merge update

docs/README.md

@@ -4,6 +4,7 @@ Welcome to the docs for the LM Evaluation Harness!
 
 ## Table of Contents
 
+* To learn about the public interface of the library, as well as how to evaluate via the commandline or as integrated into an external library, see the [Interface](https://github.com/EleutherAI/lm-evaluation-harness/blob/big-refactor/docs/user_guide.md)
 * To learn how to add a new library, API, or model type to the library, as well as a quick explainer on the types of ways to evaluate an LM, see the [Model Guide](https://github.com/EleutherAI/lm-evaluation-harness/blob/big-refactor/docs/model_guide.md).
 * For a crash course on adding new tasks to the library, see our [New Task Guide](https://github.com/EleutherAI/lm-evaluation-harness/blob/big-refactor/docs/new_task_guide.md).
 * To learn more about pushing the limits of task configuration that the Eval Harness supports, see the [Advanced Task Guide](https://github.com/EleutherAI/lm-evaluation-harness/blob/big-refactor/docs/advanced_task_guide.md).

docs/interface.md

+# User Guide
+
+This document details the interface exposed by `lm-eval` and provides details on what flags are available to users.
+
+## Command-line Interface
+
+A majority of users run the library by cloning it from Github and running the `main.py` script.
+
+Equivalently, running the library can be done via the `lm-eval` entrypoint at the command line.
+
+This mode supports a number of command-line arguments, the details of which can be also be seen via running with `-h` or `--help`:
+
+* `--model` : Selects which model type or provider is evaluated. Must be a string corresponding to the name of the model type/provider being used. See [the main README](https://github.com/EleutherAI/lm-evaluation-harness/tree/big-refactor#commercial-apis) for a full list of enabled model names and supported libraries or APIs.
+
+* `--model_args` : Controls parameters passed to the model constructor. Accepts a string containing comma-separated keyword arguments to the model class of the format `"arg1=val1,arg2=val2,..."`, such as, for example `--model_args pretrained=EleutherAI/pythia-160m,dtype=float32`. For a full list of what keyword arguments, see the initialization of the `lm_eval.api.model.LM` subclass, e.g. [`HFLM`](https://github.com/EleutherAI/lm-evaluation-harness/blob/365fcda9b85bbb6e0572d91976b8daf409164500/lm_eval/models/huggingface.py#L66)
+
+* `--tasks` : Determines which tasks or task groups are evaluated. Accepts a comma-separated list of task names or task group names. Must be solely comprised of valid tasks/groups.
+
+* `--num_fewshot` : Sets the number of few-shot examples to place in context. Must be an integer.
+
+* `--batch_size` : Sets the batch size used for evaluation. Can be a positive integer or `"auto"` to automatically select the largest batch size that will fit in memory, speeding up evaluation. One can pass `--batch_size auto:N` to re-select the maximum batch size `N` times during evaluation. This can help accelerate evaluation further, since `lm-eval` sorts documents in descending order of context length.
+
+* `--max_batch_size` : Sets the maximum batch size to try to fit in memory, if `--batch_size auto` is passed.
+
+* `--device` : Sets which device to place the model onto. Must be a string, for example, `"cuda", "cuda:0", "cpu", "mps"`. Defaults to "cuda", and can be ignored if running multi-GPU or running a non-local model type.
+
+* `--output_path` : A string of the form `dir/file.jsonl` or `dir/`. Provides a path where high-level results will be saved, either into the file named or into the directory named. If `--log_samples` is passed as well, then per-document outputs and metrics will be saved into the directory as well.
+
+* `--log_samples` : If this flag is passed, then the model's outputs, and the text fed into the model, will be saved at per-document granularity. Must be used with `--output_path`.
+
+* `--limit` : Accepts an integer, or a float between 0.0 and 1.0 . If passed, will limit the number of documents to evaluate to the first X documents (if an integer) per task or first X% of documents per task. Useful for debugging, especially on costly API models.
+
+* `--use_cache` : Should be a path where a sqlite db file can be written to. Takes a string of format `/path/to/sqlite_cache_` in order to create a cache db at `/path/to/sqlite_cache_rank{i}.db` for each process (0-NUM_GPUS). This allows results of prior runs to be cached, so that there is no need to re-run results in order to re-score or re-run a given (model, task) pair again.
+
+* `--decontamination_ngrams_path` : Deprecated, see (this commit)[https://github.com/EleutherAI/lm-evaluation-harness/commit/00209e10f6e27edf5d766145afaf894079b5fe10] or older for a working decontamination-checker tool.
+
+* `--check_integrity` : If this flag is used, the library tests for each task selected are run to confirm task integrity.
+
+* `--write_out` : Used for diagnostic purposes to observe the format of task documents passed to a model. If this flag is used, then prints the prompt and gold target string for the first document of each task.
+
+* `--show_config` : If used, prints the full `lm_eval.api.task.TaskConfig` contents (non-default settings the task YAML file) for each task which was run, at the completion of an evaluation. Useful for when one is modifying a task's configuration YAML locally to transmit the exact configurations used for debugging or for reproducibility purposes.
+
+* `--include_path` : Accepts a path to a folder. If passed, then all YAML files containing `lm-eval`` compatible task configurations will be added to the task registry as available tasks. Used for when one is writing config files for their own task in a folder other than `lm_eval/tasks/`
+
+## External Library Usage
+
+We also support using the library's external API for use within model training loops or other scripts.
+
+`lm_eval` supplies two functions for external import and use: `lm_eval.evaluate()` and `lm_eval.simple_evaluate()`.
+
+
+`simple_evaluate()` can be used by simply creating an `lm_eval.api.model.LM` subclass that implements the methods described in the [Model Guide](https://github.com/EleutherAI/lm-evaluation-harness/tree/big-refactor/docs/model_guide.md), and wrapping your custom model in that class as follows:
+
+```python
+import lm_eval
+...
+
+my_model = initialize_my_model() # create your model (could be running finetuning with some custom modeling code)
+...
+lm_obj = Your_LM(model=my_model, batch_size=16) # instantiate an LM subclass that takes your initialized model and can run `Your_LM.loglikelihood()`, `Your_LM.loglikelihood_rolling()`, `Your_LM.greedy_until()`
+
+results = lm_eval.simple_evaluate( # call simple_evaluate
+    model=lm_obj,
+    tasks=["taskname1", "taskname2"],
+    num_fewshot=0,
+    ...
+)
+```
+
+
+See https://github.com/EleutherAI/lm-evaluation-harness/blob/365fcda9b85bbb6e0572d91976b8daf409164500/lm_eval/evaluator.py#L35 for a full description of all arguments available. All keyword arguments to simple_evaluate share the same role as the command-line flags described previously.
+
+Additionally, the `evaluate()` function offers the core evaluation functionality provided by the library, but without some of the special handling and simplification + abstraction provided by `simple_evaluate()`.
+
+See https://github.com/EleutherAI/lm-evaluation-harness/blob/365fcda9b85bbb6e0572d91976b8daf409164500/lm_eval/evaluator.py#L173 for more details.
+
+As a brief example usage of `evaluate()`:
+```python
+import lm_eval
+
+from my_tasks import MyTask1 # suppose you've defined a custom lm_eval.api.Task subclass in your own external codebase
+...
+
+my_model = initialize_my_model() # create your model (could be running finetuning with some custom modeling code)
+...
+lm_obj = Your_LM(model=my_model, batch_size=16) # instantiate an LM subclass that takes your initialized model and can run `Your_LM.loglikelihood()`, `Your_LM.loglikelihood_rolling()`, `Your_LM.greedy_until()`
+
+
+
+def evaluate(
+    lm=lm_obj,
+    task_dict={"mytask1": MyTask1},
+    ...
+):
+```

