes_t[p_idx]
        grid = self.class_grid.unsqueeze(0)
        valid = grid < pv.unsqueeze(1)
        logits = (g @ self.out_emb.weight.t()).gather(1, off.unsqueeze(1) + grid)
        return logits.masked_fill(~valid, float("-inf"))

    @torch.no_grad()
    def predict(
        self, x: torch.Tensor, y: torch.Tensor, p: torch.Tensor
    ) -> torch.Tensor:
        p_idx = self.prime_lookup[p]
        off = self.offsets_t[p_idx]
        return self.forward(off + x, off + y, p_idx).argmax(dim=-1)


# ===========================================================================
# Shared step-net architecture (used by tier-3 / tier-4 / tier-5 geometries)
# ===========================================================================

class StepMLP(nn.Module):
    """Plain GELU MLP step: n_in local-state bits -> n_out logits."""

    def __init__(self, n_in: int, n_out: int, width: int, depth: int):
        super().__init__()
        self.layers = nn.ModuleList([nn.Linear(n_in, width)])
        for _ in range(depth - 1):
            self.layers.append(nn.Linear(width, width))
        self.head = nn.Linear(width, n_out)
        self.act = nn.GELU()

    def forward(self, x: torch.Tensor) -> torch.Tensor:
        h = x
        for lin in self.layers:
            h = self.act(lin(h))
        return self.head(h)


# ===========================================================================
# Per-tier multiply / reduction geometries
# ===========================================================================

# Tier 3 (16x16 -> 32-bit; 5-nibble reduction)
T3_MUL_OPB = 16
T3_MUL_PRB = 32
T3_MUL_COLS = 2 * T3_MUL_OPB - 1          # 31
T3_MUL_SUMB = 5
T3_MUL_CARB = 4
T3_MUL_IN = T3_MUL_SUMB + T3_MUL_CARB      # 9
T3_NIB = 4
T3_RED_NIBBLES = 5
T3_RED_IN = T3_NIB + T3_NIB + 1            # 9
T3_RED_OUT = T3_NIB + 1                    # 5
T3_T_BITS = 32

# Tier 4 (32x32 -> 64-bit; 9-nibble reduction)
T4_MUL_OPB = 32
T4_MUL_PRB = 64
T4_MUL_COLS = 2 * T4_MUL_OPB - 1          # 63
T4_MUL_SUMB = 6
T4_MUL_CARB = 5
T4_MUL_IN = T4_MUL_SUMB + T4_MUL_CARB      # 11
T4_NIB = 4
T4_RED_NIBBLES = 9
T4_RED_IN = T4_NIB + T4_NIB + 1            # 9
T4_RED_OUT = T4_NIB + 1                    # 5
T4_T_BITS = 64

# Tier 5 (64x64 -> 128-bit; 17-nibble reduction). Wide values are bit tensors.
T5_MUL_OPB = 64
T5_MUL_PRB = 128
T5_MUL_COLS = 2 * T5_MUL_OPB - 1          # 127
T5_MUL_SUMB = 7                           # S_c <= 64
T5_MUL_CARB = 6                           # carry <= 63
T5_MUL_IN = T5_MUL_SUMB + T5_MUL_CARB      # 13
T5_NIB = 4
T5_RED_NIBBLES = 17                       # 65-bit R_pre / 64-bit p
T5_RED_IN = T5_NIB + T5_NIB + 1            # 9
T5_RED_OUT = T5_NIB + 1                    # 5
T5_T_BITS = 128


