Use uv in GH action pipeline

.github/workflows/nightly-evals.yml

@@ -18,8 +18,8 @@ jobs:
           
       - name: Install dependencies
         run: |
-          python -m pip install --upgrade pip
-          pip install -r requirements_dev.txt
+          curl -LsSf https://astral.sh/uv/install.sh | sh
+          uv sync --frozen
           
       - name: Run evaluations
         env:
@@ -27,9 +27,8 @@ jobs:
           ELEVENLABS_API_KEY: ${{ secrets.ELEVENLABS_API_KEY }}
           HUGGINGFACE_ACCESS_TOKEN: ${{ secrets.HUGGINGFACE_ACCESS_TOKEN }}
         run: |
-          cd evals
-          python download_data.py
-          python main.py
+          uv run download_data.py
+          uv run main.py
           
       - name: Commit changes
         run: |
@@ -41,7 +40,7 @@ jobs:
       
       - name: Install Hugging Face dependencies
         run: |
-          python -m pip install huggingface_hub
+          uv add huggingface_hub
           
       - name: Upload to Hugging Face
         env:

requirements_dev.txt

@@ -1,297 +0,0 @@
-# This file was autogenerated by uv via the following command:
-#    uv pip compile pyproject.toml -o requirements_dev.txt --extra dev
-aiohappyeyeballs==2.6.1
-    # via aiohttp
-aiohttp==3.11.18
-    # via
-    #   datasets
-    #   fsspec
-aiolimiter==1.2.1
-    # via languagebench (pyproject.toml)
-aiosignal==1.3.2
-    # via aiohttp
-annotated-types==0.7.0
-    # via pydantic
-anyio==4.9.0
-    # via
-    #   httpx
-    #   openai
-    #   starlette
-async-timeout==5.0.1
-    # via aiohttp
-attrs==25.3.0
-    # via aiohttp
-bert-score==0.3.13
-    # via languagebench (pyproject.toml)
-certifi==2025.4.26
-    # via
-    #   httpcore
-    #   httpx
-    #   requests
-charset-normalizer==3.4.2
-    # via requests
-click==8.1.8
-    # via
-    #   jiwer
-    #   uvicorn
-colorama==0.4.6
-    # via sacrebleu
-contourpy==1.3.2
-    # via matplotlib
-cycler==0.12.1
-    # via matplotlib
-datasets==3.5.1
-    # via evaluate
-dill==0.3.8
-    # via
-    #   datasets
-    #   evaluate
-    #   multiprocess
-distro==1.9.0
-    # via openai
-elevenlabs==1.57.0
-    # via languagebench (pyproject.toml)
-evaluate==0.4.0
-    # via languagebench (pyproject.toml)
-exceptiongroup==1.2.2
-    # via anyio
-fastapi==0.115.12
-    # via languagebench (pyproject.toml)
-filelock==3.18.0
-    # via
-    #   datasets
-    #   huggingface-hub
-    #   torch
-    #   transformers
-fonttools==4.57.0
-    # via matplotlib
-frozenlist==1.6.0
-    # via
-    #   aiohttp
-    #   aiosignal
-fsspec==2025.3.0
-    # via
-    #   datasets
-    #   evaluate
-    #   huggingface-hub
-    #   torch
-h11==0.16.0
-    # via
-    #   httpcore
-    #   uvicorn
-httpcore==1.0.9
-    # via httpx
-httpx==0.28.1
-    # via
-    #   elevenlabs
-    #   openai
-huggingface-hub==0.30.2
-    # via
-    #   languagebench (pyproject.toml)
-    #   datasets
-    #   evaluate
-    #   tokenizers
-    #   transformers
-idna==3.10
-    # via
-    #   anyio
-    #   httpx
-    #   requests
-    #   yarl
-jinja2==3.1.6
-    # via torch
-jiter==0.9.0
-    # via openai
-jiwer==3.1.0
-    # via languagebench (pyproject.toml)
-joblib==1.5.0
-    # via languagebench (pyproject.toml)
-kiwisolver==1.4.8
-    # via matplotlib
-langcodes==3.5.0
-    # via languagebench (pyproject.toml)
-language-data==1.3.0
-    # via
-    #   languagebench (pyproject.toml)
-    #   langcodes
-lxml==5.4.0
-    # via sacrebleu
-marisa-trie==1.2.1
-    # via language-data
