add GH actions

.github/workflows/quality.yml

@@ -0,0 +1,30 @@
+name: Quality
+
+on:
+  push:
+    branches:
+      - main
+      - v*-release
+  pull_request:
+    branches:
+      - main
+
+jobs:
+
+  check_code_quality:
+    name: Check code quality
+    runs-on: ubuntu-latest
+    steps:
+      - name: Checkout code
+        uses: actions/checkout@v2
+      - name: Setup Python environment
+        uses: actions/setup-python@v2
+        with:
+          python-version: 3.10.10
+      - name: Install dependencies
+        run: |
+          python -m pip install --upgrade pip
+          python -m pip install .
+      - name: Code quality
+        run: |
+          make quality

.github/workflows/tests.yml

@@ -0,0 +1,31 @@
+name: Tests
+
+on:
+  push:
+    branches:
+      - main
+      - v*-release
+  pull_request:
+    branches:
+      - main
+
+jobs:
+
+  unit-tests:
+    name: Run unit tests
+    env:
+      HF_TOKEN: ${{ secrets.HF_TOKEN }}
+    runs-on: ubuntu-latest
+    steps:
+      - name: Checkout code
+        uses: actions/checkout@v2
+      - name: Setup Python environment
+        uses: actions/setup-python@v2
+        with:
+          python-version: 3.10.10
+      - name: Install dependencies
+        run: |
+          python -m pip install --upgrade pip
+          python -m pip install .
+      - name: Run unit tests
+        run: HF_TOKEN=$HF_TOKEN pytest tests/

scripts/run_rm.py

@@ -23,7 +23,6 @@
 import transformers
 from accelerate import Accelerator
 from accelerate.logging import get_logger
-from datasets import concatenate_datasets, load_dataset
 from fastchat.conversation import get_conv_template
 from huggingface_hub import HfApi
 from tqdm import tqdm
@@ -34,7 +33,7 @@
     pipeline,
 )
 
-from herm import load_eval_dataset, prepare_dialogue, prepare_dialogue_from_tokenizer
+from herm import load_eval_dataset
 
 # get token from HF_TOKEN env variable, but if it doesn't exist pass none
 HF_TOKEN = os.getenv("HF_TOKEN", None)
@@ -44,31 +43,6 @@
 EVAL_REPO = "ai2-rlhf-collab/rm-benchmark-results"
 PREFS_REPO = "ai2-rlhf-collab/rm-testset-results"
 
-EVAL_SUBSETS = [  # noqa
-    # custom subsets above
-    "alpacaeval-easy",
-    "alpacaeval-hard",
-    "alpacaeval-length",
-    "llmbar-adver-GPTInst",
-    "llmbar-adver-GPTOut",
-    "llmbar-adver-manual",
-    "llmbar-adver-neighbor",
-    "llmbar-natural",
-    "mt-bench-easy",
-    "mt-bench-hard",
-    "mt-bench-med",
-    "refusals-dangerous",
-    "refusals-offensive",
-    # pref subsets below
-    "anthropic",
-    "summarize",
-    "summarize_prompted",
-    "pku_better",
-    "pku_safer",
-    "shp",
-]
-
-
 def get_args():
     """
     Parse arguments strings model and chat_template