Upload folder using huggingface_hub

.gitattributes

@@ -1288,3 +1288,4 @@ classic_pythia-2.8b-step2/runs.json filter=lfs diff=lfs merge=lfs -text
 mmlu_pythia-1.4b-step2000/mmlu:subject=professional_law,method=multiple_choice_joint,model=EleutherAI_pythia-1.4b,eval_split=test,groups=mmlu_professional_law/display_requests.json filter=lfs diff=lfs merge=lfs -text
 mmlu_pythia-1.4b-step2000/mmlu:subject=professional_law,method=multiple_choice_joint,model=EleutherAI_pythia-1.4b,eval_split=test,groups=mmlu_professional_law/per_instance_stats.json filter=lfs diff=lfs merge=lfs -text
 mmlu_pythia-1.4b-step2000/mmlu:subject=professional_law,method=multiple_choice_joint,model=EleutherAI_pythia-1.4b,eval_split=test,groups=mmlu_professional_law/scenario_state.json filter=lfs diff=lfs merge=lfs -text
+lite_pythia-2.8b-step5000/gsm:model=EleutherAI_pythia-2.8b/scenario_state.json filter=lfs diff=lfs merge=lfs -text

lite_pythia-2.8b-step5000/commonsense:dataset=openbookqa,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b/run_spec.json

@@ -0,0 +1,74 @@
+{
+  "name": "commonsense:dataset=openbookqa,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+  "scenario_spec": {
+    "class_name": "helm.benchmark.scenarios.commonsense_scenario.OpenBookQA",
+    "args": {}
+  },
+  "adapter_spec": {
+    "method": "multiple_choice_joint",
+    "global_prefix": "",
+    "global_suffix": "",
+    "instructions": "The following are multiple choice questions (with answers) about common sense.\n",
+    "input_prefix": "Question: ",
+    "input_suffix": "\n",
+    "reference_prefix": "A. ",
+    "reference_suffix": "\n",
+    "chain_of_thought_prefix": "",
+    "chain_of_thought_suffix": "\n",
+    "output_prefix": "Answer: ",
+    "output_suffix": "\n",
+    "instance_prefix": "\n",
+    "substitutions": [],
+    "max_train_instances": 5,
+    "max_eval_instances": 1000,
+    "num_outputs": 5,
+    "num_train_trials": 1,
+    "num_trials": 1,
+    "sample_train": true,
+    "model_deployment": "EleutherAI/pythia-2.8b",
+    "model": "EleutherAI/pythia-2.8b",
+    "model_ability": 0.0,
+    "temperature": 0.0,
+    "max_tokens": 1,
+    "stop_sequences": [
+      "\n"
+    ],
+    "multi_label": false
+  },
+  "metric_specs": [
+    {
+      "class_name": "helm.benchmark.metrics.basic_metrics.BasicGenerationMetric",
+      "args": {
+        "names": [
+          "exact_match",
+          "quasi_exact_match",
+          "prefix_exact_match",
+          "quasi_prefix_exact_match"
+        ]
+      }
+    },
+    {
+      "class_name": "helm.benchmark.metrics.basic_metrics.BasicReferenceMetric",
+      "args": {}
+    },
+    {
+      "class_name": "helm.benchmark.metrics.basic_metrics.InstancesPerSplitMetric",
+      "args": {}
+    }
+  ],
+  "data_augmenter_spec": {
+    "perturbation_specs": [],
+    "should_augment_train_instances": false,
+    "should_include_original_train": false,
+    "should_skip_unchanged_train": false,
+    "should_augment_eval_instances": false,
+    "should_include_original_eval": false,
+    "should_skip_unchanged_eval": false,
+    "seeds_per_instance": 1
+  },
+  "groups": [
+    "openbookqa"
+  ],
+  "reeval_mode": false,
+  "reeval_max_samples": 50
+}

lite_pythia-2.8b-step5000/commonsense:dataset=openbookqa,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b/scenario.json

@@ -0,0 +1,9 @@
+{
+  "name": "openbookqa",
+  "description": "Benchmark from https://aclanthology.org/D18-1260.pdf.",
+  "tags": [
+    "knowledge",
+    "multiple_choice"
+  ],
+  "definition_path": "https://github.com/stanford-crfm/helm/blob/main/src/helm/benchmark/scenarios/commonsense_scenario.py"
+}

lite_pythia-2.8b-step5000/commonsense:dataset=openbookqa,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b/stats.json

@@ -0,0 +1,1538 @@
+[
+  {
+    "name": {
+      "name": "num_references",
+      "split": "test"
+    },
+    "count": 1,
+    "sum": 4.0,
+    "sum_squared": 16.0,
+    "min": 4.0,
+    "max": 4.0,
+    "mean": 4.0,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "num_train_trials",
+      "split": "test"
+    },
+    "count": 1,
+    "sum": 1.0,
+    "sum_squared": 1.0,
+    "min": 1.0,
+    "max": 1.0,
+    "mean": 1.0,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "num_prompt_tokens",
+      "split": "test"
+    },
+    "count": 1,
+    "sum": 251.556,
+    "sum_squared": 63280.421136000004,
+    "min": 251.556,
+    "max": 251.556,
+    "mean": 251.556,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "num_completion_tokens",
+      "split": "test"
+    },
+    "count": 1,
+    "sum": 1.0,
+    "sum_squared": 1.0,
+    "min": 1.0,
+    "max": 1.0,
+    "mean": 1.0,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "num_output_tokens",
+      "split": "test"
+    },
+    "count": 1,
+    "sum": 1.0,
+    "sum_squared": 1.0,
+    "min": 1.0,
+    "max": 1.0,
+    "mean": 1.0,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "training_co2_cost",
+      "split": "test"
+    },
+    "count": 0,
+    "sum": 0,
+    "sum_squared": 0
+  },
+  {
+    "name": {
+      "name": "training_energy_cost",
+      "split": "test"
+    },
+    "count": 0,
+    "sum": 0,
+    "sum_squared": 0
+  },
+  {
+    "name": {
+      "name": "inference_runtime",
+      "split": "test"
+    },
+    "count": 1,
+    "sum": 0.12883714818954467,
+    "sum_squared": 0.016599010753614692,
+    "min": 0.12883714818954467,
+    "max": 0.12883714818954467,
+    "mean": 0.12883714818954467,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "batch_size",
+      "split": "test"
+    },
+    "count": 1,
+    "sum": 1.0,
+    "sum_squared": 1.0,
+    "min": 1.0,
+    "max": 1.0,
+    "mean": 1.0,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "finish_reason_length",
+      "split": "test"
+    },
+    "count": 1,
+    "sum": 0.0,
+    "sum_squared": 0.0,
+    "min": 0.0,
+    "max": 0.0,
+    "mean": 0.0,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "finish_reason_stop",
+      "split": "test"
+    },
+    "count": 1,
+    "sum": 0.0,
+    "sum_squared": 0.0,
+    "min": 0.0,
+    "max": 0.0,
+    "mean": 0.0,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "finish_reason_endoftext",
+      "split": "test"
+    },
+    "count": 1,
+    "sum": 0.0,
+    "sum_squared": 0.0,
+    "min": 0.0,
+    "max": 0.0,
+    "mean": 0.0,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "finish_reason_unknown",
+      "split": "test"
+    },
+    "count": 1,
+    "sum": 1.0,
+    "sum_squared": 1.0,
+    "min": 1.0,
+    "max": 1.0,
+    "mean": 1.0,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "num_train_instances",
+      "split": "test"
+    },
+    "count": 1,
+    "sum": 5.0,
+    "sum_squared": 25.0,
+    "min": 5.0,
+    "max": 5.0,
+    "mean": 5.0,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "prompt_truncated",
+      "split": "test"
+    },
+    "count": 1,
+    "sum": 0.0,
+    "sum_squared": 0.0,
+    "min": 0.0,
+    "max": 0.0,
+    "mean": 0.0,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "max_prob",
+      "split": "test"
+    },
+    "count": 1,
+    "sum": 0.45501411941835707,
+    "sum_squared": 0.2070378488700629,
+    "min": 0.45501411941835707,
+    "max": 0.45501411941835707,
+    "mean": 0.45501411941835707,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "exact_match",
+      "split": "test"
+    },
+    "count": 1,
+    "sum": 0.22,
+    "sum_squared": 0.0484,
+    "min": 0.22,
+    "max": 0.22,
+    "mean": 0.22,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "exact_match@5",
