[Misc] Update gptq_marlin to use new vLLMParameters (#7281)

tests/weight_loading/models.txt

@@ -0,0 +1,15 @@
+gptq_marlin, robertgshaw2/zephyr-7b-beta-channelwise-gptq, main
+gptq_marlin, TheBloke/Llama-2-7B-GPTQ, main
+gptq_marlin, TheBloke/TinyLlama-1.1B-Chat-v1.0-GPTQ, main
+gptq_marlin, TheBloke/TinyLlama-1.1B-Chat-v1.0-GPTQ, gptq-8bit--1g-actorder_True
+gptq_marlin, TheBloke/TinyLlama-1.1B-Chat-v1.0-GPTQ, gptq-8bit-32g-actorder_True
+gptq_marlin, TechxGenus/gemma-1.1-2b-it-GPTQ, main
+compressed-tensors, nm-testing/tinyllama-oneshot-w8w8-test-static-shape-change, main
+compressed-tensors, nm-testing/tinyllama-oneshot-w8-channel-a8-tensor, main
+compressed-tensors, nm-testing/tinyllama-oneshot-w8a8-dynamic-token-v2, main
+compressed-tensors, nm-testing/tinyllama-oneshot-w8a8-channel-dynamic-token-v2, main
+compressed-tensors, nm-testing/tinyllama-oneshot-w4a16-group128-v2, main
+compressed-tensors, nm-testing/tinyllama-oneshot-w8a16-per-channel, main
+compressed-tensors, nm-testing/Meta-Llama-3-8B-FP8-compressed-tensors-test, main
+compressed-tensors, nm-testing/Phi-3-mini-128k-instruct-FP8, main
+compressed-tensors, neuralmagic/Phi-3-medium-128k-instruct-quantized.w4a16, main

tests/weight_loading/run_model_weight_loading_test.sh

@@ -0,0 +1,32 @@
+#!/bin/bash
+SUCCESS=0
+
+IFS=$'\n' read -d '' -r -a MODEL_CONFIGS < "weight_loading/models.txt"
+
+for MODEL_CONFIG in "${MODEL_CONFIGS[@]}"
+do
+    LOCAL_SUCCESS=0
+    IFS=', ' read -r -a array <<< "$MODEL_CONFIG"
+    
+    echo "=== RUNNING MODEL: $MODEL_CONFIG ==="
+
+    export QUANTIZATION=${array[0]}
+    export MODEL_NAME=${array[1]}
+    export REVISION=${array[2]}
+    pytest -s weight_loading/test_weight_loading.py || LOCAL_SUCCESS=$?
+
+    if [[ $LOCAL_SUCCESS == 0 ]]; then
+        echo "=== PASSED MODEL: ${MODEL_CONFIG} ==="
+    else
+        echo "=== FAILED MODEL: ${MODEL_CONFIG} ==="
+    fi
+
+    SUCCESS=$((SUCCESS + LOCAL_SUCCESS))
+
+done
+
+if [ "${SUCCESS}" -eq "0" ]; then
+    exit 0
+else
+    exit 1
+fi

tests/weight_loading/test_weight_loading.py

@@ -0,0 +1,20 @@
+import os
+
+MAX_MODEL_LEN = 1024
+MODEL_NAME = os.environ.get("MODEL_NAME",
+                            "robertgshaw2/zephyr-7b-beta-channelwise-gptq")
+REVISION = os.environ.get("REVISION", "main")
+QUANTIZATION = os.environ.get("QUANTIZATION", "gptq_marlin")
+
+
+def test_weight_loading(vllm_runner):
+    with vllm_runner(model_name=MODEL_NAME,
+                     revision=REVISION,
+                     dtype="auto",
+                     quantization=QUANTIZATION,
+                     max_model_len=MAX_MODEL_LEN,
+                     tensor_parallel_size=2) as model:
+
+        output = model.generate_greedy("Hello world!", max_tokens=20)
+        print(output)
+        assert output