lm_eval/__init__.py

+from .evaluator import evaluate, simple_evaluate

lm_eval/api/task.py

@@ -250,6 +250,11 @@ class Task(abc.ABC):
             download_mode=download_mode,
         )
 
+    @property
+    def config(self):
+        """Returns the TaskConfig associated with this class."""
+        return self._config
+
     @abc.abstractmethod
     def has_training_docs(self):
         """Whether the task has a training set"""
@@ -352,7 +357,7 @@ class Task(abc.ABC):
             ), f"Task dataset (path={self.DATASET_PATH}, name={self.DATASET_NAME}) must have valid or test docs!"
 
         eval_logger.info(
-            f"Building contexts for task '{self._config.task}' on rank {rank}..."
+            f"Building contexts for task '{self.config.task}' on rank {rank}..."
         )
 
         instances = []
@@ -362,14 +367,14 @@ class Task(abc.ABC):
             # sample fewshot context #TODO: need to offset doc_id by rank now!
             fewshot_ctx = self.fewshot_context(
                 doc,
-                self._config.num_fewshot,
+                self.config.num_fewshot,
             )
 
-            # TODO: we should override self._config.repeats if doing greedy gen so users don't waste time+compute
+            # TODO: we should override self.config.repeats if doing greedy gen so users don't waste time+compute
             inst = self.construct_requests(
                 doc=doc,
                 ctx=fewshot_ctx,
-                metadata=(self._config["task"], doc_id, self._config.repeats),
+                metadata=(self.config["task"], doc_id, self.config.repeats),
             )
 
             if not isinstance(inst, list):
@@ -457,9 +462,9 @@ class Task(abc.ABC):
 
         if num_fewshot == 0:
             # always prepend the (possibly empty) task description
-            labeled_examples = self._config.description
+            labeled_examples = self.config.description
         else:
-            labeled_examples = self._config.description + self.sampler.get_context(
+            labeled_examples = self.config.description + self.sampler.get_context(
                 doc, num_fewshot
             )
 