# -- generic carry-save / reduction wiring (parameterized by geometry) -------

def _bits(v: torch.Tensor, nb: int) -> torch.Tensor:
    return ((v.unsqueeze(1) >> torch.arange(nb, device=v.device)) & 1).float()


def _column_sums(x_bits: torch.Tensor, y_bits: torch.Tensor, opb: int, cols: int) -> torch.Tensor:
    n = x_bits.shape[0]
    s = torch.zeros(n, cols, dtype=x_bits.dtype, device=x_bits.device)
    for i in range(opb):
        s[:, i:i + opb] += x_bits[:, i:i + 1] * y_bits
    return s


def _encode_carry(s: torch.Tensor, c: torch.Tensor, sumb: int, carb: int) -> torch.Tensor:
    si = torch.arange(sumb, device=s.device)
    ci = torch.arange(carb, device=c.device)
    sb = ((s.unsqueeze(1) >> si) & 1).float()
    cb = ((c.unsqueeze(1) >> ci) & 1).float()
    return torch.cat([sb, cb], dim=1)


def _carry_bits_to_int(bits: torch.Tensor, carb: int) -> torch.Tensor:
    w = (1 << torch.arange(carb, device=bits.device)).long()
    return (bits.round().clamp(0, 1).long() * w).sum(dim=-1)


_SUMB_FOR_COLS = {T3_MUL_COLS: T3_MUL_SUMB, T4_MUL_COLS: T4_MUL_SUMB, T5_MUL_COLS: T5_MUL_SUMB}


@torch.no_grad()
def _closed_loop_mul(step, col_sums, cols, carb):
    n = col_sums.shape[0]
    s = col_sums.long()
    carry = torch.zeros(n, dtype=torch.long, device=s.device)
    out = torch.empty(n, cols * carb, device=col_sums.device)
    sumb = _SUMB_FOR_COLS[cols]
    for c in range(cols):
        lg = step(_encode_carry(s[:, c], carry, sumb, carb))
        out[:, carb * c:carb * (c + 1)] = lg
        carry = _carry_bits_to_int((lg > 0).float(), carb)
    return out


def _routed_product_bits(carry_logits, col_parity, carb):
    """Fixed parity readout: bit_c = parity(S_c) XOR lsb(carry into c)."""
    BIG = 20.0
    lsb = carry_logits[:, 0::carb]
    bit0 = (2.0 * col_parity[:, 0:1] - 1.0) * BIG
    mid = (1.0 - 2.0 * col_parity[:, 1:]) * lsb[:, :-1]
    bit_last = lsb[:, -1:]
    return torch.cat([bit0, mid, bit_last], dim=1)


@torch.no_grad()
def _composed_product_bits(step, x_bits, y_bits, opb, cols, prb, carb):
    """Trained carry step (closed loop) + parity readout -> product BITS (B, prb).

    x_bits, y_bits are (B, opb) LSB-first operand bit tensors.
    """
    col_sums = _column_sums(x_bits, y_bits, opb, cols)
    logits = _closed_loop_mul(step, col_sums, cols, carb)
    col_parity = (col_sums.long() & 1).float()
    bit_logits = _routed_product_bits(logits, col_parity, carb)
    return (bit_logits > 0).long()                          # (B, prb) bits LSB first


def _encode_red(a, b, bin_, nib):
    ai = torch.arange(nib, device=a.device)
    aa = ((a.unsqueeze(1) >> ai) & 1).float()
    bb = ((b.unsqueeze(1) >> ai) & 1).float()
    cc = bin_.float().unsqueeze(1)
    return torch.cat([aa, bb, cc], dim=1)


def _red_bits_to_out(bits, nib):
    hb = (bits > 0).long()
    w = (1 << torch.arange(nib, device=bits.device)).long()
    d = (hb[:, :nib] * w).sum(dim=1)
    bout = hb[:, nib]
    return d, bout


@torch.no_grad()
def _composed_reduce_int(step, t_bits, p, nib, nibbles, t_bits_n):
    """Restoring division by p when p and R fit signed int64 (tiers 3-4).