+      "split": "test"
+    },
+    "count": 1,
+    "sum": 0.22,
+    "sum_squared": 0.0484,
+    "min": 0.22,
+    "max": 0.22,
+    "mean": 0.22,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "quasi_exact_match",
+      "split": "test"
+    },
+    "count": 1,
+    "sum": 0.22,
+    "sum_squared": 0.0484,
+    "min": 0.22,
+    "max": 0.22,
+    "mean": 0.22,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "quasi_exact_match@5",
+      "split": "test"
+    },
+    "count": 1,
+    "sum": 0.22,
+    "sum_squared": 0.0484,
+    "min": 0.22,
+    "max": 0.22,
+    "mean": 0.22,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "prefix_exact_match",
+      "split": "test"
+    },
+    "count": 1,
+    "sum": 0.22,
+    "sum_squared": 0.0484,
+    "min": 0.22,
+    "max": 0.22,
+    "mean": 0.22,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "prefix_exact_match@5",
+      "split": "test"
+    },
+    "count": 1,
+    "sum": 0.22,
+    "sum_squared": 0.0484,
+    "min": 0.22,
+    "max": 0.22,
+    "mean": 0.22,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "quasi_prefix_exact_match",
+      "split": "test"
+    },
+    "count": 1,
+    "sum": 0.22,
+    "sum_squared": 0.0484,
+    "min": 0.22,
+    "max": 0.22,
+    "mean": 0.22,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "quasi_prefix_exact_match@5",
+      "split": "test"
+    },
+    "count": 1,
+    "sum": 0.22,
+    "sum_squared": 0.0484,
+    "min": 0.22,
+    "max": 0.22,
+    "mean": 0.22,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "logprob",
+      "split": "test"
+    },
+    "count": 1,
+    "sum": -0.805416056394577,
+    "sum_squared": 0.6486950238981924,
+    "min": -0.805416056394577,
+    "max": -0.805416056394577,
+    "mean": -0.805416056394577,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "num_perplexity_tokens",
+      "split": "test"
+    },
+    "count": 1,
+    "sum": 1.0,
+    "sum_squared": 1.0,
+    "min": 1.0,
+    "max": 1.0,
+    "mean": 1.0,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "num_bytes",
+      "split": "test"
+    },
+    "count": 1,
+    "sum": 2.0,
+    "sum_squared": 4.0,
+    "min": 2.0,
+    "max": 2.0,
+    "mean": 2.0,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "perplexity",
+      "split": "test"
+    },
+    "count": 1,
+    "sum": 2.2376272843172513,
+    "sum_squared": 5.0069758635209975,
+    "min": 2.2376272843172513,
+    "max": 2.2376272843172513,
+    "mean": 2.2376272843172513,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "bits_per_byte",
+      "split": "test"
+    },
+    "count": 1,
+    "sum": 0.580984875206401,
+    "sum_squared": 0.3375434252185973,
+    "min": 0.580984875206401,
+    "max": 0.580984875206401,
+    "mean": 0.580984875206401,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "logprob_per_byte",
+      "split": "test"
+    },
+    "count": 1,
+    "sum": -0.4027080281972885,
+    "sum_squared": 0.1621737559745481,
+    "min": -0.4027080281972885,
+    "max": -0.4027080281972885,
+    "mean": -0.4027080281972885,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "ece_10_bin",
+      "split": "test"
+    },
+    "count": 1,
+    "sum": 0.23501411941835657,
+    "sum_squared": 0.05523163632598556,
+    "min": 0.23501411941835657,
+    "max": 0.23501411941835657,
+    "mean": 0.23501411941835657,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "ece_1_bin",
+      "split": "test"
+    },
+    "count": 1,
+    "sum": 0.23501411941835657,
+    "sum_squared": 0.05523163632598556,
+    "min": 0.23501411941835657,
+    "max": 0.23501411941835657,
+    "mean": 0.23501411941835657,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "selective_cov_acc_area",
+      "split": "test"
+    },
+    "count": 1,
+    "sum": 0.2114153671720523,
+    "sum_squared": 0.044696457476493694,
+    "min": 0.2114153671720523,
+    "max": 0.2114153671720523,
+    "mean": 0.2114153671720523,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "selective_acc@10",
+      "split": "test"
+    },
+    "count": 1,
+    "sum": 0.17647058823529413,
+    "sum_squared": 0.031141868512110732,
+    "min": 0.17647058823529413,
+    "max": 0.17647058823529413,
+    "mean": 0.17647058823529413,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "platt_coef",
+      "split": "test"
+    },
+    "count": 1,
+    "sum": -0.09815250868690073,
+    "sum_squared": 0.009633914961532123,
+    "min": -0.09815250868690073,
+    "max": -0.09815250868690073,
+    "mean": -0.09815250868690073,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "platt_intercept",
+      "split": "test"
+    },
+    "count": 1,
+    "sum": -1.2836296815308241,
+    "sum_squared": 1.647705159306925,
+    "min": -1.2836296815308241,
+    "max": -1.2836296815308241,
+    "mean": -1.2836296815308241,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "platt_ece_10_bin",
+      "split": "test"
+    },
+    "count": 1,
+    "sum": 0.03173473384942972,
+    "sum_squared": 0.0010070933324941406,
+    "min": 0.03173473384942972,
+    "max": 0.03173473384942972,
+    "mean": 0.03173473384942972,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "platt_ece_1_bin",
+      "split": "test"
+    },
+    "count": 1,
+    "sum": 7.75700809032831e-05,
+    "sum_squared": 6.017117451341886e-09,
+    "min": 7.75700809032831e-05,
+    "max": 7.75700809032831e-05,
+    "mean": 7.75700809032831e-05,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "num_references",
+      "split": "test",
+      "perturbation": {
+        "name": "robustness",
+        "robustness": true,
+        "fairness": false,
+        "computed_on": "worst"
+      }
+    },
+    "count": 1,
+    "sum": 4.0,
+    "sum_squared": 16.0,
+    "min": 4.0,
+    "max": 4.0,
+    "mean": 4.0,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "num_references",
+      "split": "test",
+      "perturbation": {
+        "name": "fairness",
+        "robustness": false,
+        "fairness": true,
+        "computed_on": "worst"
+      }
+    },
+    "count": 1,
+    "sum": 4.0,
+    "sum_squared": 16.0,
+    "min": 4.0,
+    "max": 4.0,
+    "mean": 4.0,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "num_train_trials",
+      "split": "test",
+      "perturbation": {
+        "name": "robustness",
+        "robustness": true,
+        "fairness": false,
+        "computed_on": "worst"
+      }
+    },
+    "count": 1,
+    "sum": 1.0,
+    "sum_squared": 1.0,
+    "min": 1.0,
+    "max": 1.0,
+    "mean": 1.0,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "num_train_trials",
+      "split": "test",
+      "perturbation": {
+        "name": "fairness",
+        "robustness": false,
+        "fairness": true,
+        "computed_on": "worst"
+      }
+    },
+    "count": 1,
+    "sum": 1.0,
+    "sum_squared": 1.0,
+    "min": 1.0,
+    "max": 1.0,
+    "mean": 1.0,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "num_prompt_tokens",
+      "split": "test",
+      "perturbation": {
+        "name": "robustness",
+        "robustness": true,
+        "fairness": false,
+        "computed_on": "worst"
+      }
+    },
+    "count": 1,
+    "sum": 251.556,
+    "sum_squared": 63280.421136000004,
+    "min": 251.556,
+    "max": 251.556,
+    "mean": 251.556,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "num_prompt_tokens",
+      "split": "test",
+      "perturbation": {
+        "name": "fairness",
+        "robustness": false,
+        "fairness": true,
