MCPcopy Create free account
hub / github.com/0xShug0/audio.cpp / load_snake_exact

Function load_snake_exact

src/models/ace_step/vae_decoder.cpp:223–240  ·  view source on GitHub ↗

Source from the content-addressed store, hash-verified

221 core::TensorShape::from_dims({1, channels, 1}),
222 assets::TensorStorageType::F32,
223 snake_exp_transform(alpha));
224 out.beta_inv = store.make_from_f32(
225 core::TensorShape::from_dims({1, channels, 1}),
226 assets::TensorStorageType::F32,
227 snake_inverse_beta_transform(beta));
228 return out;
229}
230
231WeightNormConv1dWeights load_weight_norm_conv1d(
232 core::BackendWeightStore & store,
233 const assets::TensorSource & source,
234 const std::string & prefix,
235 int64_t out_channels,
236 int64_t in_channels,
237 int64_t kernel,
238 int stride,
239 int padding,
240 int dilation,
241 assets::TensorStorageType storage_type,
242 bool use_bias) {
243 const auto g = source.require_f32(prefix + ".weight_g", {out_channels, 1, 1});

Callers 3

load_residual_unitFunction · 0.85
load_vae_decoder_weightsFunction · 0.85
load_vae_encoder_weightsFunction · 0.85

Calls 4

snake_exp_transformFunction · 0.85
make_from_f32Method · 0.80
require_f32Method · 0.45

Tested by

no test coverage detected