Clifford Heath commited on
Commit
d94de9a
·
unverified ·
1 Parent(s): b44a680

readme : add instructions on converting to GGML + "--no-config" to wget (#874)

Browse files
Files changed (3) hide show
  1. README.md +2 -0
  2. models/README.md +17 -5
  3. models/download-ggml-model.sh +1 -1
README.md CHANGED
@@ -71,6 +71,8 @@ Then, download one of the Whisper models converted in [ggml format](models). For
71
  bash ./models/download-ggml-model.sh base.en
72
  ```
73
 
 
 
74
  Now build the [main](examples/main) example and transcribe an audio file like this:
75
 
76
  ```bash
 
71
  bash ./models/download-ggml-model.sh base.en
72
  ```
73
 
74
+ If you wish to convert the Whisper models to ggml format yourself, instructions are in [models/README.md](models/README.md).
75
+
76
  Now build the [main](examples/main) example and transcribe an audio file like this:
77
 
78
  ```bash
models/README.md CHANGED
@@ -1,15 +1,17 @@
1
  ## Whisper model files in custom ggml format
2
 
3
  The [original Whisper PyTorch models provided by OpenAI](https://github.com/openai/whisper/blob/main/whisper/__init__.py#L17-L27)
4
- have been converted to custom `ggml` format in order to be able to load them in C/C++. The conversion has been performed
5
- using the [convert-pt-to-ggml.py](convert-pt-to-ggml.py) script. You can either obtain the original models and generate
6
- the `ggml` files yourself using the conversion script, or you can use the [download-ggml-model.sh](download-ggml-model.sh)
7
- script to download the already converted models. Currently, they are hosted on the following locations:
 
 
8
 
9
  - https://huggingface.co/ggerganov/whisper.cpp
10
  - https://ggml.ggerganov.com
11
 
12
- Sample usage:
13
 
14
  ```java
15
  $ ./download-ggml-model.sh base.en
@@ -21,6 +23,16 @@ You can now use it like this:
21
  $ ./main -m models/ggml-base.en.bin -f samples/jfk.wav
22
  ```
23
 
 
 
 
 
 
 
 
 
 
 
24
  A third option to obtain the model files is to download them from Hugging Face:
25
 
26
  https://huggingface.co/ggerganov/whisper.cpp/tree/main
 
1
  ## Whisper model files in custom ggml format
2
 
3
  The [original Whisper PyTorch models provided by OpenAI](https://github.com/openai/whisper/blob/main/whisper/__init__.py#L17-L27)
4
+ are converted to custom `ggml` format in order to be able to load them in C/C++.
5
+ Conversion is performed using the [convert-pt-to-ggml.py](convert-pt-to-ggml.py) script.
6
+
7
+ You can either obtain the original models and generate the `ggml` files yourself using the conversion script,
8
+ or you can use the [download-ggml-model.sh](download-ggml-model.sh) script to download the already converted models.
9
+ Currently, they are hosted on the following locations:
10
 
11
  - https://huggingface.co/ggerganov/whisper.cpp
12
  - https://ggml.ggerganov.com
13
 
14
+ Sample download:
15
 
16
  ```java
17
  $ ./download-ggml-model.sh base.en
 
23
  $ ./main -m models/ggml-base.en.bin -f samples/jfk.wav
24
  ```
25
 
26
+ To convert the files yourself, use the convert-pt-to-ggml.py script. Here is an example usage.
27
+ The original PyTorch files are assumed to have been downloaded into ~/.cache/whisper
28
+ Change `~/path/to/repo/whisper/` to the location for your copy of the Whisper source:
29
+ ```
30
+ mkdir models/whisper-medium
31
+ python models/convert-pt-to-ggml.py ~/.cache/whisper/medium.pt ~/path/to/repo/whisper/ ./models/whisper-medium
32
+ mv ./models/whisper-medium/ggml-model.bin models/ggml-medium.bin
33
+ rmdir models/whisper-medium
34
+ ```
35
+
36
  A third option to obtain the model files is to download them from Hugging Face:
37
 
38
  https://huggingface.co/ggerganov/whisper.cpp/tree/main
models/download-ggml-model.sh CHANGED
@@ -62,7 +62,7 @@ if [ -f "ggml-$model.bin" ]; then
62
  fi
63
 
64
  if [ -x "$(command -v wget)" ]; then
65
- wget --quiet --show-progress -O ggml-$model.bin $src/$pfx-$model.bin
66
  elif [ -x "$(command -v curl)" ]; then
67
  curl -L --output ggml-$model.bin $src/$pfx-$model.bin
68
  else
 
62
  fi
63
 
64
  if [ -x "$(command -v wget)" ]; then
65
+ wget --no-config --quiet --show-progress -O ggml-$model.bin $src/$pfx-$model.bin
66
  elif [ -x "$(command -v curl)" ]; then
67
  curl -L --output ggml-$model.bin $src/$pfx-$model.bin
68
  else