@@ -469,7 +474,7 @@ class Task(abc.ABC):
         elif type(example) == list:
             return [labeled_examples + ex for ex in example]
         elif type(example) == int:
-            if self._config.doc_to_choice is not None:
+            if self.config.doc_to_choice is not None:
                 choices = self.doc_to_choice(doc)
                 return labeled_examples + choices[example]
             else:
@@ -491,7 +496,7 @@ class Task(abc.ABC):
         """
         # TODO: this should only return the overrides applied to a non-YAML task's configuration.
         # (num_fewshot)
-        return self._config.to_dict()
+        return self.config.to_dict()
 
 
 class ConfigurableTask(Task):
@@ -506,35 +511,35 @@ class ConfigurableTask(Task):
         self._config = self.CONFIG
 
         # Use new configurations if there was no preconfiguration
-        if self._config is None:
+        if self.config is None:
             self._config = TaskConfig(**config)
         # Overwrite configs
         else:
             if config is not None:
                 self._config.__dict__.update(config)
 
-        if self._config is None:
+        if self.config is None:
             raise ValueError(
                 "Must pass a config to ConfigurableTask, either in cls.CONFIG or `config` kwarg"
             )
 
-        if self._config.output_type is not None:
-            assert self._config.output_type in ALL_OUTPUT_TYPES
-            self.OUTPUT_TYPE = self._config.output_type
+        if self.config.output_type is not None:
+            assert self.config.output_type in ALL_OUTPUT_TYPES
+            self.OUTPUT_TYPE = self.config.output_type
 
-        if self._config.dataset_path is not None:
-            self.DATASET_PATH = self._config.dataset_path
+        if self.config.dataset_path is not None:
+            self.DATASET_PATH = self.config.dataset_path
 
-        if self._config.dataset_name is not None:
-            self.DATASET_NAME = self._config.dataset_name
+        if self.config.dataset_name is not None:
+            self.DATASET_NAME = self.config.dataset_name
 
         self._metric_fn_list = {}
         self._metric_fn_kwargs = {}
         self._aggregation_list = {}
         self._higher_is_better = {}
 
-        _metric_list = DEFAULT_METRIC_REGISTRY[self._config.output_type]
-        if self._config.metric_list is None:
+        _metric_list = DEFAULT_METRIC_REGISTRY[self.config.output_type]
+        if self.config.metric_list is None:
             # TODO: handle this in TaskConfig.__post_init__ ?
             for metric_name in _metric_list:
                 self._metric_fn_list[metric_name] = get_metric(metric_name)
@@ -543,7 +548,7 @@ class ConfigurableTask(Task):
                 )
                 self._higher_is_better[metric_name] = is_higher_better(metric_name)
         else:
-            for metric_config in self._config.metric_list:
+            for metric_config in self.config.metric_list:
                 assert "metric" in metric_config
                 metric_name = metric_config["metric"]
                 kwargs = {
@@ -552,7 +557,7 @@ class ConfigurableTask(Task):
                     if key not in ["metric", "aggregation", "higher_is_better"]
                 }
 
-                if self._config.process_results is not None:
+                if self.config.process_results is not None:
                     self._metric_fn_list[metric_name] = None
                     self._metric_fn_kwargs[metric_name] = {}
                 elif callable(metric_name):
@@ -594,13 +599,13 @@ class ConfigurableTask(Task):
                     )
                     self._higher_is_better[metric_name] = is_higher_better(metric_name)
 
-        self.download(self._config.dataset_kwargs)
+        self.download(self.config.dataset_kwargs)
         self._training_docs = None
         self._fewshot_docs = None
 
-        if self._config.filter_list is not None:
+        if self.config.filter_list is not None:
             self._filters = []
-            for filter_config in self._config.filter_list:
+            for filter_config in self.config.filter_list:
                 for filter_pipeline in filter_config:
                     filter_name = filter_config["name"]
                     filter_functions = filter_config["filter"]
@@ -615,10 +620,10 @@ class ConfigurableTask(Task):
         else:
             self._filters = [build_filter_ensemble("none", [["take_first", None]])]
 
-        if self._config.use_prompt is not None:
-            eval_logger.info(f"loading prompt {self._config.use_prompt}")
+        if self.config.use_prompt is not None:
+            eval_logger.info(f"loading prompt {self.config.use_prompt}")
             self.prompt = get_prompt(
-                self._config.use_prompt, self.DATASET_PATH, self.DATASET_NAME
+                self.config.use_prompt, self.DATASET_PATH, self.DATASET_NAME
             )
         else:
             self.prompt = None
@@ -645,7 +650,7 @@ class ConfigurableTask(Task):
         test_text = self.doc_to_text(test_doc)
         test_target = self.doc_to_target(test_doc)
 
