| set -o pipefail | |
| set -ex | |
| shopt -s expand_aliases | |
| if ! test -f /usr/bin/sudo; then | |
| echo "No sudo" | |
| alias sudo=' ' | |
| fi | |
| #* Optional: For document Q/A and use of DocTR. Install before other pips to avoid long conflict checks. | |
| conda install weasyprint pygobject -c conda-forge -y | |
| # Avoids library mismatch. | |
| #* Install primary dependencies | |
| # fix any bad env | |
| pip uninstall -y pandoc pypandoc pypandoc-binary flash-attn | |
| # broad support, but no training-time or data creation dependencies | |
| pip install -r requirements.txt | |
| #* Optional: Install document question-answer dependencies: | |
| # May be required for jq package: | |
| sudo apt-get -y install autoconf libtool | |
| # Required for Doc Q/A: LangChain: | |
| pip install -r reqs_optional/requirements_optional_langchain.txt | |
| # Required for CPU: LLaMa/GPT4All: | |
| pip install -r reqs_optional/requirements_optional_gpt4all.txt | |
| # Optional: PyMuPDF/ArXiv: | |
| # Note!! that pymupdf is AGPL, requiring any source code be made available, but it's like GPL and too strong a constraint for general commercial use. | |
| if [ "${GPLOK}" -eq "1" ] | |
| then | |
| pip install -r reqs_optional/requirements_optional_langchain.gpllike.txt | |
| fi | |
| # Optional: FAISS | |
| pip install -r reqs_optional/requirements_optional_faiss.txt | |
| # Optional: Selenium/PlayWright: | |
| pip install -r reqs_optional/requirements_optional_langchain.urls.txt | |
| # Optional: support docx, pptx, ArXiv, etc. required by some python packages | |
| sudo apt-get install -y libmagic-dev poppler-utils tesseract-ocr libtesseract-dev libreoffice | |
| # Optional: For DocTR | |
| pip install -r reqs_optional/requirements_optional_doctr.txt | |
| # For DocTR: go back to older onnx so Tesseract OCR still works | |
| pip install onnxruntime==1.15.0 | |
| # GPU only: | |
| pip install onnxruntime-gpu==1.15.0 | |
| # Optional: for supporting unstructured package | |
| for i in 1 2 3 4; do python -m nltk.downloader all && break || sleep 1; done # retry as frequently fails with github downloading issues | |
| # Optional: Required for PlayWright | |
| playwright install --with-deps | |
| # Audio transcription from Youtube videos and local mp3 files: | |
| pip install pydub==0.25.1 librosa==0.10.1 ffmpeg==1.4 yt_dlp==2023.10.13 wavio==0.0.8 | |
| # Audio speed-up and slowdown (best quality), if not installed can only speed-up with lower quality | |
| sudo apt-get install -y rubberband-cli | |
| pip install pyrubberband==0.3.0 | |
| # https://stackoverflow.com/questions/75813603/python-working-with-sound-librosa-and-pyrubberband-conflict | |
| pip uninstall -y pysoundfile soundfile | |
| pip install soundfile==0.12.1 | |
| # Optional: Only for testing for now | |
| pip install playsound==1.3.0 | |
| # STT from microphone (may not be required if ffmpeg installed above) | |
| sudo apt-get install ffmpeg | |
| # for any TTS: | |
| pip install torchaudio soundfile==0.12.1 | |
| # GPU Only: for Coqui XTTS (ensure CUDA_HOME set and consistent with added postfix for extra-index): | |
| # relaxed versions to avoid conflicts | |
| pip install TTS deepspeed noisereduce emoji ffmpeg-python==0.2.0 trainer pysbd coqpit | |
| # for Coqui XTTS language helpers (specific versions probably not required) | |
| pip install cutlet==0.3.0 langid==1.1.6 g2pkk==0.1.2 jamo==0.4.1 gruut[de,es,fr]==2.2.3 jieba==0.42.1 | |
| # For faster whisper: | |
| pip install git+https://github.com/SYSTRAN/faster-whisper.git | |
| # needed for librosa/soundfile to work, but violates TTS, but that's probably just too strict as we have seen before) | |
| pip install numpy==1.23.0 --no-deps --upgrade | |
| # TTS or other deps load old librosa, fix: | |
| pip install librosa==0.10.1 --no-deps --upgrade | |
| #* STT and TTS Notes: | |
| # * STT: Ensure microphone is on and in browser go to http://localhost:7860 instead of http://0.0.0.0:7860 for microphone to be possible to allow in browser. | |
