added benchmark configs

lm_eval/tasks/benchmarks/pythia.yaml

+group: pythia
+task:
+  - lambada_openai
+  - wikitext
+  - piqa
+  - sciq
+  - wsc
+  - winogrande
+  - arc_challenge
+  - arc_easy
+  - logiqa
+  - blimp_*
+  - hendrycksTest*

lm_eval/tasks/benchmarks/t0_eval.yaml

+group:
+  - t0_eval
+task:
+  - dataset_path: super_glue # Coreference Resolution
+    dataset_name: wsc.fixed
+    use_prompt: promptsource
+  - dataset_path: winogrande # Coreference Resolution
+    dataset_name: winogrande_xl
+    use_prompt: promptsource
+  - dataset_path: super_glue # Natural Language Inference
+    dataset_name: cb
+    use_prompt: promptsource
+  - dataset_path: super_glue # Natural Language Inference
+    dataset_name: rte
+    use_prompt: promptsource
+  - dataset_path: anli # Natural Language Inference
+    use_prompt: promptsource
+  - dataset_path: super_glue # Sentence Completion
+    dataset_name: copa
+    use_prompt: promptsource
+  - dataset_path: hellaswag # Natural Language Inference
+    use_prompt: promptsource
+  - dataset_path: super_glue  # Word Sense Disambiguation
+    dataset_name: wic
+    use_prompt: promptsource

lm_eval/tasks/benchmarks/t0_eval.yml

-- dataset_path: "super_glue" # Coreference Resolution
-  dataset_name: "wsc.fixed"
-- dataset_path: "winogrande" # Coreference Resolution
-  dataset_name: "winogrande_xl"
-- dataset_path: "super_glue" # Natural Language Inference
-  dataset_name: "cb"
-- dataset_path: "super_glue" # Natural Language Inference
-  dataset_name: "rte"
-- dataset_path: "anli" # Natural Language Inference
-  dataset_name: null
-- dataset_path: "super_glue" # Sentence Completion
-  dataset_name: "copa"
-- dataset_path: "hellaswag" # Natural Language Inference
-  dataset_name: null
-- dataset_path: "super_glue"  # Word Sense Disambiguation
-  dataset_name: "wic"