whisper : add support for large v3 (#1444)

* whisper : add support for large v3

* bench : fix build + fix go bindings

* bench : fix n_mels

* models : update readme

models/README.md

@@ -50,7 +50,8 @@ https://huggingface.co/ggerganov/whisper.cpp/tree/main
 | medium    | 1.5 GB | ~2.6 GB | `fd9727b6e1217c2f614f9b698455c4ffd82463b4` |
 | medium.en | 1.5 GB | ~2.6 GB | `8c30f0e44ce9560643ebd10bbe50cd20eafd3723` |
 | large-v1  | 2.9 GB | ~4.7 GB | `b1caaf735c4cc1429223d5a74f0f4d0b9b59a299` |
-| large     | 2.9 GB | ~4.7 GB | `0f4c8e34f21cf1a914c59d8b3ce882345ad349d6` |
+| large-v2  | 2.9 GB | ~4.7 GB | `0f4c8e34f21cf1a914c59d8b3ce882345ad349d6` |
+| large     | 2.9 GB | ~4.7 GB | `ad82bf6a9043ceed055076d0fd39f5f186ff8062` |
 
 ## Model files for testing purposes
 

models/convert-h5-to-coreml.py

@@ -78,14 +78,14 @@ def convert_hf_whisper(hf_model_name_or_path: str, whisper_state_path: str):
 # Ported from models/convert-whisper-to-coreml.py
 if __name__ == "__main__":
     parser = argparse.ArgumentParser()
-    parser.add_argument("--model-name", type=str, help="name of model to convert (e.g. tiny, tiny.en, base, base.en, small, small.en, medium, medium.en, large, large-v1)", required=True)
+    parser.add_argument("--model-name", type=str, help="name of model to convert (e.g. tiny, tiny.en, base, base.en, small, small.en, medium, medium.en, large, large-v1, large-v2)", required=True)
     parser.add_argument("--model-path", type=str, help="path to the model (e.g. if published on HuggingFace: Oblivion208/whisper-tiny-cantonese)", required=True)
     parser.add_argument("--encoder-only", type=bool, help="only convert encoder", default=False)
     parser.add_argument("--quantize",     type=bool, help="quantize weights to F16", default=False)
     parser.add_argument("--optimize-ane", type=bool, help="optimize for ANE execution (currently broken)", default=False)
     args = parser.parse_args()
 
-    if args.model_name not in ["tiny", "tiny.en", "base", "base.en", "small", "small.en", "medium", "medium.en", "large", "large-v1"]:
+    if args.model_name not in ["tiny", "tiny.en", "base", "base.en", "small", "small.en", "medium", "medium.en", "large", "large-v1", "large-v2"]:
         raise ValueError("Invalid model name")
 
     pt_target_path = f"models/hf-{args.model_name}.pt"

models/convert-pt-to-ggml.py

@@ -228,7 +228,7 @@ with np.load(dir_whisper / "whisper" / "assets" / "mel_filters.npz") as f:
 # for backwards compatibility, also check for older hf_transformers format tokenizer files
 # old format: dir_whisper/whisper/assets/[multilingual/gpt2]/vocab.json
 # new format: dir_whisper/whisper/assets/[multilingual/gpt2].tiktoken
-multilingual = hparams["n_vocab"] == 51865
+multilingual = hparams["n_vocab"] >= 51865
 tokenizer = dir_whisper / "whisper" / "assets" / (multilingual and "multilingual.tiktoken" or "gpt2.tiktoken")
 tokenizer_type = "tiktoken"
 if not tokenizer.is_file():

models/convert-whisper-to-coreml.py

@@ -194,7 +194,7 @@ class TextDecoderANE(TextDecoder):
         x = x.permute(0,2,3,1).squeeze(0)
 
