merge main and edit configs/bash script to interate inside subfolders

vllm-project · Dec 9, 2024 · ff8e39d · ff8e39d
1 parent 7de0afb
commit ff8e39d
Show file tree

Hide file tree

Showing 17 changed files with 19 additions and 25 deletions.
diff --git a/...e/vLLM/configs/fp8_dynamic_per_token.yaml → ...LM/configs/fp8/fp8_dynamic_per_token.yaml b/...e/vLLM/configs/fp8_dynamic_per_token.yaml → ...LM/configs/fp8/fp8_dynamic_per_token.yaml
diff --git a/...e/vLLM/configs/fp8_static_per_tensor.yaml → ...LM/configs/fp8/fp8_static_per_tensor.yaml b/...e/vLLM/configs/fp8_static_per_tensor.yaml → ...LM/configs/fp8/fp8_static_per_tensor.yaml
diff --git a/...vLLM/configs/fp8_weight_only_channel.yaml → .../configs/fp8/fp8_weight_only_channel.yaml b/...vLLM/configs/fp8_weight_only_channel.yaml → .../configs/fp8/fp8_weight_only_channel.yaml
diff --git a/.../vLLM/configs/fp8_weight_only_tensor.yaml → ...M/configs/fp8/fp8_weight_only_tensor.yaml b/.../vLLM/configs/fp8_weight_only_tensor.yaml → ...M/configs/fp8/fp8_weight_only_tensor.yaml
diff --git a/...channel_weight_static_per_tensor_act.yaml → ...channel_weight_static_per_tensor_act.yaml b/...channel_weight_static_per_tensor_act.yaml → ...channel_weight_static_per_tensor_act.yaml
diff --git a/.../vLLM/configs/int8_dynamic_per_token.yaml → .../configs/int8/int8_dynamic_per_token.yaml b/.../vLLM/configs/int8_dynamic_per_token.yaml → .../configs/int8/int8_dynamic_per_token.yaml
diff --git a/..._tensor_weight_static_per_tensor_act.yaml → ..._tensor_weight_static_per_tensor_act.yaml b/..._tensor_weight_static_per_tensor_act.yaml → ..._tensor_weight_static_per_tensor_act.yaml
diff --git a/...LLM/configs/w4a16_2of4_channel_quant.yaml → ...nfigs/w4a16/w4a16_2of4_channel_quant.yaml b/...LLM/configs/w4a16_2of4_channel_quant.yaml → ...nfigs/w4a16/w4a16_2of4_channel_quant.yaml
diff --git a/...LLM/configs/w4a16_2of4_grouped_quant.yaml → ...nfigs/w4a16/w4a16_2of4_grouped_quant.yaml b/...LLM/configs/w4a16_2of4_grouped_quant.yaml → ...nfigs/w4a16/w4a16_2of4_grouped_quant.yaml
diff --git a/...2e/vLLM/configs/w4a16_actorder_group.yaml → ...M/configs/w4a16/w4a16_actorder_group.yaml b/...2e/vLLM/configs/w4a16_actorder_group.yaml → ...M/configs/w4a16/w4a16_actorder_group.yaml
diff --git a/...e/vLLM/configs/w4a16_actorder_weight.yaml → .../configs/w4a16/w4a16_actorder_weight.yaml b/...e/vLLM/configs/w4a16_actorder_weight.yaml → .../configs/w4a16/w4a16_actorder_weight.yaml
diff --git a/...e2e/vLLM/configs/w4a16_channel_quant.yaml → ...LM/configs/w4a16/w4a16_channel_quant.yaml b/...e2e/vLLM/configs/w4a16_channel_quant.yaml → ...LM/configs/w4a16/w4a16_channel_quant.yaml
diff --git a/...e2e/vLLM/configs/w4a16_grouped_quant.yaml → ...LM/configs/w4a16/w4a16_grouped_quant.yaml b/...e2e/vLLM/configs/w4a16_grouped_quant.yaml → ...LM/configs/w4a16/w4a16_grouped_quant.yaml
diff --git a/...e2e/vLLM/configs/w8a16_channel_quant.yaml → ...LM/configs/w8a16/w8a16_channel_quant.yaml b/...e2e/vLLM/configs/w8a16_channel_quant.yaml → ...LM/configs/w8a16/w8a16_channel_quant.yaml
diff --git a/...e2e/vLLM/configs/w8a16_grouped_quant.yaml → ...LM/configs/w8a16/w8a16_grouped_quant.yaml b/...e2e/vLLM/configs/w8a16_grouped_quant.yaml → ...LM/configs/w8a16/w8a16_grouped_quant.yaml
diff --git a/tests/e2e/vLLM/run_tests.sh b/tests/e2e/vLLM/run_tests.sh
@@ -5,26 +5,30 @@ SUCCESS=0
 # Parse list of configs.
 MODEL_CONFIGS="$PWD/tests/e2e/vLLM/configs"
 
