Datasets:
aefinityAIINC
/
cis2-conformance

Dataset card Data Studio Files
xet
 Community
1
text
stringlengths
1
274
# dotted against x = [1,1,1] MUST give exactly 0.0_f32 (the `1.0` term is
# lost to rounding against the 1e8 partial sum) -- a conforming
# implementation MUST reproduce this exact cancellation, not just "close".
# Row 1 and row 2 use small exactly-representable values (no order
# sensitivity) as a plain-arithmetic sanity check alongside the
# order-sensitive row 0.
w_bits=0x4CBEBC20,0x3F800000,0xCCBEBC20,0x3F800000,0x40000000,0x40400000,0x3F000000,0xBFC00000,0x40800000
x_bits=0x3F800000,0x3F800000,0x3F800000
# CIS-2 conformance vector: rmsnorm_v1
# op=rmsnorm, spec ref: docs/CIS2_SPEC_v0.2.md §8
#
# Fields are `key=value`. Vector fields (x_bits, gamma_bits) are
# comma-separated lists of `0x`-prefixed 32-bit hex, each the raw IEEE-754
# binary32 bit pattern of one element, in index order (index 0 first).
# eps_bits is the single f32 bit pattern for CIS-2 EPS_F32 (§2.3).
op=rmsnorm
spec_ref=docs/CIS2_SPEC_v0.2.md#8-rmsnorm-normative
n=8
eps_bits=0x3727C5AC
x_bits=0x3F800000,0xC0000000,0x40400000,0x3F000000,0xBF000000,0x40800000,0xBFC00000,0x40100000
gamma_bits=0x3F000000,0x3FC00000,0x3F800000,0x40000000,0x3E800000,0x3F800000,0x40400000,0x3F400000
# CIS-2 conformance vector: rope_v1
# op=rope_table, spec ref: docs/CIS2_SPEC_v0.2.md §7 (RoPE)
#
# Fields are `key=value`. head_dim and positions are plain decimal
# integers (no ambiguity: head_dim is a small config constant, and
# `positions` values are sequence positions, cast to f32 *exactly* per
# §7.3 -- both are exact integers well within f32's exact-integer range,
# so there is no float-parsing-rounding ambiguity to route around here,
# unlike genuinely fractional float inputs). rope_theta_bits is the single
# f32 bit pattern of rope_theta (comma-separated 0x-prefixed 32-bit hex
# convention, per tests/conformance/README.md, applied to a length-1 list).
op=rope_table
spec_ref=docs/CIS2_SPEC_v0.2.md#7-rope-normative
head_dim=8
rope_theta_bits=0x461C4000
positions=0,3
# CIS-2 conformance vector: softmax_v1
# op=softmax, spec ref: docs/CIS2_SPEC_v0.2.md §9.3 (Softmax,
# `softmax_seq`, in place: strict-sequential max-scan, pinned-exp
# max-subtract, §5.3 sequential-sum denominator, elementwise divide).
# This vector exercises softmax_seq directly as its own primitive
# (unlike attention_block_v1, which exercises it as one step of a larger
# score/softmax/V-mix pipeline where the input scores come from a dot
# product) -- so the input here is a hand-sized `scores` vector, not
# derived from any dot product.
#
# Fields are `key=value`. `n` is a plain decimal integer. `scores_bits`
# is a comma-separated list of `0x`-prefixed 32-bit hex IEEE-754 binary32
# bit patterns, one per input logit, in index order (index 0 first),
# per tests/conformance/README.md's general convention.
#
# n=6. Values: 1.5, -0.5, 0.5, 1.5, -3.0, 0.0 -- deliberately includes an
# EXACT tie for the maximum at indices 0 and 3 (both 1.5), to exercise
# §9.3's "strict >, so the FIRST occurrence of the max wins ties"
# max-scan rule explicitly (the tie is invisible in the final output
# bits here, since both tied values are bit-identical, but the vector
# still documents that softmax_v1's max-scan must use strict `>` and
# must not, e.g., use the LAST occurrence or an unstable comparison).
# Also includes a large negative outlier (-3.0) to exercise the
# post-max-sub `exp_pinned` route at a value further from zero, and 0.0
# as a mid-range plain value.
op=softmax
spec_ref=docs/CIS2_SPEC_v0.2.md#93-softmax-softmax_seq-in-place-over-scores0pos
n=6
scores_bits=0x3FC00000,0xBF000000,0x3F000000,0x3FC00000,0xC0400000,0x00000000