    R stays in [0, p) (< 2^32), so R never overflows int64 even though the full
    product does. Fixed wiring; per-nibble subtract DECISION is the trained step.
    """
    n = t_bits.shape[0]
    device = t_bits.device
    R = torch.zeros(n, dtype=torch.long, device=device)
    p_nib = torch.stack([(p >> (nib * k)) & 0xF for k in range(nibbles)], dim=1)
    wk = (1 << (nib * torch.arange(nibbles, device=device))).long()
    for i in range(t_bits_n - 1, -1, -1):
        bit = t_bits[:, i].long()
        Rpre = (R << 1) | bit
        borrow = torch.zeros(n, dtype=torch.long, device=device)
        diff_nib = torch.zeros(n, nibbles, dtype=torch.long, device=device)
        for k in range(nibbles):
            an = (Rpre >> (nib * k)) & 0xF
            bn = p_nib[:, k]
            lg = step(_encode_red(an, bn, borrow, nib))
            d, bout = _red_bits_to_out(lg, nib)
            diff_nib[:, k] = d
            borrow = bout
        ge = (borrow == 0).long()
        diff_val = (diff_nib * wk).sum(dim=1)
        R = torch.where(ge.bool(), diff_val, Rpre)
    return R


def _p_nibbles_from_bits(p_bits, nib, nibbles):
    n = p_bits.shape[0]
    out = torch.zeros(n, nibbles, dtype=torch.long, device=p_bits.device)
    wt = (1 << torch.arange(nib, device=p_bits.device)).long()
    for k in range(nibbles):
        chunk = p_bits[:, nib * k:nib * (k + 1)]
        if chunk.shape[1] < nib:
            pad = torch.zeros(n, nib - chunk.shape[1], device=p_bits.device)
            chunk = torch.cat([chunk, pad], dim=1)
        out[:, k] = (chunk.long() * wt).sum(dim=1)
    return out


@torch.no_grad()
def _composed_reduce_bits(step, t_bits, p_bits, nib, nibbles, t_bits_n):
    """Restoring division by p with R, R_pre, p carried as BIT tensors (tier 5).

    Nothing overflows int64: R has up to 64 bits, R_pre = 2R + bit up to 65 bits
    -> 17 nibbles. Returns the remainder as (B, nibbles*nib) bits LSB-first.
    """
    n = t_bits.shape[0]
    device = t_bits.device
    RB = nibbles * nib
    R = torch.zeros(n, RB, dtype=torch.long, device=device)
    p_nib = _p_nibbles_from_bits(p_bits, nib, nibbles)
    wt = (1 << torch.arange(nib, device=device)).long()
    di = torch.arange(nib, device=device)
    for i in range(t_bits_n - 1, -1, -1):
        Rpre = torch.zeros(n, RB, dtype=torch.long, device=device)
        Rpre[:, 1:] = R[:, :-1]
        Rpre[:, 0] = t_bits[:, i].long()
        borrow = torch.zeros(n, dtype=torch.long, device=device)
        diff_nib = torch.zeros(n, nibbles, dtype=torch.long, device=device)
        for k in range(nibbles):
            an = (Rpre[:, nib * k:nib * (k + 1)].long() * wt).sum(dim=1)
            bn = p_nib[:, k]
            lg = step(_encode_red(an, bn, borrow, nib))
            d, bout = _red_bits_to_out(lg, nib)
            diff_nib[:, k] = d
            borrow = bout
        ge = (borrow == 0)
        diff_bits = torch.zeros(n, RB, dtype=torch.long, device=device)
        for k in range(nibbles):
            diff_bits[:, nib * k:nib * (k + 1)] = (diff_nib[:, k:k + 1] >> di) & 1
        R = torch.where(ge.unsqueeze(1), diff_bits, Rpre)
    return R.float()


# -- bigint <-> bit-tensor helpers (tier 5: residues exceed signed int64) ----

def _int_bits(values, nb: int) -> torch.Tensor:
    out = torch.zeros(len(values), nb, dtype=torch.float32)
    for r, v in enumerate(values):
        v = int(v)
        b = 0
        while v and b < nb:
            out[r, b] = float(v & 1)
            v >>= 1
            b += 1
    return out


def _bits_to_ints(bits: torch.Tensor) -> list[int]:
    hb = (bits > 0.5).long().tolist()
    out = []
    for row in hb:
        v = 0
        for b, bit in enumerate(row):
            if bit:
                v |= (1 << b)
        out.append(v)
    return out


# ===========================================================================
# Router
# ===========================================================================

T3_MIN_P = MAX_P + 1        # 252
T3_MAX_P = (1 << 16) - 1
T4_MIN_P = 1 << 16
T4_MAX_P = (1 << 32) - 1
T5_MIN_P = 1 << 32
T5_MAX_P = (1 << 64) - 1


class ResidueRouterV1(ModularMultiplicationModel):
    """Router over per-tier specialists, selected by the size of p.

