Fixes https://github.com/EleutherAI/lm-evaluation-harness/issues/1416 (#1418)

* Fixes https://github.com/EleutherAI/lm-evaluation-harness/issues/1416

Sets `do_sample = False` if `temperature == 0.0` and `do_sample = None`

* Update huggingface.py

* Update huggingface.py

making linter happy

lm_eval/models/huggingface.py

@@ -727,6 +727,11 @@ class HFLM(LM):
         # and we don't want a warning from HF
         generation_kwargs["temperature"] = generation_kwargs.get("temperature", 0.0)
         do_sample = generation_kwargs.get("do_sample", None)
+
+        # The temperature has to be a strictly positive float -- if it is 0.0, use greedy decoding strategies
+        if generation_kwargs.get("temperature") == 0.0 and do_sample is None:
+            generation_kwargs["do_sample"] = do_sample = False
+
         if do_sample is False and generation_kwargs.get("temperature") == 0.0:
             generation_kwargs.pop("temperature")
         # build stopping criteria