+        "computed_on": "worst"
+      }
+    },
+    "count": 1,
+    "sum": 251.556,
+    "sum_squared": 63280.421136000004,
+    "min": 251.556,
+    "max": 251.556,
+    "mean": 251.556,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "num_completion_tokens",
+      "split": "test",
+      "perturbation": {
+        "name": "robustness",
+        "robustness": true,
+        "fairness": false,
+        "computed_on": "worst"
+      }
+    },
+    "count": 1,
+    "sum": 1.0,
+    "sum_squared": 1.0,
+    "min": 1.0,
+    "max": 1.0,
+    "mean": 1.0,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "num_completion_tokens",
+      "split": "test",
+      "perturbation": {
+        "name": "fairness",
+        "robustness": false,
+        "fairness": true,
+        "computed_on": "worst"
+      }
+    },
+    "count": 1,
+    "sum": 1.0,
+    "sum_squared": 1.0,
+    "min": 1.0,
+    "max": 1.0,
+    "mean": 1.0,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "num_output_tokens",
+      "split": "test",
+      "perturbation": {
+        "name": "robustness",
+        "robustness": true,
+        "fairness": false,
+        "computed_on": "worst"
+      }
+    },
+    "count": 1,
+    "sum": 1.0,
+    "sum_squared": 1.0,
+    "min": 1.0,
+    "max": 1.0,
+    "mean": 1.0,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "num_output_tokens",
+      "split": "test",
+      "perturbation": {
+        "name": "fairness",
+        "robustness": false,
+        "fairness": true,
+        "computed_on": "worst"
+      }
+    },
+    "count": 1,
+    "sum": 1.0,
+    "sum_squared": 1.0,
+    "min": 1.0,
+    "max": 1.0,
+    "mean": 1.0,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "inference_runtime",
+      "split": "test",
+      "perturbation": {
+        "name": "robustness",
+        "robustness": true,
+        "fairness": false,
+        "computed_on": "worst"
+      }
+    },
+    "count": 1,
+    "sum": 0.12883714818954467,
+    "sum_squared": 0.016599010753614692,
+    "min": 0.12883714818954467,
+    "max": 0.12883714818954467,
+    "mean": 0.12883714818954467,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "inference_runtime",
+      "split": "test",
+      "perturbation": {
+        "name": "fairness",
+        "robustness": false,
+        "fairness": true,
+        "computed_on": "worst"
+      }
+    },
+    "count": 1,
+    "sum": 0.12883714818954467,
+    "sum_squared": 0.016599010753614692,
+    "min": 0.12883714818954467,
+    "max": 0.12883714818954467,
+    "mean": 0.12883714818954467,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "batch_size",
+      "split": "test",
+      "perturbation": {
+        "name": "robustness",
+        "robustness": true,
+        "fairness": false,
+        "computed_on": "worst"
+      }
+    },
+    "count": 1,
+    "sum": 1.0,
+    "sum_squared": 1.0,
+    "min": 1.0,
+    "max": 1.0,
+    "mean": 1.0,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "batch_size",
+      "split": "test",
+      "perturbation": {
+        "name": "fairness",
+        "robustness": false,
+        "fairness": true,
+        "computed_on": "worst"
+      }
+    },
+    "count": 1,
+    "sum": 1.0,
+    "sum_squared": 1.0,
+    "min": 1.0,
+    "max": 1.0,
+    "mean": 1.0,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "finish_reason_length",
+      "split": "test",
+      "perturbation": {
+        "name": "robustness",
+        "robustness": true,
+        "fairness": false,
+        "computed_on": "worst"
+      }
+    },
+    "count": 1,
+    "sum": 0.0,
+    "sum_squared": 0.0,
+    "min": 0.0,
+    "max": 0.0,
+    "mean": 0.0,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "finish_reason_length",
+      "split": "test",
+      "perturbation": {
+        "name": "fairness",
+        "robustness": false,
+        "fairness": true,
+        "computed_on": "worst"
+      }
+    },
+    "count": 1,
+    "sum": 0.0,
+    "sum_squared": 0.0,
+    "min": 0.0,
+    "max": 0.0,
+    "mean": 0.0,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "finish_reason_stop",
+      "split": "test",
+      "perturbation": {
+        "name": "robustness",
+        "robustness": true,
+        "fairness": false,
+        "computed_on": "worst"
+      }
+    },
+    "count": 1,
+    "sum": 0.0,
+    "sum_squared": 0.0,
+    "min": 0.0,
+    "max": 0.0,
+    "mean": 0.0,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "finish_reason_stop",
+      "split": "test",
+      "perturbation": {
+        "name": "fairness",
+        "robustness": false,
+        "fairness": true,
+        "computed_on": "worst"
+      }
+    },
+    "count": 1,
+    "sum": 0.0,
+    "sum_squared": 0.0,
+    "min": 0.0,
+    "max": 0.0,
+    "mean": 0.0,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "finish_reason_endoftext",
+      "split": "test",
+      "perturbation": {
+        "name": "robustness",
+        "robustness": true,
+        "fairness": false,
+        "computed_on": "worst"
+      }
+    },
+    "count": 1,
+    "sum": 0.0,
+    "sum_squared": 0.0,
+    "min": 0.0,
+    "max": 0.0,
+    "mean": 0.0,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "finish_reason_endoftext",
+      "split": "test",
+      "perturbation": {
+        "name": "fairness",
+        "robustness": false,
+        "fairness": true,
+        "computed_on": "worst"
+      }
+    },
+    "count": 1,
+    "sum": 0.0,
+    "sum_squared": 0.0,
+    "min": 0.0,
+    "max": 0.0,
+    "mean": 0.0,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "finish_reason_unknown",
+      "split": "test",
+      "perturbation": {
+        "name": "robustness",
+        "robustness": true,
+        "fairness": false,
+        "computed_on": "worst"
+      }
+    },
+    "count": 1,
+    "sum": 1.0,
+    "sum_squared": 1.0,
+    "min": 1.0,
+    "max": 1.0,
+    "mean": 1.0,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "finish_reason_unknown",
+      "split": "test",
+      "perturbation": {
+        "name": "fairness",
+        "robustness": false,
+        "fairness": true,
+        "computed_on": "worst"
+      }
+    },
+    "count": 1,
+    "sum": 1.0,
+    "sum_squared": 1.0,
+    "min": 1.0,
+    "max": 1.0,
+    "mean": 1.0,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "num_train_instances",
+      "split": "test",
+      "perturbation": {
+        "name": "robustness",
+        "robustness": true,
+        "fairness": false,
+        "computed_on": "worst"
+      }
+    },
+    "count": 1,
+    "sum": 5.0,
+    "sum_squared": 25.0,
+    "min": 5.0,
+    "max": 5.0,
+    "mean": 5.0,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "num_train_instances",
+      "split": "test",
+      "perturbation": {
+        "name": "fairness",
+        "robustness": false,
+        "fairness": true,
+        "computed_on": "worst"
+      }
+    },
+    "count": 1,
+    "sum": 5.0,
+    "sum_squared": 25.0,
+    "min": 5.0,
+    "max": 5.0,
+    "mean": 5.0,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "prompt_truncated",
+      "split": "test",
+      "perturbation": {
+        "name": "robustness",
+        "robustness": true,
+        "fairness": false,
+        "computed_on": "worst"
+      }
+    },
+    "count": 1,
+    "sum": 0.0,
+    "sum_squared": 0.0,
+    "min": 0.0,
+    "max": 0.0,
+    "mean": 0.0,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "prompt_truncated",
+      "split": "test",
+      "perturbation": {
+        "name": "fairness",
