Update interface

lm_eval/base.py

@@ -6,7 +6,9 @@ import collections
 class LM(abc.ABC):
     @abc.abstractmethod
     def loglikelihood(self, requests):
-        """Compute log-likelihood of generating a continuation from a context
+        """Compute log-likelihood of generating a continuation from a context.
+        Downstream tasks should attempt to use loglikelihood instead of other 
+        LM calls whenever possible.
 
         :param requests: list
             A list of pairs (context, continuation)
@@ -35,7 +37,7 @@ class LM(abc.ABC):
                 Context string
             until: str
                 The string sequence to generate until. This string sequence may 
-                span across msultiple tokens, or may be part of one token.
+                span across multiple tokens, or may be part of one token.
         :return: list
             A list of strings continuation
             continuation: str
@@ -112,7 +114,13 @@ class Dataset(abc.ABC):
         pass
 
     @abc.abstractmethod
-    def construct_requests(self, doc, nshot=0, prompt=False):
+    def construct_requests(self, ctx):
+        """ Uses RequestFactory to construct Requests and returns an iterable of 
+        Requests which will be sent to the LM.
+
+        :param ctx: str
+            The context string, generated by fewshot_context.
+        """
         pass
     
     @abc.abstractmethod
@@ -134,6 +142,11 @@ class Dataset(abc.ABC):
             aggregates them into one float. This should be the same for all 
             submetrics of the same name; if it differs, an error should be 
             raised.
+
+        :param doc:
+            The document as returned from training_docs, validation_docs, or test_docs.
+        :param results:
+            The results of the requests created in construct_requests.
         """
         pass