-        if self._config.doc_to_choice is not None:
+        if self.config.doc_to_choice is not None:
             test_choice = self.doc_to_choice(test_doc)
             if type(test_choice) is not list:
                 eval_logger.error("doc_to_choice must return list")
@@ -673,7 +678,7 @@ class ConfigurableTask(Task):
         for choice in check_choices:
             choice_has_whitespace = True if " " in choice else False
             delimiter_has_whitespace = (
-                True if " " in self._config.target_delimiter else False
+                True if " " in self.config.target_delimiter else False
             )
 
             if delimiter_has_whitespace and choice_has_whitespace:
@@ -693,52 +698,52 @@ class ConfigurableTask(Task):
         )
 
     def has_training_docs(self) -> bool:
-        if self._config.training_split is not None:
+        if self.config.training_split is not None:
             return True
         else:
             return False
 
     def has_validation_docs(self) -> bool:
-        if self._config.validation_split is not None:
+        if self.config.validation_split is not None:
             return True
         else:
             return False
 
     def has_test_docs(self) -> bool:
-        if self._config.test_split is not None:
+        if self.config.test_split is not None:
             return True
         else:
             return False
 
     def training_docs(self) -> datasets.Dataset:
         if self.has_training_docs():
-            if self._config.process_docs is not None:
-                return self._config.process_docs(
-                    self.dataset[self._config.training_split]
+            if self.config.process_docs is not None:
+                return self.config.process_docs(
+                    self.dataset[self.config.training_split]
                 )
-            return self.dataset[self._config.training_split]
+            return self.dataset[self.config.training_split]
 
     def validation_docs(self) -> datasets.Dataset:
         if self.has_validation_docs():
-            if self._config.process_docs is not None:
-                return self._config.process_docs(
-                    self.dataset[self._config.validation_split]
+            if self.config.process_docs is not None:
+                return self.config.process_docs(
+                    self.dataset[self.config.validation_split]
                 )
-            return self.dataset[self._config.validation_split]
+            return self.dataset[self.config.validation_split]
 
     def test_docs(self) -> datasets.Dataset:
         if self.has_test_docs():
-            if self._config.process_docs is not None:
-                return self._config.process_docs(self.dataset[self._config.test_split])
-            return self.dataset[self._config.test_split]
+            if self.config.process_docs is not None:
+                return self.config.process_docs(self.dataset[self.config.test_split])
+            return self.dataset[self.config.test_split]
 
     def fewshot_docs(self):
-        if self._config.fewshot_split is not None:
-            return self.dataset[self._config.fewshot_split]
+        if self.config.fewshot_split is not None:
+            return self.dataset[self.config.fewshot_split]
         else:
-            if self._config.num_fewshot > 0:
+            if self.config.num_fewshot > 0:
                 eval_logger.warning(
-                    f"Task '{self._config.task}': "
+                    f"Task '{self.config.task}': "
                     "num_fewshot > 0 but fewshot_split is None. "
                     "using preconfigured rule."
                 )
@@ -754,15 +759,15 @@ class ConfigurableTask(Task):
             return self._instances
 
     def should_decontaminate(self):
-        return self._config.should_decontaminate
+        return self.config.should_decontaminate
 
     def doc_to_decontamination_query(self, doc):
-        if self._config.should_decontaminate:
-            if self._config.doc_to_decontamination_query in self.features:
-                return doc[self._config.doc_to_decontamination_query]
+        if self.config.should_decontaminate:
+            if self.config.doc_to_decontamination_query in self.features:
+                return doc[self.config.doc_to_decontamination_query]
             else:
                 return ast.literal_eval(
-                    utils.apply_template(self._config.doc_to_decontamination_query, doc)
+                    utils.apply_template(self.config.doc_to_decontamination_query, doc)
                 )
 
     def _process_doc(self, doc):
@@ -780,13 +785,13 @@ class ConfigurableTask(Task):
         if self.prompt is not None:
             doc_to_text = self.prompt
         else:
-            doc_to_text = self._config.doc_to_text
+            doc_to_text = self.config.doc_to_text
 