+        "robustness": false,
+        "fairness": true,
+        "computed_on": "worst"
+      }
+    },
+    "count": 1,
+    "sum": 0.0,
+    "sum_squared": 0.0,
+    "min": 0.0,
+    "max": 0.0,
+    "mean": 0.0,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "max_prob",
+      "split": "test",
+      "perturbation": {
+        "name": "robustness",
+        "robustness": true,
+        "fairness": false,
+        "computed_on": "worst"
+      }
+    },
+    "count": 1,
+    "sum": 0.45501411941835707,
+    "sum_squared": 0.2070378488700629,
+    "min": 0.45501411941835707,
+    "max": 0.45501411941835707,
+    "mean": 0.45501411941835707,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "max_prob",
+      "split": "test",
+      "perturbation": {
+        "name": "fairness",
+        "robustness": false,
+        "fairness": true,
+        "computed_on": "worst"
+      }
+    },
+    "count": 1,
+    "sum": 0.45501411941835707,
+    "sum_squared": 0.2070378488700629,
+    "min": 0.45501411941835707,
+    "max": 0.45501411941835707,
+    "mean": 0.45501411941835707,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "exact_match",
+      "split": "test",
+      "perturbation": {
+        "name": "robustness",
+        "robustness": true,
+        "fairness": false,
+        "computed_on": "worst"
+      }
+    },
+    "count": 1,
+    "sum": 0.22,
+    "sum_squared": 0.0484,
+    "min": 0.22,
+    "max": 0.22,
+    "mean": 0.22,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "exact_match",
+      "split": "test",
+      "perturbation": {
+        "name": "fairness",
+        "robustness": false,
+        "fairness": true,
+        "computed_on": "worst"
+      }
+    },
+    "count": 1,
+    "sum": 0.22,
+    "sum_squared": 0.0484,
+    "min": 0.22,
+    "max": 0.22,
+    "mean": 0.22,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "exact_match@5",
+      "split": "test",
+      "perturbation": {
+        "name": "robustness",
+        "robustness": true,
+        "fairness": false,
+        "computed_on": "worst"
+      }
+    },
+    "count": 1,
+    "sum": 0.22,
+    "sum_squared": 0.0484,
+    "min": 0.22,
+    "max": 0.22,
+    "mean": 0.22,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "exact_match@5",
+      "split": "test",
+      "perturbation": {
+        "name": "fairness",
+        "robustness": false,
+        "fairness": true,
+        "computed_on": "worst"
+      }
+    },
+    "count": 1,
+    "sum": 0.22,
+    "sum_squared": 0.0484,
+    "min": 0.22,
+    "max": 0.22,
+    "mean": 0.22,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "quasi_exact_match",
+      "split": "test",
+      "perturbation": {
+        "name": "robustness",
+        "robustness": true,
+        "fairness": false,
+        "computed_on": "worst"
+      }
+    },
+    "count": 1,
+    "sum": 0.22,
+    "sum_squared": 0.0484,
+    "min": 0.22,
+    "max": 0.22,
+    "mean": 0.22,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "quasi_exact_match",
+      "split": "test",
+      "perturbation": {
+        "name": "fairness",
+        "robustness": false,
+        "fairness": true,
+        "computed_on": "worst"
+      }
+    },
+    "count": 1,
+    "sum": 0.22,
+    "sum_squared": 0.0484,
+    "min": 0.22,
+    "max": 0.22,
+    "mean": 0.22,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "quasi_exact_match@5",
+      "split": "test",
+      "perturbation": {
+        "name": "robustness",
+        "robustness": true,
+        "fairness": false,
+        "computed_on": "worst"
+      }
+    },
+    "count": 1,
+    "sum": 0.22,
+    "sum_squared": 0.0484,
+    "min": 0.22,
+    "max": 0.22,
+    "mean": 0.22,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "quasi_exact_match@5",
+      "split": "test",
+      "perturbation": {
+        "name": "fairness",
+        "robustness": false,
+        "fairness": true,
+        "computed_on": "worst"
+      }
+    },
+    "count": 1,
+    "sum": 0.22,
+    "sum_squared": 0.0484,
+    "min": 0.22,
+    "max": 0.22,
+    "mean": 0.22,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "prefix_exact_match",
+      "split": "test",
+      "perturbation": {
+        "name": "robustness",
+        "robustness": true,
+        "fairness": false,
+        "computed_on": "worst"
+      }
+    },
+    "count": 1,
+    "sum": 0.22,
+    "sum_squared": 0.0484,
+    "min": 0.22,
+    "max": 0.22,
+    "mean": 0.22,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "prefix_exact_match",
+      "split": "test",
+      "perturbation": {
+        "name": "fairness",
+        "robustness": false,
+        "fairness": true,
+        "computed_on": "worst"
+      }
+    },
+    "count": 1,
+    "sum": 0.22,
+    "sum_squared": 0.0484,
+    "min": 0.22,
+    "max": 0.22,
+    "mean": 0.22,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "prefix_exact_match@5",
+      "split": "test",
+      "perturbation": {
+        "name": "robustness",
+        "robustness": true,
+        "fairness": false,
+        "computed_on": "worst"
+      }
+    },
+    "count": 1,
+    "sum": 0.22,
+    "sum_squared": 0.0484,
+    "min": 0.22,
+    "max": 0.22,
+    "mean": 0.22,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "prefix_exact_match@5",
+      "split": "test",
+      "perturbation": {
+        "name": "fairness",
+        "robustness": false,
+        "fairness": true,
+        "computed_on": "worst"
+      }
+    },
+    "count": 1,
+    "sum": 0.22,
+    "sum_squared": 0.0484,
+    "min": 0.22,
+    "max": 0.22,
+    "mean": 0.22,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "quasi_prefix_exact_match",
+      "split": "test",
+      "perturbation": {
+        "name": "robustness",
+        "robustness": true,
+        "fairness": false,
+        "computed_on": "worst"
+      }
+    },
+    "count": 1,
+    "sum": 0.22,
+    "sum_squared": 0.0484,
+    "min": 0.22,
+    "max": 0.22,
+    "mean": 0.22,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "quasi_prefix_exact_match",
+      "split": "test",
+      "perturbation": {
+        "name": "fairness",
+        "robustness": false,
+        "fairness": true,
+        "computed_on": "worst"
+      }
+    },
+    "count": 1,
+    "sum": 0.22,
+    "sum_squared": 0.0484,
+    "min": 0.22,
+    "max": 0.22,
+    "mean": 0.22,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "quasi_prefix_exact_match@5",
+      "split": "test",
+      "perturbation": {
+        "name": "robustness",
+        "robustness": true,
+        "fairness": false,
+        "computed_on": "worst"
+      }
+    },
+    "count": 1,
+    "sum": 0.22,
+    "sum_squared": 0.0484,
+    "min": 0.22,
+    "max": 0.22,
+    "mean": 0.22,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "quasi_prefix_exact_match@5",
+      "split": "test",
+      "perturbation": {
+        "name": "fairness",
+        "robustness": false,
+        "fairness": true,
+        "computed_on": "worst"
+      }
+    },
+    "count": 1,
+    "sum": 0.22,
+    "sum_squared": 0.0484,
+    "min": 0.22,
+    "max": 0.22,
+    "mean": 0.22,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "logprob",
+      "split": "test",
+      "perturbation": {
+        "name": "robustness",
+        "robustness": true,
+        "fairness": false,
+        "computed_on": "worst"
+      }
+    },
+    "count": 1,
+    "sum": -0.805416056394577,
+    "sum_squared": 0.6486950238981924,
+    "min": -0.805416056394577,
+    "max": -0.805416056394577,
