2 days ago
-
Add the quantized codec checkpoints as safetensors: decoder int8, encoder int8 and int4
-
Add the XNNPACK encoder: fp32 DAC, fp16 HuBERT with int8 transformer linears (25-500 frames)
9 days ago
-
Add the Vulkan encoders, QAT int8 and int4 (25-500 frames)
10 days ago
-
Add get_model_schema to the programs (dynamic lengths for react-native-executorch)
11 days ago
-
Codec programs with one method, forward (was decode / encode); outputs identical
-
Revert the plain weights: remove decoder/plain and encoder/plain, README as at 10e8a97
-
Add the plain weights: decoder fp16 and int8, encoder int8 and int4; README
-
Lay the programs out as encoder/coreml and decoder/coreml