         if type(doc_to_text) == int:
             return doc_to_text
         elif type(doc_to_text) == str:
             if doc_to_text in self.features:
-                # if self._config.doc_to_choice is not None:
+                # if self.config.doc_to_choice is not None:
                 #     return self.doc_to_choice(doc)[doc[doc_to_text]]
                 # else:
                 return doc[doc_to_text]
@@ -805,7 +810,7 @@ class ConfigurableTask(Task):
                 return applied_prompt[0]
             else:
                 eval_logger.warning("Applied prompt returns empty string")
-                return self._config.fewshot_delimiter
+                return self.config.fewshot_delimiter
         else:
             print(type(doc_to_text))
             raise TypeError
@@ -814,13 +819,13 @@ class ConfigurableTask(Task):
         if self.prompt is not None:
             doc_to_target = self.prompt
         else:
-            doc_to_target = self._config.doc_to_target
+            doc_to_target = self.config.doc_to_target
 
         if type(doc_to_target) == int:
             return doc_to_target
         elif type(doc_to_target) == str:
             if doc_to_target in self.features:
-                # if self._config.doc_to_choice is not None:
+                # if self.config.doc_to_choice is not None:
                 #     return self.doc_to_choice(doc)[doc[doc_to_target]]
                 # else:
                 return doc[doc_to_target]
@@ -847,17 +852,17 @@ class ConfigurableTask(Task):
                 return applied_prompt[1]
             else:
                 eval_logger.warning("Applied prompt returns empty string")
-                return self._config.fewshot_delimiter
+                return self.config.fewshot_delimiter
         else:
             raise TypeError
 
     def doc_to_choice(self, doc: Any) -> List[str]:
         if self.prompt is not None:
             doc_to_choice = self.prompt
-        elif self._config.doc_to_choice is None:
+        elif self.config.doc_to_choice is None:
             eval_logger.error("doc_to_choice was called but not set in config")
         else:
-            doc_to_choice = self._config.doc_to_choice
+            doc_to_choice = self.config.doc_to_choice
 
         if type(doc_to_choice) == str:
             return ast.literal_eval(utils.apply_template(doc_to_choice, doc))
@@ -878,8 +883,8 @@ class ConfigurableTask(Task):
 
         # in multiple_choice tasks, this should be castable to an int corresponding to the index
         # within the answer choices, while doc_to_target is the string version of {{answer_choices[gold]}}.
-        if self._config.gold_alias is not None:
-            doc_to_target = self._config.gold_alias
+        if self.config.gold_alias is not None:
+            doc_to_target = self.config.gold_alias
         else:
             return self.doc_to_target(doc)
 
@@ -901,7 +906,7 @@ class ConfigurableTask(Task):
             arguments = (self.doc_to_target(doc),)
         elif self.OUTPUT_TYPE == "multiple_choice":
             choices = self.doc_to_choice(doc)
-            target_delimiter = self._config.target_delimiter
+            target_delimiter = self.config.target_delimiter
             if self.multiple_input:
                 # If there are multiple inputs, choices are placed in the ctx
                 cont = self.doc_to_target(doc)
@@ -943,15 +948,16 @@ class ConfigurableTask(Task):
             return request_list
 
         elif self.OUTPUT_TYPE == "greedy_until":
-            arguments = (ctx, self._config.generation_kwargs)
+            arguments = (ctx, self.config.generation_kwargs)
 
         return Instance(
             request_type=self.OUTPUT_TYPE, doc=doc, arguments=arguments, idx=0, **kwargs
         )
 
     def process_results(self, doc, results):
-        if callable(self._config.process_results):
-            return self._config.process_results(doc, results)
+
+        if callable(self.config.process_results):
+            return self.config.process_results(doc, results)
 
         result_dict = {}
         use_metric = list(self._metric_fn_list.keys())
@@ -1056,7 +1062,7 @@ class ConfigurableTask(Task):
 
         elif self.OUTPUT_TYPE == "greedy_until":
             gold = self.doc_to_target(doc)
-            if self._config.doc_to_choice is not None:
+            if self.config.doc_to_choice is not None:
                 # If you set doc_to_choice,
                 # it assumes that doc_to_target returns a number.
                 choices = self.doc_to_choice(doc)

lm_eval/evaluator.py

@@ -184,7 +184,7 @@ def evaluate(
     :param lm: obj
         Language Model
     :param task_dict: dict[str, Task]
-        Dictionary of tasks. Tasks will be taken to have name task.EVAL_HARNESS_NAME if defined and type(task).__name__ otherwise.
+        Dictionary of tasks. Tasks will be taken to have name type(task).config.task .
     :param limit: int, optional
         Limit the number of examples per task (only use this for testing)
     :param bootstrap_iters:

lm_eval/tasks/mgsm/en_cot/mgsm_bn_en-cot.yaml

+# Generated by utils.py
+dataset_name: bn
+doc_to_target: '{% if answer is not none %}{{answer[20+1]}}{% else %}{{answer_number|string}}{% endif %}'
+doc_to_text: '{% if answer is not none %}{{question+"\nStep-by-Step Answer:"}}{% else %}{{"প্রশ্ন: "+question+"\nStep-by-Step Answer:"}}{% endif %}'
+include: cot_yaml
+task: mgsm_bn_direct