| # * TTS: For XTT models, ensure `CUDA_HOME` is set correctly, because deepspeed compiles at runtime using torch and nvcc. Those must match CUDA version. E.g. if used `--extra-index https://download.pytorch.org/whl/cu118`, then must have ENV `CUDA_HOME=/usr/local/cuda-11.7` or ENV from conda must be that version. Since conda only has up to cuda 11.7 for dev toolkit, but H100+ need cuda 11.8, for those cases one should download the toolkit from NVIDIA. | |
| # Vision/Image packages | |
| pip install fiftyone | |
| pip install pytube | |
| pip install diffusers==0.24.0 | |
| #* HNSW issue: | |
| # In some cases old chroma migration package will install old hnswlib and that may cause issues when making a database, then do: | |
| pip uninstall -y hnswlib chroma-hnswlib | |
| # restore correct version | |
| pip install chroma-hnswlib==0.7.3 --upgrade | |
| #* Selenium needs to have chrome installed, e.g. on Ubuntu: | |
| sudo apt install -y unzip xvfb libxi6 libgconf-2-4 libu2f-udev | |
| sudo apt install -y default-jdk | |
| if [ 1 -eq 0 ]; then | |
| sudo bash -c 'curl -sS -o - https://dl-ssl.google.com/linux/linux_signing_key.pub | apt-key add' | |
| sudo bash -c "echo 'deb [arch=amd64] http://dl.google.com/linux/chrome/deb/ stable main' >> /etc/apt/sources.list.d/google-chrome.list" | |
| sudo apt -y update | |
| sudo apt -y install google-chrome-stable # e.g. Google Chrome 114.0.5735.198 | |
| fi | |
| wget http://dl.google.com/linux/chrome/deb/pool/main/g/google-chrome-stable/google-chrome-stable_114.0.5735.198-1_amd64.deb | |
| sudo dpkg -i google-chrome-stable_114.0.5735.198-1_amd64.deb | |
| sudo google-chrome --version # e.g. Google Chrome 114.0.5735.198 | |
| # visit https://chromedriver.chromium.org/downloads and download matching version | |
| # E.g. | |
| sudo rm -rf chromedriver_linux64.zip chromedriver LICENSE.chromedriver | |
| sudo wget https://chromedriver.storage.googleapis.com/114.0.5735.90/chromedriver_linux64.zip | |
| sudo unzip chromedriver_linux64.zip | |
| sudo mv chromedriver /usr/bin/chromedriver | |
| sudo chown root:root /usr/bin/chromedriver | |
| sudo chmod +x /usr/bin/chromedriver | |
| #* GPU Optional: For AutoGPTQ support on x86_64 linux | |
| pip uninstall -y auto-gptq ; pip install auto-gptq==0.6.0 --extra-index-url https://huggingface.github.io/autogptq-index/whl/cu118/ | |
| # in-transformers support of AutoGPTQ, requires also auto-gptq above to be installed since used internally by transformers/optimum | |
| pip install optimum==1.16.1 | |
| # See [AutoGPTQ](README_GPU.md#autogptq) about running AutoGPT models. | |
| #* GPU Optional: For AutoAWQ support on x86_64 linux | |
| pip uninstall -y autoawq ; pip install https://github.com/casper-hansen/AutoAWQ/releases/download/v0.1.8/autoawq-0.1.8+cu118-cp310-cp310-linux_x86_64.whl | |
| # fix version since don't need lm-eval to have its version of 1.5.0 | |
| pip install sacrebleu==2.3.1 --upgrade | |
| # If this has issues, you need to build: | |
| if [ 1 -eq 0 ] | |
| then | |
| pip uninstall -y autoawq | |
| git clone https://github.com/casper-hansen/AutoAWQ | |
| cd AutoAWQ | |
| pip install . | |
| fi | |
| #* GPU Optional: For exllama support on x86_64 linux | |
| pip uninstall -y exllama ; pip install https://github.com/jllllll/exllama/releases/download/0.0.18/exllama-0.0.18+cu118-cp310-cp310-linux_x86_64.whl --no-cache-dir | |
| # See [exllama](README_GPU.md#exllama) about running exllama models. | |
| # * If any issues with llama_cpp_python, then must compile llama-cpp-python with CUDA support: | |
| if [ 1 -eq 0 ] | |
| then | |
| pip uninstall -y llama-cpp-python llama-cpp-python-cuda | |
| export LLAMA_CUBLAS=1 | |
| export CMAKE_ARGS=-DLLAMA_CUBLAS=on | |
| export FORCE_CMAKE=1 | |
| CMAKE_ARGS="-DLLAMA_CUBLAS=on" FORCE_CMAKE=1 pip install llama-cpp-python==0.2.26 --no-cache-dir --verbose | |
| fi | |
| # * By default, we set `n_gpu_layers` to large value, so llama.cpp offloads all layers for maximum GPU performance. You can control this by passing `--llamacpp_dict="{'n_gpu_layers':20}"` for value 20, or setting in UI. For highest performance, offload *all* layers. | |