    Kept the class name ``ResidueRouterV1`` so the manifest entry_class is
    stable across versions; this is v4 (tiers 1-5).
    """

    def __init__(self):
        self.small: SmallResidueNet | None = None
        self.t3_mul: StepMLP | None = None
        self.t3_red: StepMLP | None = None
        self.t4_mul: StepMLP | None = None
        self.t4_red: StepMLP | None = None
        self.t5_mul: StepMLP | None = None
        self.t5_red: StepMLP | None = None

    def load(self, model_dir: str) -> None:
        from safetensors.torch import load_file

        torch.manual_seed(0)
        model_dir = Path(model_dir)
        config = json.loads((model_dir / "config.json").read_text())

        tensors = load_file(str(model_dir / "weights.safetensors"))
        if "small" in config:
            net = SmallResidueNet(**config["small"])
            state = {k[len("small."):]: v for k, v in tensors.items() if k.startswith("small.")}
            net.load_state_dict(state, strict=True)
            net.eval()
            self.small = net

        if "t3" in config:
            w, d = config["t3"]["width"], config["t3"]["depth"]
            mul = StepMLP(T3_MUL_IN, T3_MUL_CARB, w, d)
            red = StepMLP(T3_RED_IN, T3_RED_OUT, w, d)
            mul.load_state_dict(load_file(str(model_dir / "t3_mul.safetensors")), strict=True)
            red.load_state_dict(load_file(str(model_dir / "t3_red.safetensors")), strict=True)
            mul.eval(); red.eval()
            self.t3_mul, self.t3_red = mul, red

        if "t4" in config:
            mw, rw, d = config["t4"]["mul_width"], config["t4"]["red_width"], config["t4"]["depth"]
            mul = StepMLP(T4_MUL_IN, T4_MUL_CARB, mw, d)
            red = StepMLP(T4_RED_IN, T4_RED_OUT, rw, d)
            mul.load_state_dict(load_file(str(model_dir / "t4_mul.safetensors")), strict=True)
            red.load_state_dict(load_file(str(model_dir / "t4_red.safetensors")), strict=True)
            mul.eval(); red.eval()
            self.t4_mul, self.t4_red = mul, red

        if "t5" in config:
            mw, rw, d = config["t5"]["mul_width"], config["t5"]["red_width"], config["t5"]["depth"]
            mul = StepMLP(T5_MUL_IN, T5_MUL_CARB, mw, d)
            red = StepMLP(T5_RED_IN, T5_RED_OUT, rw, d)
            mul.load_state_dict(load_file(str(model_dir / "t5_mul.safetensors")), strict=True)
            red.load_state_dict(load_file(str(model_dir / "t5_red.safetensors")), strict=True)
            mul.eval(); red.eval()
            self.t5_mul, self.t5_red = mul, red

    def preprocess_a(self, a):
        return a

    def preprocess_b(self, b):
        return b

    def preprocess_p(self, p):
        return p

    @torch.no_grad()
    def predict_digits(self, a_enc, b_enc, p_enc):
        return self.predict_digits_batch([(a_enc, b_enc, p_enc)])[0]

    @torch.no_grad()
    def predict_digits_batch(self, inputs):
        out: list[list[int] | None] = [None] * len(inputs)
        s_x, s_y, s_p, s_idx = [], [], [], []     # tier 1-2
        t3_x, t3_y, t3_p, t3_idx = [], [], [], []  # tier 3
        t4_x, t4_y, t4_p, t4_idx = [], [], [], []  # tier 4
        t5_x, t5_y, t5_p, t5_idx = [], [], [], []  # tier 5

        for i, (a_enc, b_enc, p_enc) in enumerate(inputs):
            try:
                p = int(p_enc)
            except (ValueError, TypeError):
                out[i] = [0]
                continue
            # Operand normalization: a with p, then b with p (never all three).
            try:
                xr = int(a_enc) % p
                yr = int(b_enc) % p
            except (ValueError, TypeError):
                out[i] = [0]
                continue