+    "mean": -0.805416056394577,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "logprob",
+      "split": "test",
+      "perturbation": {
+        "name": "fairness",
+        "robustness": false,
+        "fairness": true,
+        "computed_on": "worst"
+      }
+    },
+    "count": 1,
+    "sum": -0.805416056394577,
+    "sum_squared": 0.6486950238981924,
+    "min": -0.805416056394577,
+    "max": -0.805416056394577,
+    "mean": -0.805416056394577,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "num_perplexity_tokens",
+      "split": "test",
+      "perturbation": {
+        "name": "robustness",
+        "robustness": true,
+        "fairness": false,
+        "computed_on": "worst"
+      }
+    },
+    "count": 1,
+    "sum": 1.0,
+    "sum_squared": 1.0,
+    "min": 1.0,
+    "max": 1.0,
+    "mean": 1.0,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "num_perplexity_tokens",
+      "split": "test",
+      "perturbation": {
+        "name": "fairness",
+        "robustness": false,
+        "fairness": true,
+        "computed_on": "worst"
+      }
+    },
+    "count": 1,
+    "sum": 1.0,
+    "sum_squared": 1.0,
+    "min": 1.0,
+    "max": 1.0,
+    "mean": 1.0,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "num_bytes",
+      "split": "test",
+      "perturbation": {
+        "name": "robustness",
+        "robustness": true,
+        "fairness": false,
+        "computed_on": "worst"
+      }
+    },
+    "count": 1,
+    "sum": 2.0,
+    "sum_squared": 4.0,
+    "min": 2.0,
+    "max": 2.0,
+    "mean": 2.0,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "num_bytes",
+      "split": "test",
+      "perturbation": {
+        "name": "fairness",
+        "robustness": false,
+        "fairness": true,
+        "computed_on": "worst"
+      }
+    },
+    "count": 1,
+    "sum": 2.0,
+    "sum_squared": 4.0,
+    "min": 2.0,
+    "max": 2.0,
+    "mean": 2.0,
+    "variance": 0.0,
+    "stddev": 0.0
+  },
+  {
+    "name": {
+      "name": "num_instances",
+      "split": "test"
+    },
+    "count": 1,
+    "sum": 500.0,
+    "sum_squared": 250000.0,
+    "min": 500.0,
+    "max": 500.0,
+    "mean": 500.0,
+    "variance": 0.0,
+    "stddev": 0.0
+  }
+]

lite_pythia-2.8b-step5000/costs.json

@@ -0,0 +1 @@
+{}

lite_pythia-2.8b-step5000/groups.json

@@ -0,0 +1,491 @@
+[
+  {
+    "title": "All scenarios",
+    "header": [
+      {
+        "value": "Group",
+        "markdown": false,
+        "metadata": {}
+      },
+      {
+        "value": "Description",
+        "markdown": false,
+        "metadata": {}
+      },
+      {
+        "value": "Adaptation method",
+        "description": "Adaptation strategy (e.g., generation)",
+        "markdown": false,
+        "metadata": {}
+      },
+      {
+        "value": "# instances",
+        "description": "Number of instances evaluated on",
+        "markdown": false,
+        "metadata": {}
+      },
+      {
+        "value": "# references",
+        "description": "Number of references provided per instance",
+        "markdown": false,
+        "metadata": {}
+      },
+      {
+        "value": "# prompt tokens",
+        "description": "Total number of prompt tokens",
+        "markdown": false,
+        "metadata": {}
+      },
+      {
+        "value": "# completion tokens",
+        "description": "Total number of completion tokens",
+        "markdown": false,
+        "metadata": {}
+      },
+      {
+        "value": "# models",
+        "description": "Number of models we're evaluating",
+        "markdown": false,
+        "metadata": {}
+      }
+    ],
+    "rows": [
+      [
+        {
+          "value": "Core scenarios",
+          "href": "?group=core_scenarios",
+          "markdown": false
+        },
+        {
+          "value": "The scenarios where we evaluate all the models.",
+          "markdown": true
+        },
+        {
+          "value": "multiple_choice_joint, generation",
+          "markdown": false
+        },
+        {
+          "value": 242.0,
+          "description": "min=8, mean=242, max=1000, sum=4114 (17)",
+          "markdown": false
+        },
+        {
+          "value": 2.9411764705882355,
+          "description": "min=1, mean=2.941, max=4, sum=150 (51)",
+          "markdown": false
+        },
+        {
+          "value": 26153.916629538737,
+          "description": "min=206.779, mean=512.822, max=1497.455, sum=26153.917 (51)",
+          "markdown": false
+        },
+        {
+          "value": 563.0038753823058,
+          "description": "min=1, mean=11.039, max=168.459, sum=563.004 (51)",
+          "markdown": false
+        },
+        {
+          "value": 1,
+          "markdown": false
+        }
+      ]
+    ],
+    "links": []
+  },
+  {
+    "title": "Scenarios",
+    "header": [
+      {
+        "value": "Group",
+        "markdown": false,
+        "metadata": {}
+      },
+      {
+        "value": "Description",
+        "markdown": false,
+        "metadata": {}
+      },
+      {
+        "value": "Adaptation method",
+        "description": "Adaptation strategy (e.g., generation)",
+        "markdown": false,
+        "metadata": {}
+      },
+      {
+        "value": "# instances",
+        "description": "Number of instances evaluated on",
+        "markdown": false,
+        "metadata": {}
+      },
+      {
+        "value": "# references",
+        "description": "Number of references provided per instance",
+        "markdown": false,
+        "metadata": {}
+      },
+      {
+        "value": "# prompt tokens",
+        "description": "Total number of prompt tokens",
+        "markdown": false,
+        "metadata": {}
+      },
+      {
+        "value": "# completion tokens",
+        "description": "Total number of completion tokens",
+        "markdown": false,
+        "metadata": {}
+      },
+      {
+        "value": "# models",
+        "description": "Number of models we're evaluating",
+        "markdown": false,
+        "metadata": {}
+      }
+    ],
+    "rows": [
+      [
+        {
+          "value": "NarrativeQA",
+          "href": "?group=narrative_qa",
+          "markdown": false
+        },
+        {
+          "value": "The NarrativeQA benchmark for reading comprehension over narratives [(Ko\u010disk\u00fd et al., 2017)](https://aclanthology.org/Q18-1023/).",
+          "markdown": true
+        },
+        {
+          "value": "",
+          "markdown": false
+        },
+        {
+          "markdown": false
+        },
+        {
+          "markdown": false
+        },
+        {
+          "markdown": false
+        },
+        {
+          "markdown": false
+        },
+        {
+          "value": 0,
+          "markdown": false
+        }
+      ],
+      [
+        {
+          "value": "NaturalQuestions (closed-book)",
+          "href": "?group=natural_qa_closedbook",
+          "markdown": false
+        },
+        {
+          "value": "The NaturalQuestions [(Kwiatkowski et al., 2019)](https://aclanthology.org/Q19-1026/) benchmark for question answering based on naturally-occurring queries through Google Search. The input does not include the Wikipedia page with the answer.",
+          "markdown": true
+        },
+        {
+          "value": "",
+          "markdown": false
+        },
+        {
+          "markdown": false
+        },
+        {
+          "markdown": false
+        },
+        {
+          "markdown": false
+        },
+        {
