Merge remote-tracking branch 'upstream/big-refactor' into seq2seq-refactor

lm_eval/tasks/arc.py

@@ -16,7 +16,7 @@ from lm_eval import utils
 from lm_eval.prompts import get_prompt
 from lm_eval.api.task import MultipleChoiceTask
 
-from lm_eval.api.register import register_task, register_group
+from lm_eval.api.registry import register_task, register_group
 
 _CITATION = """
 @article{Clark2018ThinkYH,

lm_eval/tasks/gsm8k.py

@@ -24,7 +24,7 @@ from lm_eval.api.instance import Instance
 
 from lm_eval.prompts import get_prompt
 
-from lm_eval.api.register import register_task, register_group
+from lm_eval.api.registry import register_task, register_group
 
 _CITATION = """
 @misc{cobbe2021training,
@@ -92,7 +92,7 @@ class GradeSchoolMath8K(Task):
         return Instance(
             request_type=self.OUTPUT_TYPE,
             doc=doc,
-            arguments=(ctx, ["\n"]),
+            arguments=(ctx, ["\n\n"]),
             idx=0,
             **kwargs
         )
@@ -113,7 +113,7 @@ class GradeSchoolMath8K(Task):
         assert gold != INVALID_ANS, "No ground truth answer found in the document."
         # return self._extract_answer(completion) == gold
         # print(completion)
-        return completion == gold
+        return self._extract_answer(completion) == gold
 
     def process_results(self, doc, results):
         """Take a single document and the LM results and evaluates, returning a

lm_eval/tasks/gsm8k/README.md

+# GSM8k
+
+## Paper
+Training Verifiers to Solve Math Word Problems
+https://arxiv.org/abs/2110.14168
+
+State-of-the-art language models can match human performance on many tasks, but
+they still struggle to robustly perform multi-step mathematical reasoning. To
+diagnose the failures of current models and support research, we introduce GSM8K,
+a dataset of 8.5K high quality linguistically diverse grade school math word problems.
+We find that even the largest transformer models fail to achieve high test performance,
+despite the conceptual simplicity of this problem distribution.
+
+NOTE: See the official implementation of the task:
+    https://github.com/openai/grade-school-math/blob/master/grade_school_math/calculator.py
+for how to make use of the dataset's calculator annotations in your language
+model's sample/generation function.
+
+Homepage: https://github.com/openai/grade-school-math
+
+
+## Citation
+```
+@misc{cobbe2021training,
+      title={Training Verifiers to Solve Math Word Problems},
+      author={Karl Cobbe and Vineet Kosaraju and Mohammad Bavarian and Jacob Hilton and Reiichiro Nakano and Christopher Hesse and John Schulman},
+      year={2021},
+      eprint={2110.14168},
+      archivePrefix={arXiv},
+      primaryClass={cs.LG}
+}
+```

lm_eval/tasks/gsm8k/cot-gsm8k.yaml

-# "Training Verifiers to Solve Math Word Problems"
-# https://arxiv.org/abs/2110.14168
-
-# State-of-the-art language models can match human performance on many tasks, but
-# they still struggle to robustly perform multi-step mathematical reasoning. To
-# diagnose the failures of current models and support research, we introduce GSM8K,
-# a dataset of 8.5K high quality linguistically diverse grade school math word problems.
-# We find that even the largest transformer models fail to achieve high test performance,
-# despite the conceptual simplicity of this problem distribution.
-
-# NOTE: See the official implementation of the task:
-#     https://github.com/openai/grade-school-math/blob/master/grade_school_math/calculator.py
-# for how to make use of the dataset's calculator annotations in your language
-# model's sample/generation function.
-
-# Homepage: https://github.com/openai/grade-school-math
-
-# _CITATION = """
-# @misc{cobbe2021training,
-#       title={Training Verifiers to Solve Math Word Problems},
-#       author={Karl Cobbe and Vineet Kosaraju and Mohammad Bavarian and Jacob Hilton and Reiichiro Nakano and Christopher Hesse and John Schulman},
-#       year={2021},
-#       eprint={2110.14168},
-#       archivePrefix={arXiv},
-#       primaryClass={cs.LG}
-# }
-# """
-
-task: gsm8k_yaml
-dataset_path: gsm8k
-dataset_name: main
-training_split: train
-test_split: test
-use_prompt: "qa-basic:question-newline-answer"
-doc_to_target: "{{answer.split('### ')[-1]}}"
-metric_list:
-  - metric: exact_match
-    aggregation: mean
-    higher_is_better: true
-    ignore_case: true
-    ignore_punctuation: true
-delimiter: "\n"
-repeats: 4
-# filter_list:
-#   - name: "get-answer"
-#     filter:
-#       - function: "regex"
-#         regex_pattern: "#### (\-?[0-9\.\,]+)"