         # ANE can only load tensors with dim size of at most 16,384 - whisper uses 51,864 (en) or 51,865 (multi-lang) tokens so we need to compute in chunks
-        if self.token_embedding.weight.shape[0] == 51865:
+        if self.token_embedding.weight.shape[0] >= 51865:
             # split in 11 chunks - 4715 each
             splits = self.token_embedding.weight.split(self.token_embedding.weight.shape[0]//11, dim=0)
             logits = torch.cat([torch.einsum('bid,jd->bij', x, split) for split in splits]).view(*x.shape[:2], -1)
@@ -296,13 +296,13 @@ def convert_decoder(hparams, model, quantize=False):
 
 if __name__ == "__main__":
     parser = argparse.ArgumentParser()
-    parser.add_argument("--model", type=str, help="model to convert (e.g. tiny, tiny.en, base, base.en, small, small.en, medium, medium.en, large, large-v1)", required=True)
+    parser.add_argument("--model", type=str, help="model to convert (e.g. tiny, tiny.en, base, base.en, small, small.en, medium, medium.en, large, large-v1, large-v2)", required=True)
     parser.add_argument("--encoder-only", type=bool, help="only convert encoder", default=False)
     parser.add_argument("--quantize",     type=bool, help="quantize weights to F16", default=False)
     parser.add_argument("--optimize-ane", type=bool, help="optimize for ANE execution (currently broken)", default=False)
     args = parser.parse_args()
 
-    if args.model not in ["tiny", "tiny.en", "base", "base.en", "small", "small.en", "medium", "medium.en", "large", "large-v1"]:
+    if args.model not in ["tiny", "tiny.en", "base", "base.en", "small", "small.en", "medium", "medium.en", "large", "large-v1", "large-v2"]:
         raise ValueError("Invalid model name")
 
     whisper = load_model(args.model).cpu()

models/convert-whisper-to-openvino.py

@@ -38,10 +38,10 @@ def convert_encoder(hparams, encoder, mname):
 
 if __name__ == "__main__":
     parser = argparse.ArgumentParser()
-    parser.add_argument("--model", type=str, help="model to convert (e.g. tiny, tiny.en, base, base.en, small, small.en, medium, medium.en, large, large-v1)", required=True)
+    parser.add_argument("--model", type=str, help="model to convert (e.g. tiny, tiny.en, base, base.en, small, small.en, medium, medium.en, large, large-v1, large-v2)", required=True)
     args = parser.parse_args()
 
-    if args.model not in ["tiny", "tiny.en", "base", "base.en", "small", "small.en", "medium", "medium.en", "large", "large-v1"]:
+    if args.model not in ["tiny", "tiny.en", "base", "base.en", "small", "small.en", "medium", "medium.en", "large", "large-v1", "large-v2"]:
         raise ValueError("Invalid model name")
 
     whisper = load_model(args.model).cpu()

models/download-coreml-model.sh

@@ -19,7 +19,7 @@ function get_script_path() {
 models_path="$(get_script_path)"
 
 # Whisper models
-models=( "tiny.en" "tiny" "base.en" "base" "small.en" "small" "medium.en" "medium" "large-v1" "large" )
+models=( "tiny.en" "tiny" "base.en" "base" "small.en" "small" "medium.en" "medium" "large-v1" "large-v2" "large" )
 
 # list available models
 function list_models {

models/download-ggml-model.cmd

@@ -8,7 +8,7 @@ popd
 set argc=0
 for %%x in (%*) do set /A argc+=1
 
-set models=tiny.en tiny base.en base small.en small medium.en medium large-v1 large
+set models=tiny.en tiny base.en base small.en small medium.en medium large-v1 large-v2 large
 
 if %argc% neq 1 (
   echo.
@@ -57,8 +57,8 @@ goto :eof
 :list_models
   echo.
   echo Available models:
-  (for %%a in (%models%) do ( 
-    echo %%a 
+  (for %%a in (%models%) do (
+    echo %%a
   ))
   echo.
   exit /b

models/download-ggml-model.sh

@@ -41,6 +41,7 @@ models=(
     "medium-q5_0"
     "medium.en-q5_0"
     "large-v1"
+    "large-v2"
     "large"
     "large-q5_0"
 )