+          "markdown": false
+        },
+        {
+          "value": 0,
+          "markdown": false
+        }
+      ],
+      [
+        {
+          "value": "NaturalQuestions (open-book)",
+          "href": "?group=natural_qa_openbook_longans",
+          "markdown": false
+        },
+        {
+          "value": "The NaturalQuestions [(Kwiatkowski et al., 2019)](https://aclanthology.org/Q19-1026/) benchmark for question answering based on naturally-occurring queries through Google Search. The input includes the Wikipedia page with the answer.",
+          "markdown": true
+        },
+        {
+          "value": "",
+          "markdown": false
+        },
+        {
+          "markdown": false
+        },
+        {
+          "markdown": false
+        },
+        {
+          "markdown": false
+        },
+        {
+          "markdown": false
+        },
+        {
+          "value": 0,
+          "markdown": false
+        }
+      ],
+      [
+        {
+          "value": "OpenbookQA",
+          "href": "?group=openbookqa",
+          "markdown": false
+        },
+        {
+          "value": "The OpenbookQA benchmark for commonsense-intensive open book question answering [(Mihaylov et al., 2018)](https://aclanthology.org/D18-1260/).",
+          "markdown": true
+        },
+        {
+          "value": "multiple_choice_joint",
+          "markdown": false
+        },
+        {
+          "value": 500.0,
+          "description": "min=500, mean=500, max=500, sum=500 (1)",
+          "markdown": false
+        },
+        {
+          "value": 4.0,
+          "description": "min=4, mean=4, max=4, sum=12 (3)",
+          "markdown": false
+        },
+        {
+          "value": 754.668,
+          "description": "min=251.556, mean=251.556, max=251.556, sum=754.668 (3)",
+          "markdown": false
+        },
+        {
+          "value": 3.0,
+          "description": "min=1, mean=1, max=1, sum=3 (3)",
+          "markdown": false
+        },
+        {
+          "value": 1,
+          "markdown": false
+        }
+      ],
+      [
+        {
+          "value": "MMLU (Massive Multitask Language Understanding)",
+          "href": "?group=mmlu",
+          "markdown": false
+        },
+        {
+          "value": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).",
+          "markdown": true
+        },
+        {
+          "value": "multiple_choice_joint",
+          "markdown": false
+        },
+        {
+          "value": 56.7,
+          "description": "min=8, mean=56.7, max=114, sum=567 (10)",
+          "markdown": false
+        },
+        {
+          "value": 4.0,
+          "description": "min=4, mean=4, max=4, sum=120 (30)",
+          "markdown": false
+        },
+        {
+          "value": 14170.841555023922,
+          "description": "min=354.455, mean=472.361, max=612.798, sum=14170.842 (30)",
+          "markdown": false
+        },
+        {
+          "value": 30.0,
+          "description": "min=1, mean=1, max=1, sum=30 (30)",
+          "markdown": false
+        },
+        {
+          "value": 1,
+          "markdown": false
+        }
+      ],
+      [
+        {
+          "value": "GSM8K (Grade School Math)",
+          "href": "?group=gsm",
+          "markdown": false
+        },
+        {
+          "value": "The grade school math word problems dataset (GSM8K) for testing mathematical reasoning on grade-school math problems [(Cobbe et al., 2021)](https://arxiv.org/pdf/2110.14168.pdf).",
+          "markdown": true
+        },
+        {
+          "value": "generation",
+          "markdown": false
+        },
+        {
+          "value": 1000.0,
+          "description": "min=1000, mean=1000, max=1000, sum=1000 (1)",
+          "markdown": false
+        },
+        {
+          "value": 1.0,
+          "description": "min=1, mean=1, max=1, sum=3 (3)",
+          "markdown": false
+        },
+        {
+          "value": 2818.746,
+          "description": "min=939.582, mean=939.582, max=939.582, sum=2818.746 (3)",
+          "markdown": false
+        },
+        {
+          "value": 505.377,
+          "description": "min=168.459, mean=168.459, max=168.459, sum=505.377 (3)",
+          "markdown": false
+        },
+        {
+          "value": 1,
+          "markdown": false
+        }
+      ],
+      [
+        {
+          "value": "MATH",
+          "href": "?group=math_chain_of_thought",
+          "markdown": false
+        },
+        {
+          "value": "The MATH benchmark for measuring mathematical problem solving on competition math problems with chain-of-thought style reasoning [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2103.03874.pdf).",
+          "markdown": true
+        },
+        {
+          "value": "",
+          "markdown": false
+        },
+        {
+          "markdown": false
+        },
+        {
+          "markdown": false
+        },
+        {
+          "markdown": false
+        },
+        {
+          "markdown": false
+        },
+        {
+          "value": 0,
+          "markdown": false
+        }
+      ],
+      [
+        {
+          "value": "LegalBench",
+          "href": "?group=legalbench",
+          "markdown": false
+        },
+        {
+          "value": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).",
+          "markdown": true
+        },
+        {
+          "value": "generation",
+          "markdown": false
+        },
+        {
+          "value": 409.4,
+          "description": "min=95, mean=409.4, max=1000, sum=2047 (5)",
+          "markdown": false
+        },
+        {
+          "value": 1.0,
+          "description": "min=1, mean=1, max=1, sum=15 (15)",
+          "markdown": false
+        },
+        {
+          "value": 8409.66107451482,
+          "description": "min=206.779, mean=560.644, max=1497.455, sum=8409.661 (15)",
+          "markdown": false
+        },
+        {
+          "value": 24.626875382305506,
+          "description": "min=1, mean=1.642, max=4.124, sum=24.627 (15)",
+          "markdown": false
+        },
+        {
+          "value": 1,
+          "markdown": false
+        }
+      ],
+      [
+        {
+          "value": "MedQA",
+          "href": "?group=med_qa",
+          "markdown": false
+        },
+        {
+          "value": "MedQA is an open domain question answering dataset composed of questions from professional medical board exams ([Jin et al. 2020](https://arxiv.org/pdf/2009.13081.pdf)).",
+          "markdown": true
+        },
+        {
+          "value": "",
+          "markdown": false
+        },
+        {
+          "markdown": false
+        },
+        {
+          "markdown": false
+        },
+        {
+          "markdown": false
+        },
+        {
+          "markdown": false
+        },
+        {
+          "value": 0,
+          "markdown": false
+        }
+      ],
+      [
+        {
+          "value": "WMT 2014",
+          "href": "?group=wmt_14",
+          "markdown": false
+        },
+        {
+          "value": "WMT 2014 is a collection of machine translation datasets [(website)](https://www.statmt.org/wmt14/index.html).",
+          "markdown": true
+        },
+        {
+          "value": "",
+          "markdown": false
+        },
+        {
+          "markdown": false
+        },
+        {
+          "markdown": false
+        },
+        {
+          "markdown": false
+        },
+        {
+          "markdown": false
