[CB][FP8] throw error for batch size 1 (#467)

### [CB][FP8] throw error for batch size 1

intermediate solution until #466 can be merged safely.

Signed-off-by: Yannick Schnider <[email protected]>

Author: yannicks1

vllm_spyre/platform.py

@@ -146,6 +146,10 @@ def check_and_update_config(cls, vllm_config: VllmConfig) -> None:
             if envs_spyre.VLLM_SPYRE_ENABLE_PROMPT_LOGPROBS:
                 raise ValueError("Prompt logprobs not supported with " \
                 "continuous batching")
+            if (vllm_config.model_config.quantization
+                    and vllm_config.scheduler_config.max_num_seqs == 1):
+                raise ValueError(
+                    "Batch size 1 not supported for fp8 continuous batching.")
         else:
             # Static batching or embedding model.
             # Override --max-num-seqs to the biggest warmup batch size