[VLLM] Allows for max tokens to be set in model config file (#547)

* commit

* commit

* Update src/lighteval/main_vllm.py

* commit

* change doc

* change doc

* change doc

* allow max new token to be set in model config file

Author: NathanHB

examples/model_configs/vllm_model_config.yaml

@@ -10,5 +10,5 @@ model:
     top_k: -1
     min_p: 0.0
     top_p: 0.9
-    max_new_tokens: 100
+    max_new_tokens: 256
     stop_tokens: ["<EOS>", "<PAD>"]

src/lighteval/models/vllm/vllm_model.py

@@ -316,7 +316,9 @@ def _generate(
         sampling_params = self.sampling_params.clone() or SamplingParams()
         if generate:
             sampling_params.n = num_samples
-            sampling_params.max_tokens = max_new_tokens
+            sampling_params.max_tokens = (
+                max_new_tokens if sampling_params.max_tokens is None else sampling_params.max_tokens
+            )
             sampling_params.stop = stop_tokens
             sampling_params.logprobs = 1 if returns_logits else 0