+        },
+        {
+          "value": 0,
+          "markdown": false
+        }
+      ]
+    ],
+    "links": []
+  }
+]

lite_pythia-2.8b-step5000/groups/core_scenarios.json

@@ -0,0 +1,1264 @@
+[
+  {
+    "title": "Accuracy",
+    "header": [
+      {
+        "value": "Model",
+        "markdown": false,
+        "metadata": {}
+      },
+      {
+        "value": "Mean win rate",
+        "description": "How many models this model outperforms on average (over columns).",
+        "markdown": false,
+        "lower_is_better": false,
+        "metadata": {}
+      },
+      {
+        "value": "NarrativeQA - F1",
+        "description": "The NarrativeQA benchmark for reading comprehension over narratives [(Ko\u010disk\u00fd et al., 2017)](https://aclanthology.org/Q18-1023/).\n\nF1: Average F1 score in terms of word overlap between the model output and correct reference.",
+        "markdown": false,
+        "lower_is_better": false,
+        "metadata": {
+          "metric": "F1",
+          "run_group": "NarrativeQA"
+        }
+      },
+      {
+        "value": "NaturalQuestions (open-book) - F1",
+        "description": "The NaturalQuestions [(Kwiatkowski et al., 2019)](https://aclanthology.org/Q19-1026/) benchmark for question answering based on naturally-occurring queries through Google Search. The input includes the Wikipedia page with the answer.\n\nF1: Average F1 score in terms of word overlap between the model output and correct reference.",
+        "markdown": false,
+        "lower_is_better": false,
+        "metadata": {
+          "metric": "F1",
+          "run_group": "NaturalQuestions (open-book)"
+        }
+      },
+      {
+        "value": "NaturalQuestions (closed-book) - F1",
+        "description": "The NaturalQuestions [(Kwiatkowski et al., 2019)](https://aclanthology.org/Q19-1026/) benchmark for question answering based on naturally-occurring queries through Google Search. The input does not include the Wikipedia page with the answer.\n\nF1: Average F1 score in terms of word overlap between the model output and correct reference.",
+        "markdown": false,
+        "lower_is_better": false,
+        "metadata": {
+          "metric": "F1",
+          "run_group": "NaturalQuestions (closed-book)"
+        }
+      },
+      {
+        "value": "OpenbookQA - EM",
+        "description": "The OpenbookQA benchmark for commonsense-intensive open book question answering [(Mihaylov et al., 2018)](https://aclanthology.org/D18-1260/).\n\nExact match: Fraction of instances that the predicted output matches a correct reference exactly.",
+        "markdown": false,
+        "lower_is_better": false,
+        "metadata": {
+          "metric": "EM",
+          "run_group": "OpenbookQA"
+        }
+      },
+      {
+        "value": "MMLU - EM",
+        "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\nExact match: Fraction of instances that the predicted output matches a correct reference exactly.",
+        "markdown": false,
+        "lower_is_better": false,
+        "metadata": {
+          "metric": "EM",
+          "run_group": "MMLU"
+        }
+      },
+      {
+        "value": "MATH - Equivalent (CoT)",
+        "description": "The MATH benchmark for measuring mathematical problem solving on competition math problems with chain-of-thought style reasoning [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2103.03874.pdf).\n\nEquivalent (CoT): Fraction of model outputs that are mathematically equivalent to the correct reference when using chain-of-thought prompting.",
+        "markdown": false,
+        "lower_is_better": false,
+        "metadata": {
+          "metric": "Equivalent (CoT)",
+          "run_group": "MATH"
+        }
+      },
+      {
+        "value": "GSM8K - EM",
+        "description": "The grade school math word problems dataset (GSM8K) for testing mathematical reasoning on grade-school math problems [(Cobbe et al., 2021)](https://arxiv.org/pdf/2110.14168.pdf).\n\nExact match (final number): Fraction of instances that the predicted output matches a correct reference exactly, ignoring text preceding the specified indicator.",
+        "markdown": false,
+        "lower_is_better": false,
+        "metadata": {
+          "metric": "EM",
+          "run_group": "GSM8K"
+        }
+      },
+      {
+        "value": "LegalBench - EM",
+        "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\nQuasi-exact match: Fraction of instances that the predicted output matches a correct reference up to light processing.",
+        "markdown": false,
+        "lower_is_better": false,
+        "metadata": {
+          "metric": "EM",
+          "run_group": "LegalBench"
+        }
+      },
+      {
+        "value": "MedQA - EM",
+        "description": "MedQA is an open domain question answering dataset composed of questions from professional medical board exams ([Jin et al. 2020](https://arxiv.org/pdf/2009.13081.pdf)).\n\nQuasi-exact match: Fraction of instances that the predicted output matches a correct reference up to light processing.",
+        "markdown": false,
+        "lower_is_better": false,
+        "metadata": {
+          "metric": "EM",
+          "run_group": "MedQA"
+        }
+      },
+      {
+        "value": "WMT 2014 - BLEU-4",
+        "description": "WMT 2014 is a collection of machine translation datasets [(website)](https://www.statmt.org/wmt14/index.html).\n\nBLEU-4: Average BLEU score [(Papineni et al., 2002)](https://aclanthology.org/P02-1040/) based on 4-gram overlap.",
+        "markdown": false,
+        "lower_is_better": false,
+        "metadata": {
+          "metric": "BLEU-4",
+          "run_group": "WMT 2014"
+        }
+      }
+    ],
+    "rows": [
+      [
+        {
+          "value": "EleutherAI/pythia-2.8b",
+          "description": "",
+          "markdown": false
+        },
+        {
+          "markdown": false
+        },
+        {
+          "description": "No matching runs",
+          "markdown": false
+        },
+        {
+          "description": "No matching runs",
+          "markdown": false
+        },
+        {
+          "description": "No matching runs",
+          "markdown": false
+        },
+        {
+          "value": 0.22,
+          "description": "min=0.22, mean=0.22, max=0.22, sum=0.22 (1)",
+          "style": {
+            "font-weight": "bold"
+          },
+          "markdown": false,
+          "run_spec_names": [
+            "commonsense:dataset=openbookqa,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b"
+          ]
+        },
+        {
+          "value": 0.27487719298245616,
+          "description": "min=0.25, mean=0.275, max=0.32, sum=1.374 (5)",
+          "style": {
+            "font-weight": "bold"
+          },
+          "markdown": false,
+          "run_spec_names": [
+            "mmlu:subject=abstract_algebra,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=college_chemistry,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=computer_security,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=econometrics,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=us_foreign_policy,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b"
+          ]
+        },
+        {
+          "description": "No matching runs",
+          "markdown": false
+        },
+        {
+          "value": 0.01,
+          "description": "min=0.01, mean=0.01, max=0.01, sum=0.01 (1)",
+          "style": {
+            "font-weight": "bold"
