Metavoice - first cut (#1717)

* Add the metavoice transformer.

* Sketch the speaker-encoder module.

* Adding to the metavoice model.

* Start adding the metavoice example.

* Get some logits out.

* Load the second stage model.

* Get the second step to run.

* Tweak the example.

* Add encodec tilting.

* Glue the different bits together.

* Fix a shape issue.

* Use a constant.

* BPE tokenization.

* Add a warning.
This commit is contained in:
Laurent Mazare
2024-03-02 18:50:01 +01:00
committed by GitHub
parent 314630638d
commit 4fff5b51f5
6 changed files with 1117 additions and 0 deletions

View File

@ -42,6 +42,7 @@ candle-transformers = { path = "./candle-transformers", version = "0.4.1" }
clap = { version = "4.2.4", features = ["derive"] }
criterion = { version = "0.5.1", default-features=false }
cudarc = { version = "0.10.0", features = ["f16"] }
fancy-regex = "0.13.0"
gemm = { version = "0.17.0", features = ["wasm-simd128-enable"] }
hf-hub = "0.3.0"
half = { version = "2.3.1", features = ["num-traits", "use-intrinsics", "rand_distr"] }