| # That is, one gets maximum performance if one sees in startup of h2oGPT all layers offloaded: | |
| # ```text | |
| # llama_model_load_internal: offloaded 35/35 layers to GPU | |
| # ``` | |
| # but this requires sufficient GPU memory. Reduce if you have low memory GPU, say 15. | |
| # * Pass to `generate.py` the option `--max_seq_len=2048` or some other number if you want model have controlled smaller context, else default (relatively large) value is used that will be slower on CPU. | |
| # * For LLaMa2, can set `max_tokens` to a larger value for longer output. | |
| # * If one sees `/usr/bin/nvcc` mentioned in errors, that file needs to be removed as would likely conflict with version installed for conda. | |
| # * Note that once `llama-cpp-python` is compiled to support CUDA, it no longer works for CPU mode, so one would have to reinstall it without the above options to recovers CPU mode or have a separate h2oGPT env for CPU mode. | |
| #* GPU Optional: Support amazon/MistralLite with flash attention 2 | |
| if [[ -v CUDA_HOME ]]; | |
| then | |
| pip install flash-attn==2.3.4 --no-build-isolation | |
| fi | |
| #* Control Core Count for chroma < 0.4 using chromamigdb package: | |
| # * Duckdb used by Chroma < 0.4 uses DuckDB 0.8.1 that has no control over number of threads per database, `import duckdb` leads to all virtual cores as threads and each db consumes another number of threads equal to virtual cores. To prevent this, one can rebuild duckdb using [this modification](https://github.com/h2oai/duckdb/commit/dcd8c1ffc53dd020623630efb99ba6a3a4cbc5ad) or one can try to use the prebuild wheel for x86_64 built on Ubuntu 20. | |
| pip uninstall -y pyduckdb duckdb | |
| pip install https://h2o-release.s3.amazonaws.com/h2ogpt/duckdb-0.8.2.dev4025%2Bg9698e9e6a8.d20230907-cp310-cp310-linux_x86_64.whl --no-cache-dir --force-reinstall --no-deps | |
| #* SERP for search: | |
| pip install -r reqs_optional/requirements_optional_agents.txt | |
| # For more info see [SERP Docs](README_SerpAPI.md). | |
| #* Deal with not-thread-safe things in LangChain: | |
| pwd0=`pwd` | |
| sp=`python3.10 -c 'import site; print(site.getsitepackages()[0])'` | |
| cd $sp | |
| sed -i 's/with HiddenPrints():/if True:/g' langchain/utilities/serpapi.py | |
| #sed -i 's/"progress": Status.PROGRESS,/"progress": Status.PROGRESS,\n "heartbeat": Status.PROGRESS,/g' gradio_client/utils.py | |
| #sed -i 's/async for line in response.aiter_text():/async for line in response.aiter_lines():\n if len(line) == 0:\n continue\n if line == """{"detail":"Not Found"}""":\n continue/g' gradio_client/utils.py | |
| cd $pwd0 | |
| # fix pytube to avoid errors for restricted content | |
| sp=`python3.10 -c 'import site; print(site.getsitepackages()[0])'` | |
| sed -i "s/client='ANDROID_MUSIC'/client='ANDROID'/g" $sp/pytube/innertube.py | |
| # fix asyncio same way websockets was fixed, else keep hitting errors in async calls | |
| # https://github.com/python-websockets/websockets/commit/f9fd2cebcd42633ed917cd64e805bea17879c2d7 | |
| sp=`python3.10 -c 'import site; print(site.getsitepackages()[0])'` | |
| sed -i "s/except OSError:/except (OSError, RuntimeError):/g" $sp/anyio/_backends/_asyncio.py | |
| #* PDF View support | |
| # only if using gradio4 | |
| #pip install https://h2o-release.s3.amazonaws.com/h2ogpt/gradio_pdf-0.0.3-py3-none-any.whl | |
| ### Compile Install Issues | |
| # * `/usr/local/cuda/include/crt/host_config.h:132:2: error: #error -- unsupported GNU version! gcc versions later than 11 are not supported!` | |
| # * gcc > 11 is not currently supported by nvcc. Install GCC with a maximum version: | |
| if [ 1 -eq 0 ] | |
| then | |
| MAX_GCC_VERSION=11 | |
| sudo apt install gcc-$MAX_GCC_VERSION g++-$MAX_GCC_VERSION | |
| sudo update-alternatives --config gcc | |
| # pick version 11 | |
| sudo update-alternatives --config g++ | |
| # pick version 11 | |
| fi | |