-for MODEL_CONFIG in "$MODEL_CONFIGS"/*
-do
+# Function to process a file
+process_model_config() {
+    local MODEL_CONFIG=$1
     LOCAL_SUCCESS=0
-
+    
     echo "=== RUNNING MODEL: $MODEL_CONFIG ==="
-
-    export TEST_DATA_FILE="$MODEL_CONFIG"
-    pytest \
-        --capture=tee-sys \
-        --junitxml="test-results/e2e-$(date +%s).xml" \
-        "$PWD/tests/e2e/vLLM/test_vllm.py" || LOCAL_SUCCESS=$?
-
+    export TEST_DATA_FILE=${MODEL_CONFIG}
+    pytest -s $PWD/tests/e2e/vLLM/test_vllm.py || LOCAL_SUCCESS=$?
     if [[ $LOCAL_SUCCESS == 0 ]]; then
-        echo "=== PASSED MODEL: $MODEL_CONFIG ==="
+        echo "=== PASSED MODEL: ${MODEL_CONFIG} ==="
     else
-        echo "=== FAILED MODEL: $MODEL_CONFIG ==="
+        echo "=== FAILED MODEL: ${MODEL_CONFIG} ==="
     fi
-
     SUCCESS=$((SUCCESS + LOCAL_SUCCESS))
+}
 
+# Use find to locate all files in MODEL_CONFIGS and process them
+find "$MODEL_CONFIGS" -type f | while read -r MODEL_CONFIG
+do
+    process_model_config "$MODEL_CONFIG"
 done
 
-exit "$SUCCESS"
+if [ "${SUCCESS}" -eq "0" ]; then
+    exit 0
+else
+    exit 1
+fi
diff --git a/tests/e2e/vLLM/test_vllm.py b/tests/e2e/vLLM/test_vllm.py
@@ -19,20 +19,10 @@
     vllm_installed = False
     logger.warning("vllm is not installed. This test will be skipped")
 
-# Defines the file paths to the directories containing the test configs
-# for each of the quantization schemes
-WNA16 = "tests/e2e/vLLM/configs/WNA16"
-FP8 = "tests/e2e/vLLM/configs/FP8"
-INT8 = "tests/e2e/vLLM/configs/INT8"
-ACTORDER = "tests/e2e/vLLM/configs/actorder"
-KV_CACHE = "tests/e2e/vLLM/configs/kv_cache"
-WNA16_2of4 = "tests/e2e/vLLM/configs/WNA16_2of4"
-
-CONFIGS = [WNA16, FP8, INT8, ACTORDER, WNA16_2of4, KV_CACHE]
-
 HF_MODEL_HUB_NAME = "nm-testing"
 TEST_DATA_FILE = os.environ.get("TEST_DATA_FILE", None)
 
+
 # Will run each test case in its own process through run_tests.sh
 # emulating vLLM CI testing
 @requires_gpu_count(1)