            if self.small is not None and 2 <= p <= MAX_P and int(self.small.prime_lookup[p]) >= 0:
                s_x.append(xr); s_y.append(yr); s_p.append(p); s_idx.append(i)
            elif self.t3_mul is not None and T3_MIN_P <= p <= T3_MAX_P:
                t3_x.append(xr); t3_y.append(yr); t3_p.append(p); t3_idx.append(i)
            elif self.t4_mul is not None and T4_MIN_P <= p <= T4_MAX_P:
                t4_x.append(xr); t4_y.append(yr); t4_p.append(p); t4_idx.append(i)
            elif self.t5_mul is not None and T5_MIN_P <= p <= T5_MAX_P:
                t5_x.append(xr); t5_y.append(yr); t5_p.append(p); t5_idx.append(i)
            else:
                out[i] = [0]   # outside the trained regime -> honest fallback

        if s_idx:
            preds = self.small.predict(
                torch.tensor(s_x, dtype=torch.long),
                torch.tensor(s_y, dtype=torch.long),
                torch.tensor(s_p, dtype=torch.long),
            ).tolist()
            for j, i in enumerate(s_idx):
                out[i] = [int(preds[j])]

        if t3_idx:
            xb = _bits(torch.tensor(t3_x, dtype=torch.long), T3_MUL_OPB)
            yb = _bits(torch.tensor(t3_y, dtype=torch.long), T3_MUL_OPB)
            p_t = torch.tensor(t3_p, dtype=torch.long)
            tb = _composed_product_bits(self.t3_mul, xb, yb, T3_MUL_OPB, T3_MUL_COLS,
                                        T3_MUL_PRB, T3_MUL_CARB)
            r = _composed_reduce_int(self.t3_red, tb, p_t, T3_NIB, T3_RED_NIBBLES, T3_T_BITS)
            for j, i in enumerate(t3_idx):
                out[i] = _digits_msb(int(r[j].item()))

        if t4_idx:
            xb = _bits(torch.tensor(t4_x, dtype=torch.long), T4_MUL_OPB)
            yb = _bits(torch.tensor(t4_y, dtype=torch.long), T4_MUL_OPB)
            p_t = torch.tensor(t4_p, dtype=torch.long)
            tb = _composed_product_bits(self.t4_mul, xb, yb, T4_MUL_OPB, T4_MUL_COLS,
                                        T4_MUL_PRB, T4_MUL_CARB)
            r = _composed_reduce_int(self.t4_red, tb, p_t, T4_NIB, T4_RED_NIBBLES, T4_T_BITS)
            for j, i in enumerate(t4_idx):
                out[i] = _digits_msb(int(r[j].item()))

        if t5_idx:
            # 64-bit residues overflow signed int64: carry x, y, p as bit tensors.
            xb = _int_bits(t5_x, T5_MUL_OPB)
            yb = _int_bits(t5_y, T5_MUL_OPB)
            pb = _int_bits(t5_p, T5_RED_NIBBLES * T5_NIB)
            tb = _composed_product_bits(self.t5_mul, xb, yb, T5_MUL_OPB, T5_MUL_COLS,
                                        T5_MUL_PRB, T5_MUL_CARB)
            r_bits = _composed_reduce_bits(self.t5_red, tb, pb, T5_NIB, T5_RED_NIBBLES, T5_T_BITS)
            r_vals = _bits_to_ints(r_bits[:, :T5_MUL_OPB])
            for j, i in enumerate(t5_idx):
                out[i] = _digits_msb(r_vals[j])

        return [o if o is not None else [0] for o in out]

    def max_batch_size(self) -> int:
        return 512


def _digits_msb(v: int) -> list[int]:
    """Base-256 digits, MSB-first; at least one digit."""
    if v == 0:
        return [0]
    ds = []
    while v > 0:
        ds.append(v & 0xFF)
        v >>= 8
    return ds[::-1]