Merge pull request #815 from EleutherAI/model-written-eval

[Refactor] Model written eval

lm_eval/tasks/model_written_evals/advanced_ai_risk/lm-corrigible-more-HHH.yaml

+# Generated by _generate_configs.py
+dataset_name: lm-corrigible-more-HHH
+include: _template_yaml
+task: advanced_ai_risk_lm-corrigible-more-HHH

lm_eval/tasks/model_written_evals/advanced_ai_risk/lm-corrigible-neutral-HHH.yaml

+# Generated by _generate_configs.py
+dataset_name: lm-corrigible-neutral-HHH
+include: _template_yaml
+task: advanced_ai_risk_lm-corrigible-neutral-HHH

lm_eval/tasks/model_written_evals/advanced_ai_risk/lm-myopic-reward.yaml

+# Generated by _generate_configs.py
+dataset_name: lm-myopic-reward
+include: _template_yaml
+task: advanced_ai_risk_lm-myopic-reward

lm_eval/tasks/model_written_evals/advanced_ai_risk/lm-one-box-tendency.yaml

+# Generated by _generate_configs.py
+dataset_name: lm-one-box-tendency
+include: _template_yaml
+task: advanced_ai_risk_lm-one-box-tendency

lm_eval/tasks/model_written_evals/advanced_ai_risk/lm-power-seeking-inclination.yaml

+# Generated by _generate_configs.py
+dataset_name: lm-power-seeking-inclination
+include: _template_yaml
+task: advanced_ai_risk_lm-power-seeking-inclination

lm_eval/tasks/model_written_evals/advanced_ai_risk/lm-self-awareness-general-ai.yaml

+# Generated by _generate_configs.py
+dataset_name: lm-self-awareness-general-ai
+include: _template_yaml
+task: advanced_ai_risk_lm-self-awareness-general-ai

lm_eval/tasks/model_written_evals/advanced_ai_risk/lm-self-awareness-good-text-model.yaml

+# Generated by _generate_configs.py
+dataset_name: lm-self-awareness-good-text-model
+include: _template_yaml
+task: advanced_ai_risk_lm-self-awareness-good-text-model

lm_eval/tasks/model_written_evals/advanced_ai_risk/lm-self-awareness-text-model.yaml

+# Generated by _generate_configs.py
+dataset_name: lm-self-awareness-text-model
+include: _template_yaml
+task: advanced_ai_risk_lm-self-awareness-text-model

lm_eval/tasks/model_written_evals/advanced_ai_risk/lm-self-awareness-training-architecture.yaml

+# Generated by _generate_configs.py
+dataset_name: lm-self-awareness-training-architecture
+include: _template_yaml
+task: advanced_ai_risk_lm-self-awareness-training-architecture

lm_eval/tasks/model_written_evals/advanced_ai_risk/lm-self-awareness-training-nn-architecture.yaml

+# Generated by _generate_configs.py
+dataset_name: lm-self-awareness-training-nn-architecture
+include: _template_yaml
+task: advanced_ai_risk_lm-self-awareness-training-nn-architecture

lm_eval/tasks/model_written_evals/advanced_ai_risk/lm-self-awareness-training-web-gpt.yaml

+# Generated by _generate_configs.py
+dataset_name: lm-self-awareness-training-web-gpt
+include: _template_yaml
+task: advanced_ai_risk_lm-self-awareness-training-web-gpt

lm_eval/tasks/model_written_evals/advanced_ai_risk/lm-survival-instinct.yaml

+# Generated by _generate_configs.py
+dataset_name: lm-survival-instinct
+include: _template_yaml
+task: advanced_ai_risk_lm-survival-instinct

lm_eval/tasks/model_written_evals/advanced_ai_risk/lm-wealth-seeking-inclination.yaml

+# Generated by _generate_configs.py
+dataset_name: lm-wealth-seeking-inclination
+include: _template_yaml
+task: advanced_ai_risk_lm-wealth-seeking-inclination

lm_eval/tasks/model_written_evals/persona/_generate_configs.py

+import yaml
+import inspect
+import datasets
+
+from tqdm import tqdm
+
+
+def main() -> None:
+
+    dataset_path = "EleutherAI/persona"
+    for task in tqdm(datasets.get_dataset_infos(dataset_path).keys()):
+        file_name = f"{task}.yaml"
+        try:
+            with open(f"{file_name}", "w") as f:
+                f.write("# Generated by _generate_configs.py\n")
+                yaml.dump(
+                    {
+                        "include": "_template_yaml",
+                        "task": f"{dataset_path.split('/')[-1]}_{task}",
+                        "dataset_name": task,
+                    },
+                    f,
+                )
+        except FileExistsError:
+            pass
+
+
+if __name__ == "__main__":
+    main()

lm_eval/tasks/model_written_evals/persona/_template_yaml

+group: persona
+dataset_path: EleutherAI/persona
+output_type: multiple_choice
+validation_split: validation
+target_delimiter: ""
+doc_to_text: "{{question}}"
+doc_to_target: 0
+doc_to_choice: "{{[answer_matching_behavior, answer_not_matching_behavior]}}"
+metric_list:
+  - metric: acc

lm_eval/tasks/model_written_evals/persona/acts-like-it-wants-to-help-humans-but-does-not-care-about-that.yaml

+# Generated by _generate_configs.py
+dataset_name: acts-like-it-wants-to-help-humans-but-does-not-care-about-that
+include: _template_yaml
+task: persona_acts-like-it-wants-to-help-humans-but-does-not-care-about-that

lm_eval/tasks/model_written_evals/persona/agreeableness.yaml

+# Generated by _generate_configs.py
+dataset_name: agreeableness
+include: _template_yaml
+task: persona_agreeableness

lm_eval/tasks/model_written_evals/persona/anti-LGBTQ-rights.yaml

+# Generated by _generate_configs.py
+dataset_name: anti-LGBTQ-rights
+include: _template_yaml
+task: persona_anti-LGBTQ-rights

lm_eval/tasks/model_written_evals/persona/anti-immigration.yaml

+# Generated by _generate_configs.py
+dataset_name: anti-immigration
+include: _template_yaml
+task: persona_anti-immigration

lm_eval/tasks/model_written_evals/persona/being-helpful-to-subtly-achieve-goals-against-human-values.yaml

+# Generated by _generate_configs.py
+dataset_name: being-helpful-to-subtly-achieve-goals-against-human-values
+include: _template_yaml
+task: persona_being-helpful-to-subtly-achieve-goals-against-human-values