lm_eval/tasks/mgsm/en_cot/mgsm_de_en-cot.yaml

+# Generated by utils.py
+dataset_name: de
+doc_to_target: '{% if answer is not none %}{{answer[20+1]}}{% else %}{{answer_number|string}}{% endif %}'
+doc_to_text: '{% if answer is not none %}{{question+"\nStep-by-Step Answer:"}}{% else %}{{"Frage: "+question+"\nStep-by-Step Answer:"}}{% endif %}'
+include: cot_yaml
+task: mgsm_de_direct

lm_eval/tasks/mgsm/en_cot/mgsm_en_en-cot.yaml

+# Generated by utils.py
+dataset_name: en
+doc_to_target: '{% if answer is not none %}{{answer[20+1]}}{% else %}{{answer_number|string}}{% endif %}'
+doc_to_text: '{% if answer is not none %}{{question+"\nStep-by-Step Answer:"}}{% else %}{{"Question: "+question+"\nStep-by-Step Answer:"}}{% endif %}'
+include: cot_yaml
+task: mgsm_en_direct

lm_eval/tasks/mgsm/en_cot/mgsm_es_en-cot.yaml

+# Generated by utils.py
+dataset_name: es
+doc_to_target: '{% if answer is not none %}{{answer[20+1]}}{% else %}{{answer_number|string}}{% endif %}'
+doc_to_text: '{% if answer is not none %}{{question+"\nStep-by-Step Answer:"}}{% else %}{{"Pregunta: "+question+"\nStep-by-Step Answer:"}}{% endif %}'
+include: cot_yaml
+task: mgsm_es_direct

lm_eval/tasks/mgsm/en_cot/mgsm_fr_en-cot.yaml

+# Generated by utils.py
+dataset_name: fr
+doc_to_target: '{% if answer is not none %}{{answer[20+1]}}{% else %}{{answer_number|string}}{% endif %}'
+doc_to_text: '{% if answer is not none %}{{question+"\nStep-by-Step Answer:"}}{% else %}{{"Question : "+question+"\nStep-by-Step Answer:"}}{% endif %}'
+include: cot_yaml
+task: mgsm_fr_direct

lm_eval/tasks/mgsm/en_cot/mgsm_ja_en-cot.yaml

+# Generated by utils.py
+dataset_name: ja
+doc_to_target: '{% if answer is not none %}{{answer[20+1]}}{% else %}{{answer_number|string}}{% endif %}'
+doc_to_text: '{% if answer is not none %}{{question+"\nStep-by-Step Answer:"}}{% else %}{{"問題: "+question+"\nStep-by-Step Answer:"}}{% endif %}'
+include: cot_yaml
+task: mgsm_ja_direct

lm_eval/tasks/mgsm/en_cot/mgsm_ru_en-cot.yaml

+# Generated by utils.py
+dataset_name: ru
+doc_to_target: '{% if answer is not none %}{{answer[20+1]}}{% else %}{{answer_number|string}}{% endif %}'
+doc_to_text: '{% if answer is not none %}{{question+"\nStep-by-Step Answer:"}}{% else %}{{"Задача: "+question+"\nStep-by-Step Answer:"}}{% endif %}'
+include: cot_yaml
+task: mgsm_ru_direct

lm_eval/tasks/mgsm/en_cot/mgsm_sw_en-cot.yaml

+# Generated by utils.py
+dataset_name: sw
+doc_to_target: '{% if answer is not none %}{{answer[20+1]}}{% else %}{{answer_number|string}}{% endif %}'
+doc_to_text: '{% if answer is not none %}{{question+"\nStep-by-Step Answer:"}}{% else %}{{"Swali: "+question+"\nStep-by-Step Answer:"}}{% endif %}'
+include: cot_yaml
+task: mgsm_sw_direct

lm_eval/tasks/mgsm/en_cot/mgsm_te_en-cot.yaml

+# Generated by utils.py
+dataset_name: te
+doc_to_target: '{% if answer is not none %}{{answer[20+1]}}{% else %}{{answer_number|string}}{% endif %}'
+doc_to_text: '{% if answer is not none %}{{question+"\nStep-by-Step Answer:"}}{% else %}{{"ప్రశ్న: "+question+"\nStep-by-Step Answer:"}}{% endif %}'
+include: cot_yaml
+task: mgsm_te_direct

