Merge pull request #27 from SmudgedWings/main

ci_update

Author: llmc-reviewer

.github/workflows/lint.yml

@@ -4,6 +4,9 @@ on:
   pull_request:
     branches:
       - main
+  push:
+    branches:
+      - main
 
 concurrency:
   group: ${{ github.workflow }}-${{ github.ref }}

.github/workflows/main.yml

@@ -4,6 +4,9 @@ on:
   pull_request:
     branches:
       - main
+  push:
+    branches:
+      - main
 
 concurrency:
   group: ${{ github.workflow }}-${{ github.ref }}

ci_check/change_files.py

@@ -40,7 +40,15 @@ def main():
                 "modifications": [
                     (
                         "torch.cuda.empty_cache()",
-                        "if use_cuda: torch.cuda.empty_cache()",
+                        "if use_cuda: torch.cuda.empty_cache()"
+                    ),
+                    (
+                        "init_process_group(backend='nccl')",
+                        "init_process_group(backend='gloo')"
+                    ),
+                    (
+                        "torch.cuda.set_device(int(os.environ['LOCAL_RANK']))",
+                        "# torch.cuda.set_device(int(os.environ['LOCAL_RANK']))"
                     )
                 ],
             }

ci_check/run.sh

@@ -4,7 +4,22 @@ current_directory=$(pwd)
 llmc=$(echo "$current_directory" | sed 's/\/ci_check$//')
 export PYTHONPATH=$llmc:$PYTHONPATH
 
+config=${llmc}/ci_check/awq_w4a16_fakequant_eval.yml 
+
+nnodes=1
+nproc_per_node=1
+MASTER_ADDR=127.0.0.1
+MASTER_PORT=$((10000 + RANDOM % 20000))
+
+RANDOM=$(python -c 'import uuid; print(uuid.uuid4())')
+task_id=$RANDOM
 
 cd ../scripts
 
-python -m llmc --config ../ci_check/awq_w4a16_fakequant_eval.yml 
+torchrun \
+    --nnodes $nnodes \
+    --nproc_per_node $nproc_per_node \
+    --rdzv_id $task_id \
+    --rdzv_backend c10d \
+    --rdzv_endpoint $MASTER_ADDR:$MASTER_PORT \
+    ${llmc}/llmc/__main__.py --config $config --task_id $task_id \