Update with dockerfile

dockerfile

@@ -0,0 +1,30 @@
+# docker build -t whisper-webui --build-arg WHISPER_IMPLEMENTATION=faster-whisper .
+
+FROM huggingface/transformers-pytorch-gpu
+EXPOSE 7860
+
+ARG WHISPER_IMPLEMENTATION=faster-whisper
+ENV WHISPER_IMPLEMENTATION=${WHISPER_IMPLEMENTATION}
+
+ADD . /opt/whisper-webui/
+
+# Latest version of transformers-pytorch-gpu seems to lack tk. 
+# Further, pip install fails, so we must upgrade pip first.
+RUN apt-get -y install python3-tk
+RUN  python3 -m pip install --upgrade pip
+
+RUN if [ "${WHISPER_IMPLEMENTATION}" = "whisper" ]; then \
+    python3 -m pip install -r /opt/whisper-webui/requirements-whisper.txt; \
+  else \
+    python3 -m pip install -r /opt/whisper-webui/requirements-fasterWhisper.txt; \
+  fi
+
+# Note: Models will be downloaded on demand to the directory /root/.cache/whisper.
+# You can also bind this directory in the container to somewhere on the host.
+
+# To be able to see logs in real time
+ENV PYTHONUNBUFFERED=1
+
+WORKDIR /opt/whisper-webui/
+ENTRYPOINT ["python3"]
+CMD ["app.py", "--whisper_implementation", "faster-whisper", "--input_audio_max_duration", "-1", "--server_name", "0.0.0.0", "--auto_parallel", "True"]

pre-requirements.txt

@@ -1,5 +0,0 @@
-ctranslate2==4.4.0
-torch
-torchaudio
-nvidia-cublas-cu12
-nvidia-cudnn-cu12==9.*

requirements-fasterWhisper.txt

@@ -0,0 +1,36 @@
+transformers>=4.45.2
+ctranslate2>=4.4.0
+faster-whisper>=1.0.3
+ffmpeg-python==0.2.0
+gradio==5.9.1
+yt-dlp
+json5
+torch
+torchaudio
+more_itertools
+zhconv
+sentencepiece
+
+pyannote.audio
+torchmetrics
+lightning
+hydra-core==1.3.2
+hydra-colorlog==1.2.0
+llama-cookbook
+wandb
+rich
+git+https://github.com/lucas-ventura/lutils.git
+git+https://github.com/idriscnrs/idr_torch.git
+pycocoevalcap
+prettytable
+hf_transfer
+bitsandbytes
+
+# Needed by ALMA-GPTQ
+accelerate
+auto-gptq
+optimum
+# Needed by ALMA-GGUL
+ctransformers[cuda]
+# Needed by load_in_4bit parameters in transformers
+bitsandbytes

requirements-whisper.txt

@@ -0,0 +1,36 @@
+transformers>=4.45.2
+ctranslate2>=4.4.0
+git+https://github.com/openai/whisper.git
+ffmpeg-python==0.2.0
+gradio==5.9.1
+yt-dlp
+json5
+torch
+torchaudio
+altair
+zhconv
+sentencepiece
+
+pyannote.audio
+torchmetrics
+lightning
+hydra-core==1.3.2
+hydra-colorlog==1.2.0
+llama-cookbook
+wandb
+rich
+git+https://github.com/lucas-ventura/lutils.git
+git+https://github.com/idriscnrs/idr_torch.git
+pycocoevalcap
+prettytable
+hf_transfer
+bitsandbytes
+
+# Needed by ALMA-GPTQ
+accelerate
+auto-gptq
+optimum
+# Needed by ALMA-GGUL
+ctransformers[cuda]
+# Needed by load_in_4bit parameters in transformers
+bitsandbytes

requirements.txt

@@ -1,3 +1,16 @@
+transformers>=4.45.2
+ctranslate2>=4.4.0
+faster-whisper>=1.0.3
+ffmpeg-python==0.2.0
+gradio==5.9.1
+yt-dlp
+json5
+torch
+torchaudio
+more_itertools
+zhconv
+sentencepiece
+
 pyannote.audio
 torchmetrics
 lightning
@@ -11,8 +24,13 @@ git+https://github.com/idriscnrs/idr_torch.git
 pycocoevalcap
 prettytable
 hf_transfer
-faster-whisper
-ctranslate2==4.4.0
-gradio
 bitsandbytes
-yt-dlp
+
+# Needed by ALMA-GPTQ
+accelerate
+auto-gptq
+optimum
+# Needed by ALMA-GGUL
+ctransformers[cuda]
+# Needed by load_in_4bit parameters in transformers
+bitsandbytes