fix(linear): add bias once, not once per SIMD lane

acc was seeded with @splat(bias) and reduced with .Add, so every
linear output carried 7 extra copies of its bias. conv2d was unaffected
(bias via memset), which is why the trunk matched zig-solver to 5e-6
while fc/head activations diverged — the 81% labels regression.

Also adds a scalar tail for in_n < 8 (pose_fc has in_n=4) and a
known-answer selftest (test_linear).

👾 Generated with [Letta Code](https://letta.com)

Co-Authored-By: Letta Code <noreply@letta.com>

build.zig

@@ -5,7 +5,7 @@ pub fn build(b: *std.Build) void {
     const optimize = b.standardOptimizeOption(.{});
 
     // The library module — what downstream consumers import.
-    const crucible_mod = b.addModule("crucible", .{
+    const cruc_mod = b.addModule("crucible", .{
         .root_source_file = b.path("src/crucible.zig"),
         .target = target,
         .optimize = optimize,
@@ -14,19 +14,18 @@ pub fn build(b: *std.Build) void {
     // Static lib artifact (optional install).
     const lib = b.addLibrary(.{
         .name = "crucible",
-        .root_module = crucible_mod,
+        .root_module = cruc_mod,
     });
     b.installArtifact(lib);
 
-    // Example executable: stripsolver (needs a tests/ png decoder? no —
-    // example is self-contained: it uses crucible only for state dict +
-    // layers; arm input comes from CSV).
+    // Example executable: stripsolver (self-contained: uses crucible only
+    // for state dict + layers; arm input comes from CSV).
     const example_mod = b.createModule(.{
         .root_source_file = b.path("examples/stripsolver.zig"),
         .target = target,
         .optimize = optimize,
         .imports = &.{
-            .{ .name = "crucible", .module = crucible_mod },
+            .{ .name = "crucible", .module = cruc_mod },
         },
     });
     const example = b.addExecutable(.{
@@ -41,8 +40,20 @@ pub fn build(b: *std.Build) void {
     const run_step = b.step("run", "Run the stripsolver example");
     run_step.dependOn(&run_example.step);
 
+    // linear selftest
+    const tl_mod = b.createModule(.{
+        .root_source_file = b.path("examples/test_linear.zig"),
+        .target = target,
+        .optimize = optimize,
+        .imports = &.{
+            .{ .name = "crucible", .module = cruc_mod },
+        },
+    });
+    const tl = b.addExecutable(.{ .name = "test_linear", .root_module = tl_mod });
+    b.installArtifact(tl);
+
     // Unit tests.
-    const tests = b.addTest(.{ .root_module = crucible_mod });
+    const tests = b.addTest(.{ .root_module = cruc_mod });
     const run_tests = b.addRunArtifact(tests);
     const test_step = b.step("test", "Run unit tests");
     test_step.dependOn(&run_tests.step);

examples/test_linear.zig

@@ -0,0 +1,29 @@
+//! Minimal known-answer test for layers.linear.
+const std = @import("std");
+const crucible = @import("crucible");
+
+pub fn main(init: std.process.Init) !void {
+    const alloc = init.gpa;
+
+    // in_n = 8, out_n = 1, bias 0. w = 1..8, x = 1..8 -> expect 204.
+    var w: [8]f32 = .{ 1, 2, 3, 4, 5, 6, 7, 8 };
+    var x: [8]f32 = .{ 1, 2, 3, 4, 5, 6, 7, 8 };
+    var b: [1]f32 = .{0};
+    var out: [1]f32 = undefined;
+    crucible.layers.linear(&x, &w, &b, &out);
+    std.debug.print("case1 (expect 204): {e}\n", .{out[0]});
+
+    // bias 100
+    b[0] = 100;
+    crucible.layers.linear(&x, &w, &b, &out);
+    std.debug.print("case2 (expect 304): {e}\n", .{out[0]});
+
+    // in_n = 4 (the pose_fc case!) w=1..4 x=1..4 -> 30
+    var w4: [4]f32 = .{ 1, 2, 3, 4 };
+    var x4: [4]f32 = .{ 1, 2, 3, 4 };
+    var b1: [1]f32 = .{0};
+    var out1: [1]f32 = undefined;
+    crucible.layers.linear(&x4, &w4, &b1, &out1);
+    std.debug.print("case3 (expect 30): {e}\n", .{out1[0]});
+    _ = alloc;
+}

src/layers.zig

@@ -79,7 +79,8 @@ pub fn conv2d(
 pub fn linear(input: []const f32, weight: []const f32, bias: ?[]const f32, output: []f32) void {
     const in_n = weight.len / output.len;
     for (0..output.len) |o| {
-        var acc: V = @splat(if (bias) |b| b[o] else 0);
+        var acc: V = @splat(0);
+        var tail: f32 = 0;
         const w_row = weight[o * in_n ..][0..in_n];
         var x: usize = 0;
         while (x + 8 <= in_n) : (x += 8) {
@@ -87,7 +88,11 @@ pub fn linear(input: []const f32, weight: []const f32, bias: ?[]const f32, outpu
             const wv: V = w_row[x..][0..8].*;
             acc = @mulAdd(V, iv, wv, acc);
         }
-        output[o] = @reduce(.Add, acc);
+        while (x < in_n) : (x += 1) {
+            tail += input[x] * w_row[x];
+        }
+        // bias added ONCE — splatting it into acc lanes would multiply it by 8
+        output[o] = @reduce(.Add, acc) + tail + (if (bias) |b| b[o] else 0);
     }
 }