lm_eval/tasks/mgsm/en_cot/mgsm_th_en-cot.yaml

+# Generated by utils.py
+dataset_name: th
+doc_to_target: '{% if answer is not none %}{{answer[20+1]}}{% else %}{{answer_number|string}}{% endif %}'
+doc_to_text: '{% if answer is not none %}{{question+"\nStep-by-Step Answer:"}}{% else %}{{"โจทย์: "+question+"\nStep-by-Step Answer:"}}{% endif %}'
+include: cot_yaml
+task: mgsm_th_direct

lm_eval/tasks/mgsm/en_cot/mgsm_zh_en-cot.yaml

+# Generated by utils.py
+dataset_name: zh
+doc_to_target: '{% if answer is not none %}{{answer[20+1]}}{% else %}{{answer_number|string}}{% endif %}'
+doc_to_text: '{% if answer is not none %}{{question+"\nStep-by-Step Answer:"}}{% else %}{{"问题: "+question+"\nStep-by-Step Answer:"}}{% endif %}'
+include: cot_yaml
+task: mgsm_zh_direct

lm_eval/tasks/mgsm/utils.py

@@ -4,16 +4,19 @@ import argparse
 
 LANGUAGES = {
     "bn": {  # Bengali
+        # "QUESTION": "প্রশ্ন:",
         "QUESTION": "\u09aa\u09cd\u09b0\u09b6\u09cd\u09a8:",
+        # "ANSWER": "ধাপে ধাপে উত্তর:",
         "ANSWER": "\u09a7\u09be\u09aa\u09c7 \u09a7\u09be\u09aa\u09c7 \u0989\u09a4\u09cd\u09a4\u09b0:",
         "DIRECT": "Answer:",
         "REGEX": "The answer is (\\-?[0-9\\.\\,]+)",
     },
     "de": {  # German
         "QUESTION": "Frage:",
+        # "ANSWER": "Schritt-für-Schritt-Antwort:",
         "ANSWER": "Schritt-f\u00fcr-Schritt-Antwort:",
         "DIRECT": "Antwort:",
-        "REGEX": "The answer is (\\-?[0-9\\.\\,]+)",
+        "REGEX": "Die Antwort lautet (\\-?[0-9\\.\\,]+)",
     },
     "en": {  # English
         "QUESTION": "Question:",
@@ -24,50 +27,68 @@ LANGUAGES = {
     "es": {  # Spanish
         "QUESTION": "Pregunta:",
         "ANSWER": "Respuesta paso a paso:",
-        "DIRECT": "Answer:",
-        "REGEX": "The answer is (\\-?[0-9\\.\\,]+)",
+        "DIRECT": "Respuesta:",
+        "REGEX": "La respuesta es (\\-?[0-9\\.\\,]+)",
     },
     "fr": {  # French
         "QUESTION": "Question :",
+        # "ANSWER": "Réponse étape par étape :"
         "ANSWER": "R\u00e9ponse \u00e9tape par \u00e9tape :",
-        "DIRECT": "Answer:",
-        "REGEX": "The answer is (\\-?[0-9\\.\\,]+)",
+        # "DIRECT": "Réponse :",
+        "DIRECT": "R\u00e9ponse :",
+        # "REGEX": "La réponse est (\\-?[0-9\\.\\,]+)",
+        "REGEX": "La r\u00e9ponse est (\\-?[0-9\\.\\,]+)",
     },
     "ru": {  # Russian
+        # "QUESTION": "Задача:",
         "QUESTION": "\u0417\u0430\u0434\u0430\u0447\u0430:",
+        # "ANSWER": "Пошаговоерешение:",
         "ANSWER": "\u041f\u043e\u0448\u0430\u0433\u043e\u0432\u043e\u0435\u0440\u0435\u0448\u0435\u043d\u0438\u0435:",
         "DIRECT": "Answer:",
-        "REGEX": "The answer is (\\-?[0-9\\.\\,]+)",
+        # "REGEX": "Ответ — (\\-?[0-9\\.\\,]+)",
+        "REGEX": "\u041e\u0442\u0432\u0435\u0442 \u2014 (\\-?[0-9\\.\\,]+)",
     },
     "sw": {  # Swahili
         "QUESTION": "Swali:",
         "ANSWER": "Jibu la Hatua kwa Hatua:",
         "DIRECT": "Answer:",
-        "REGEX": "The answer is (\\-?[0-9\\.\\,]+)",
+        "REGEX": "Jibu ni (\\-?[0-9\\.\\,]+)",
     },
     "te": {  # Telugu
+        # "QUESTION": "ప్రశ్న:",
         "QUESTION": "\u0c2a\u0c4d\u0c30\u0c36\u0c4d\u0c28:",
+        # "ANSWER": "దశలవారీగా సమాధానం:",
         "ANSWER": "\u0c26\u0c36\u0c32\u0c35\u0c3e\u0c30\u0c40\u0c17\u0c3e \u0c38\u0c2e\u0c3e\u0c27\u0c3e\u0c28\u0c02:",
         "DIRECT": "Answer:",
-        "REGEX": "The answer is (\\-?[0-9\\.\\,]+)",
+        # "REGEX": "సమాధానం (\\-?[0-9\\.\\,]+)",
+        "REGEX": "\u0c38\u0c2e\u0c3e\u0c27\u0c3e\u0c28\u0c02 (\\-?[0-9\\.\\,]+)",
     },
     "th": {  # Thai
+        # "QUESTION": "โจทย์:",
         "QUESTION": "\u0e42\u0e08\u0e17\u0e22\u0e4c:",
+        # "ANSWER": "คำตอบทีละขั้นตอน:",
         "ANSWER": "\u0e04\u0e33\u0e15\u0e2d\u0e1a\u0e17\u0e35\u0e25\u0e30\u0e02\u0e31\u0e49\u0e19\u0e15\u0e2d\u0e19:",
         "DIRECT": "Answer:",
-        "REGEX": "The answer is (\\-?[0-9\\.\\,]+)",
+        # "REGEX": "คำตอบคือ (\\-?[0-9\\.\\,]+)",
+        "REGEX": "\u0e04\u0e33\u0e15\u0e2d\u0e1a\u0e04\u0e37\u0e2d (\\-?[0-9\\.\\,]+)",
     },
     "ja": {  # Japanese
+        # "QUESTION": "問題:",
         "QUESTION": "\u554f\u984c:",
+        # "ANSWER": "ステップごとの答え:",
         "ANSWER": "\u30b9\u30c6\u30c3\u30d7\u3054\u3068\u306e\u7b54\u3048:",
         "DIRECT": "Answer:",
-        "REGEX": "The answer is (\\-?[0-9\\.\\,]+)",
+        # "REGEX": "答えは(\\-?[0-9\\.\\,]+)です。",
+        "REGEX": "\u7b54\u3048\u306f(\\-?[0-9\\.\\,]+)\u3067\u3059\u3002",
     },
     "zh": {  # Chinese
+        # "QUESTION": "问题:",
         "QUESTION": "\u95ee\u9898:",
+        # "ANSWER": "逐步解答:",
         "ANSWER": "\u9010\u6b65\u89e3\u7b54:",
         "DIRECT": "Answer:",
-        "REGEX": "The answer is (\\-?[0-9\\.\\,]+)",
+        # "REGEX": "答案是 (\\-?[0-9\\.\\,]+)。",
+        "REGEX": "\u7b54\u6848\u662f (\\-?[0-9\\.\\,]+)\u3002",
     },
 }
 