lm_eval/tasks/gsm8k/gsm8k-cot-self-consistency.yaml

+include: gsm8k-cot.yaml
+group:
+  - chain_of_thought
+  - self_consistency
+task: gsm8k_cot_self_consistency
+generation_kwargs:
+  until:
+    - "Q:"
+    - "\n\n"
+  do_sample: true
+  temperature: 0.2
+repeats: 8
+filter_list:
+  - name: "score-first" # pick only the first response, and report metrics on that
+    filter:
+      - function: "regex"
+        regex_pattern: "The answer is (\\-?[0-9\\.\\,]*[0-9]+)"
+      - function: "take_first"
+  - name: "maj@64"
+    filter:
+      - function: "regex"
+        regex_pattern: "The answer is (\\-?[0-9\\.\\,]*[0-9]+)"
+      - function: "majority_vote"
+      - function: "take_first"
+  - name: "maj@8" # get Maj@8 , via selecting the first 8 responses. Using a better estimator would be optimal.
+    filter:
+      - function: "take_first_k"
+        k: 8
+      - function: "regex"
+        regex_pattern: "The answer is (\\-?[0-9\\.\\,]*[0-9]+)"
+      - function: "majority_vote"
+      - function: "take_first"

lm_eval/tasks/gsm8k/gsm8k-cot.yaml

+group:
+  - chain_of_thought
+task: gsm8k_cot
+dataset_path: gsm8k
+dataset_name: main
+output_type: greedy_until
+test_split: test
+doc_to_text: "Q: There are 15 trees in the grove. Grove workers will plant trees in the grove today. After they are done, there will be 21 trees. How many trees did the grove workers plant today?\n\nA: There are 15 trees originally. Then there were 21 trees after some more were planted. So there must have been 21 - 15 = 6. The answer is 6.\n\n\
+Q: If there are 3 cars in the parking lot and 2 more cars arrive, how many cars are in the parking lot?\n\nA: There are originally 3 cars. 2 more cars arrive. 3 + 2 = 5. The answer is 5.\n\n\
+Q: Leah had 32 chocolates and her sister had 42. If they ate 35, how many pieces do they have left in total?\n\nA: Originally, Leah had 32 chocolates. Her sister had 42. So in total they had 32 + 42 = 74. After eating 35, they had 74 - 35 = 39. The answer is 39.\n\n\
+Q: Jason had 20 lollipops. He gave Denny some lollipops. Now Jason has 12 lollipops. How many lollipops did Jason give to Denny?\n\nA: Jason started with 20 lollipops. Then he had 12 after giving some to Denny. So he gave Denny 20 - 12 = 8. The answer is 8.\n\n\
+Q: Shawn has five toys. For Christmas, he got two toys each from his mom and dad. How many toys does he have now?\n\nA: Shawn started with 5 toys. If he got 2 toys each from his mom and dad, then that is 4 more toys. 5 + 4 = 9. The answer is 9.\n\n\
+Q: There were nine computers in the server room. Five more computers were installed each day, from monday to thursday. How many computers are now in the server room?\n\nA: There were originally 9 computers. For each of 4 days, 5 more computers were added. So 5 * 4 = 20 computers were added. 9 + 20 is 29. The answer is 29.\n\n\
+Q: Michael had 58 golf balls. On tuesday, he lost 23 golf balls. On wednesday, he lost 2 more. How many golf balls did he have at the end of wednesday?\n\nA: Michael started with 58 golf balls. After losing 23 on tuesday, he had 58 - 23 = 35. After losing 2 more, he had 35 - 2 = 33 golf balls. The answer is 33.\n\n\
+Q: Olivia has $23. She bought five bagels for $3 each. How much money does she have left?\n\nA: Olivia had 23 dollars. 5 bagels for 3 dollars each will be 5 x 3 = 15 dollars. So she has 23 - 15 dollars left. 23 - 15 is 8. The answer is 8.\n\n\
+Q: {{question}}\n\nA:"
+doc_to_target: "{{answer}}" #" {{answer.split('### ')[-1].rstrip()}}"
+gold_alias: "{{answer.split('### ')[-1].rstrip()}}" # this post-processes the reference that we'll score against
+metric_list:
+  - metric: exact_match
+    aggregation: mean
+    higher_is_better: true
+    ignore_case: true
+    ignore_punctuation: false
+    regexes_to_ignore:
+      - ","
+      - "\\$"
+delimiter: "\n\n"
+generation_kwargs:
+  until:
+    - "Q:"
+    - "\n\n"
+  do_sample: false
+  temperature: 0.0
+repeats: 1
+num_fewshot: 0
+filter_list:
+  - name: "get-answer"
+    filter:
+      - function: "regex"
+        regex_pattern: "The answer is (\\-?[0-9\\.\\,]+)"
+      - function: "take_first"

