Fix gemma3 workload execution failure

[shepark]

Fixed gemma3 workload execution failures.
Tested with gemma3 4b it model.
For your testing, you might need to use cache for model, and need to set no_proxy for client connection.

• With 1.21.0
  Server command:

VLLM_PROMPT_BS_BUCKET_MIN=1 VLLM_PROMPT_BS_BUCKET_STEP=1 VLLM_PROMPT_BS_BUCKET_MAX=1 \
VLLM_PROMPT_SEQ_BUCKET_MIN=384 VLLM_PROMPT_SEQ_BUCKET_MAX=384 \
VLLM_DECODE_BS_BUCKET_MIN=1 VLLM_DECODE_BS_BUCKET_MAX=1 \
VLLM_DECODE_BLOCK_BUCKET_MIN=512 VLLM_DECODE_BLOCK_BUCKET_MAX=512 \
python -m vllm.entrypoints.openai.api_server \
--model google/gemma-3-4b-it --max-num-batched-tokens 8192 --max-model-len 8192 --port 8000

• With 1.20.0
  There's an issue on hpu graph side, when it deals with pixel_values.
  So, you need to add PT_HPUGRAPH_DISABLE_TENSOR_CACHE=false

• Client command (from: https://rocm.blogs.amd.com/artificial-intelligence/deployingGemma-vllm/README.html)

for text and image input

curl -X POST "http://localhost:8000/v1/chat/completions" \
  -H "Content-Type: application/json" \
  --data '{
    "model": "google/gemma-3-4b-it",
    "messages": [
      {
        "role": "user",
        "content": [
          {
            "type": "text",
            "text": "Describe this image in one sentence."
          },
          {
            "type": "image_url",
            "image_url": {
              "url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
            }
          }
        ]
      }
    ]
  }' | jq

for text only

curl -X POST "http://localhost:8000/v1/chat/completions" \
  -H "Content-Type: application/json" \
  --data '{
    "model": "google/gemma-3-4b-it",
    "messages": [
      {
        "role": "user",
        "content": [
          {
            "type": "text",
            "text": "Describe the shape of an apple."
          }
        ]
      }
    ]
  }' | jq

[michalkuligowski]

/run-gaudi-tests

[michalkuligowski]

/run-gaudi-tests

[michalkuligowski]

/run-gaudi-tests

[michalkuligowski]

Please add hpu specific method for this and in the invocation call "current_platform.is_hpu()"

[shepark]

Please add hpu specific method for this and in the invocation call "current_platform.is_hpu()"

Thank you for the review. updated.

[michalkuligowski]

Precommit suite fails with the white space at the end of this line