Skip to content

fix bpb for llama tokenizer (which has weird space handling) #4622

fix bpb for llama tokenizer (which has weird space handling)

fix bpb for llama tokenizer (which has weird space handling) #4622

Workflow file for this run

name: Run tests
on: [push, pull_request]
jobs:
build:
if: github.event_name == 'push' || (github.event_name == 'pull_request' && github.event.pull_request.head.repo.full_name != github.repository)
runs-on: ubuntu-latest
strategy:
matrix:
python-version: ["3.10"]
jax-version: ["0.4.26"]
steps:
- uses: actions/checkout@v3
- name: Set up Python ${{ matrix.python-version }}
uses: actions/setup-python@v4
with:
python-version: ${{ matrix.python-version }}
- name: Install dependencies
run: |
python -m pip install --upgrade pip
pip install .[test] "jax[cpu]==${{ matrix.jax-version }}" "jaxlib==${{ matrix.jax-version }}"
pip install -r ./tests/requirements.txt
- name: Test with pytest
run: |
XLA_FLAGS=--xla_force_host_platform_device_count=8 PYTHONPATH=tests:src:. pytest tests -m "not entry and not slow and not ray"