lm_eval/tasks/gsm8k/gsm8k.yaml

+task: gsm8k_yaml
+dataset_path: gsm8k
+dataset_name: main
+output_type: greedy_until
+training_split: train
+fewshot_split: train
+test_split: test
+doc_to_text: "Question: {{question}}\nAnswer:"
+doc_to_target: "{{answer}}" #" {{answer.split('### ')[-1].rstrip()}}"
+gold_alias: "{{answer.split('### ')[-1].rstrip()}}" # this post-processes the reference that we'll score against
+metric_list:
+  - metric: exact_match
+    aggregation: mean
+    higher_is_better: true
+    ignore_case: true
+    ignore_punctuation: false
+    regexes_to_ignore:
+      - ","
+      - "\\$"
+      - ".*### "
+delimiter: "\n\n"
+generation_kwargs:
+  until:
+    - "\n\n"
+    - "Question:"
+  do_sample: false
+  temperature: 0.0
+repeats: 2
+num_fewshot: 5
+# filter_list:
+#   - name: "get-answer"
+#     filter:
+#       - function: "regex"
+#         regex_pattern: "### (\\-?[0-9\\.\\,]+)"
+#       - function: "take_first"

lm_eval/tasks/lambada.py

@@ -16,7 +16,7 @@ from lm_eval.api.task import Task
 from lm_eval.api.instance import Instance
 from lm_eval.api.metrics import mean, perplexity
 
-from lm_eval.api.register import register_task, register_group
+from lm_eval.api.registry import register_task, register_group
 
 _CITATION = """
 @misc{

lm_eval/tasks/lambada/README.md

 # LAMBADA
 
 ### Paper
-The LAMBADA dataset: Word prediction requiring a broad discourse context 
+The LAMBADA dataset: Word prediction requiring a broad discourse context
 https://arxiv.org/pdf/1606.06031.pdf
 
 LAMBADA is a dataset to evaluate the capabilities of computational models for text
@@ -23,4 +23,4 @@ Homepage: https://zenodo.org/record/2630551#.X4Xzn5NKjUI
     publisher={Zenodo},
     year={2016},
     month={Aug}
-}
+}
\ No newline at end of file

lm_eval/tasks/pile.py

@@ -12,7 +12,7 @@ Homepage: https://pile.eleuther.ai/
 
 from lm_eval.api.task import PerplexityTask
 
-from lm_eval.api.register import register_task, register_group
+from lm_eval.api.registry import register_task, register_group
 
 _CITATION = """
 @article{pile,

lm_eval/tasks/pile/README.md

@@ -20,4 +20,4 @@ Homepage: https://pile.eleuther.ai/
   journal={arXiv preprint arXiv:2101.00027},
   year={2020}
 }
-```
+```
\ No newline at end of file

lm_eval/tasks/pile/pile_arxiv.yaml

@@ -19,4 +19,4 @@ metric_list:
     higher_is_better: false
   - metric: bits_per_byte
     aggregation: bits_per_byte
-    higher_is_better: false
+    higher_is_better: false
\ No newline at end of file

lm_eval/tasks/pile/pile_bookcorpus2.yaml

@@ -19,4 +19,4 @@ metric_list:
     higher_is_better: false
   - metric: bits_per_byte
     aggregation: bits_per_byte
-    higher_is_better: false
+    higher_is_better: false
\ No newline at end of file

lm_eval/tasks/pile/pile_books3.yaml

@@ -19,4 +19,4 @@ metric_list:
     higher_is_better: false
   - metric: bits_per_byte
     aggregation: bits_per_byte
-    higher_is_better: false
+    higher_is_better: false
\ No newline at end of file

lm_eval/tasks/pile/pile_dm-mathematics.yaml

@@ -19,4 +19,4 @@ metric_list:
     higher_is_better: false
   - metric: bits_per_byte
     aggregation: bits_per_byte
-    higher_is_better: false
+    higher_is_better: false
\ No newline at end of file

lm_eval/tasks/pile/pile_europarl.yaml

@@ -19,4 +19,4 @@ metric_list:
     higher_is_better: false
   - metric: bits_per_byte
     aggregation: bits_per_byte
-    higher_is_better: false
+    higher_is_better: false
\ No newline at end of file

lm_eval/tasks/pile/pile_freelaw.yaml

@@ -19,4 +19,4 @@ metric_list:
     higher_is_better: false
   - metric: bits_per_byte
     aggregation: bits_per_byte
-    higher_is_better: false
+    higher_is_better: false
\ No newline at end of file

lm_eval/tasks/pile/pile_github.yaml

@@ -19,4 +19,4 @@ metric_list:
     higher_is_better: false
   - metric: bits_per_byte
     aggregation: bits_per_byte
-    higher_is_better: false
+    higher_is_better: false
\ No newline at end of file

lm_eval/tasks/pile/pile_gutenberg.yaml

@@ -19,4 +19,4 @@ metric_list:
     higher_is_better: false
   - metric: bits_per_byte
     aggregation: bits_per_byte
-    higher_is_better: false
+    higher_is_better: false
\ No newline at end of file

lm_eval/tasks/pile/pile_hackernews.yaml

@@ -19,4 +19,4 @@ metric_list:
     higher_is_better: false
   - metric: bits_per_byte
     aggregation: bits_per_byte
-    higher_is_better: false
+    higher_is_better: false
\ No newline at end of file