-markdown-it-py==3.0.0
-    # via rich
-markupsafe==3.0.2
-    # via jinja2
-matplotlib==3.10.1
-    # via bert-score
-mdurl==0.1.2
-    # via markdown-it-py
-mpmath==1.3.0
-    # via sympy
-multidict==6.4.3
-    # via
-    #   aiohttp
-    #   yarl
-multiprocess==0.70.16
-    # via
-    #   datasets
-    #   evaluate
-networkx==3.4.2
-    # via torch
-numpy==2.2.5
-    # via
-    #   languagebench (pyproject.toml)
-    #   bert-score
-    #   contourpy
-    #   datasets
-    #   evaluate
-    #   matplotlib
-    #   pandas
-    #   sacrebleu
-    #   transformers
-openai==1.77.0
-    # via languagebench (pyproject.toml)
-packaging==25.0
-    # via
-    #   bert-score
-    #   datasets
-    #   evaluate
-    #   huggingface-hub
-    #   matplotlib
-    #   transformers
-pandas==2.2.3
-    # via
-    #   languagebench (pyproject.toml)
-    #   bert-score
-    #   datasets
-    #   evaluate
-pillow==11.2.1
-    # via matplotlib
-portalocker==3.1.1
-    # via sacrebleu
-propcache==0.3.1
-    # via
-    #   aiohttp
-    #   yarl
-protobuf==6.30.2
-    # via languagebench (pyproject.toml)
-pyarrow==20.0.0
-    # via datasets
-pydantic==2.11.4
-    # via
-    #   elevenlabs
-    #   fastapi
-    #   openai
-pydantic-core==2.33.2
-    # via
-    #   elevenlabs
-    #   pydantic
-pygments==2.19.1
-    # via rich
-pyparsing==3.2.3
-    # via matplotlib
-python-dateutil==2.9.0.post0
-    # via
-    #   matplotlib
-    #   pandas
-python-dotenv==1.1.0
-    # via languagebench (pyproject.toml)
-pytz==2025.2
-    # via pandas
-pyyaml==6.0.2
-    # via
-    #   datasets
-    #   huggingface-hub
-    #   transformers
-rapidfuzz==3.13.0
-    # via jiwer
-regex==2024.11.6
-    # via
-    #   sacrebleu
-    #   tiktoken
-    #   transformers
-requests==2.32.3
-    # via
-    #   bert-score
-    #   datasets
-    #   elevenlabs
-    #   evaluate
-    #   huggingface-hub
-    #   responses
-    #   tiktoken
-    #   transformers
-responses==0.18.0
-    # via evaluate
-rich==14.0.0
-    # via languagebench (pyproject.toml)
-sacrebleu==2.5.1
-    # via languagebench (pyproject.toml)
-safetensors==0.5.3
-    # via transformers
-sentencepiece==0.2.0
-    # via languagebench (pyproject.toml)
-setuptools==80.3.1
-    # via marisa-trie
-six==1.17.0
-    # via python-dateutil
-sniffio==1.3.1
-    # via
-    #   anyio
-    #   openai
-starlette==0.46.2
-    # via fastapi
-sympy==1.14.0
-    # via torch
-tabulate==0.9.0
-    # via sacrebleu
-tiktoken==0.9.0
-    # via languagebench (pyproject.toml)
-tokenizers==0.21.1
-    # via transformers
-torch==2.7.0
-    # via bert-score
-tqdm==4.67.1
-    # via
-    #   languagebench (pyproject.toml)
-    #   bert-score
-    #   datasets
-    #   evaluate
-    #   huggingface-hub
-    #   openai
-    #   transformers
-transformers==4.51.3
-    # via
-    #   languagebench (pyproject.toml)
-    #   bert-score
-typing-extensions==4.13.2
-    # via
-    #   anyio
-    #   elevenlabs
-    #   fastapi
-    #   huggingface-hub
-    #   multidict
-    #   openai
-    #   pydantic
-    #   pydantic-core
-    #   rich
-    #   torch
-    #   typing-inspection
-    #   uvicorn
-typing-inspection==0.4.0
-    # via pydantic
-tzdata==2025.2
-    # via pandas
-urllib3==2.4.0
-    # via
-    #   requests
-    #   responses
-uvicorn==0.34.2
-    # via languagebench (pyproject.toml)
-websockets==15.0.1
-    # via elevenlabs
-xxhash==3.5.0
-    # via
-    #   datasets
-    #   evaluate
-yarl==1.20.0
-    # via aiohttp