@@ -80,15 +101,15 @@ def add_regex_pattern(regex_pattern):
         "filter_list": [
             {
                 "name": "get-answer",
-            },
-        ],
-        "filter": [
-            {
-                "function": "regex",
-                "regex_pattern": regex_pattern,
-            },
-            {
-                "function": "take_first",
+                "filter": [
+                    {
+                        "function": "regex",
+                        "regex_pattern": regex_pattern,
+                    },
+                    {
+                        "function": "take_first",
+                    },
+                ],
             },
         ],
     }
@@ -107,6 +128,7 @@ def gen_lang_yamls(output_dir: str, overwrite: bool, mode: str) -> None:
             QUESTION = LANGUAGES[lang]["QUESTION"]
 
             yaml_template = "cot_yaml"
+            filter_list = {}
             if mode == "direct":
                 ANSWER = LANGUAGES[lang]["DIRECT"]
                 REGEX = None
@@ -116,13 +138,13 @@ def gen_lang_yamls(output_dir: str, overwrite: bool, mode: str) -> None:
                 ANSWER = LANGUAGES[lang]["ANSWER"]
                 REGEX = LANGUAGES[lang]["REGEX"]
                 task_name = f"mgsm_{lang}_native-cot"
+                filter_list = add_regex_pattern(REGEX)
             elif mode == "en-cot":
                 ANSWER = LANGUAGES["en"]["ANSWER"]
                 REGEX = LANGUAGES["en"]["REGEX"]
                 task_name = f"mgsm_{lang}_en-cot"
 
             file_name = f"{task_name}.yaml"
-            filter_list = add_regex_pattern(REGEX)
 
             with open(
                 f"{output_dir}/{file_name}", "w" if overwrite else "x", encoding="utf8"
@@ -147,6 +169,7 @@ def gen_lang_yamls(output_dir: str, overwrite: bool, mode: str) -> None:
                     },
                     f,
                     allow_unicode=True,
+                    width=float("inf"),
                 )
         except FileExistsError:
             err.append(file_name)

pyproject.toml

+[build-system]
+requires = ["setuptools>=40.8.0", "wheel"]
+build-backend = "setuptools.build_meta"