fix: dtype kwarg (torch_dtype deprecated), vLLM max_model_len=4096

- grpo_train: torch_dtype → dtype (transformers 4.45+ renamed it)
- secalign_agent: add max_model_len=4096 to vLLM LLM() init; vLLM was
trying to allocate KV cache for full 131072-token context (16 GiB)
but only 5.4 GiB is free after Qwen bf16 occupies ~3 GiB. With
max_model_len=4096 the KV cache drops to ~500 MB — fits easily.
SecAlign agent tasks never need more than 4096 tokens.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

env/defenses/secalign_agent.py

@@ -84,6 +84,7 @@ class SecAlignAgent:
                 max_lora_rank=64,
                 trust_remote_code=True,
                 gpu_memory_utilization=self._gpu_memory_utilization,
+                max_model_len=4096,
             )
             self._sampling_params = SamplingParams(
                 temperature=0.0,

train/grpo_train.py

@@ -115,7 +115,7 @@ def _load_model_and_tokenizer(model_id: str, seed: int):
 
     model = AutoModelForCausalLM.from_pretrained(
         model_id,
-        torch_dtype=torch.bfloat16,
+        dtype=torch.bfloat16,
         device_map="auto",
         trust_remote_code=True,
     )