+          },
+          "markdown": false,
+          "run_spec_names": [
+            "gsm:model=EleutherAI_pythia-2.8b"
+          ]
+        },
+        {
+          "value": 0.3189986232611502,
+          "description": "min=0.144, mean=0.319, max=0.619, sum=1.595 (5)",
+          "style": {
+            "font-weight": "bold"
+          },
+          "markdown": false,
+          "run_spec_names": [
+            "legalbench:subset=abercrombie,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=corporate_lobbying,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=function_of_decision_section,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=international_citizenship_questions,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=proa,model=EleutherAI_pythia-2.8b"
+          ]
+        },
+        {
+          "description": "No matching runs",
+          "markdown": false
+        },
+        {
+          "description": "No matching runs",
+          "markdown": false
+        }
+      ]
+    ],
+    "links": [
+      {
+        "text": "LaTeX",
+        "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/latex/core_scenarios_accuracy.tex"
+      },
+      {
+        "text": "JSON",
+        "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/json/core_scenarios_accuracy.json"
+      }
+    ],
+    "name": "accuracy"
+  },
+  {
+    "title": "Efficiency",
+    "header": [
+      {
+        "value": "Model",
+        "markdown": false,
+        "metadata": {}
+      },
+      {
+        "value": "Mean win rate",
+        "description": "How many models this model outperforms on average (over columns).",
+        "markdown": false,
+        "lower_is_better": false,
+        "metadata": {}
+      },
+      {
+        "value": "NarrativeQA - Observed inference time (s)",
+        "description": "The NarrativeQA benchmark for reading comprehension over narratives [(Ko\u010disk\u00fd et al., 2017)](https://aclanthology.org/Q18-1023/).\n\nObserved inference runtime (s): Average observed time to process a request to the model (via an API, and thus depends on particular deployment).",
+        "markdown": false,
+        "lower_is_better": true,
+        "metadata": {
+          "metric": "Observed inference time (s)",
+          "run_group": "NarrativeQA"
+        }
+      },
+      {
+        "value": "NaturalQuestions (open-book) - Observed inference time (s)",
+        "description": "The NaturalQuestions [(Kwiatkowski et al., 2019)](https://aclanthology.org/Q19-1026/) benchmark for question answering based on naturally-occurring queries through Google Search. The input includes the Wikipedia page with the answer.\n\nObserved inference runtime (s): Average observed time to process a request to the model (via an API, and thus depends on particular deployment).",
+        "markdown": false,
+        "lower_is_better": true,
+        "metadata": {
+          "metric": "Observed inference time (s)",
+          "run_group": "NaturalQuestions (open-book)"
+        }
+      },
+      {
+        "value": "NaturalQuestions (closed-book) - Observed inference time (s)",
+        "description": "The NaturalQuestions [(Kwiatkowski et al., 2019)](https://aclanthology.org/Q19-1026/) benchmark for question answering based on naturally-occurring queries through Google Search. The input does not include the Wikipedia page with the answer.\n\nObserved inference runtime (s): Average observed time to process a request to the model (via an API, and thus depends on particular deployment).",
+        "markdown": false,
+        "lower_is_better": true,
+        "metadata": {
+          "metric": "Observed inference time (s)",
+          "run_group": "NaturalQuestions (closed-book)"
+        }
+      },
+      {
+        "value": "OpenbookQA - Observed inference time (s)",
+        "description": "The OpenbookQA benchmark for commonsense-intensive open book question answering [(Mihaylov et al., 2018)](https://aclanthology.org/D18-1260/).\n\nObserved inference runtime (s): Average observed time to process a request to the model (via an API, and thus depends on particular deployment).",
+        "markdown": false,
+        "lower_is_better": true,
+        "metadata": {
+          "metric": "Observed inference time (s)",
+          "run_group": "OpenbookQA"
+        }
+      },
+      {
+        "value": "MMLU - Observed inference time (s)",
+        "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\nObserved inference runtime (s): Average observed time to process a request to the model (via an API, and thus depends on particular deployment).",
+        "markdown": false,
+        "lower_is_better": true,
+        "metadata": {
+          "metric": "Observed inference time (s)",
+          "run_group": "MMLU"
+        }
+      },
+      {
+        "value": "MATH - Observed inference time (s)",
+        "description": "The MATH benchmark for measuring mathematical problem solving on competition math problems with chain-of-thought style reasoning [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2103.03874.pdf).\n\nObserved inference runtime (s): Average observed time to process a request to the model (via an API, and thus depends on particular deployment).",
+        "markdown": false,
+        "lower_is_better": true,
+        "metadata": {
+          "metric": "Observed inference time (s)",
+          "run_group": "MATH"
+        }
+      },
+      {
+        "value": "GSM8K - Observed inference time (s)",
+        "description": "The grade school math word problems dataset (GSM8K) for testing mathematical reasoning on grade-school math problems [(Cobbe et al., 2021)](https://arxiv.org/pdf/2110.14168.pdf).\n\nObserved inference runtime (s): Average observed time to process a request to the model (via an API, and thus depends on particular deployment).",
+        "markdown": false,
+        "lower_is_better": true,
+        "metadata": {
+          "metric": "Observed inference time (s)",
+          "run_group": "GSM8K"
+        }
+      },
+      {
+        "value": "LegalBench - Observed inference time (s)",
+        "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\nObserved inference runtime (s): Average observed time to process a request to the model (via an API, and thus depends on particular deployment).",
+        "markdown": false,
+        "lower_is_better": true,
+        "metadata": {
+          "metric": "Observed inference time (s)",
+          "run_group": "LegalBench"
+        }
+      },
+      {
+        "value": "MedQA - Observed inference time (s)",
+        "description": "MedQA is an open domain question answering dataset composed of questions from professional medical board exams ([Jin et al. 2020](https://arxiv.org/pdf/2009.13081.pdf)).\n\nObserved inference runtime (s): Average observed time to process a request to the model (via an API, and thus depends on particular deployment).",
+        "markdown": false,
+        "lower_is_better": true,
+        "metadata": {
+          "metric": "Observed inference time (s)",
+          "run_group": "MedQA"
+        }
+      },
+      {
+        "value": "WMT 2014 - Observed inference time (s)",
+        "description": "WMT 2014 is a collection of machine translation datasets [(website)](https://www.statmt.org/wmt14/index.html).\n\nObserved inference runtime (s): Average observed time to process a request to the model (via an API, and thus depends on particular deployment).",
+        "markdown": false,
+        "lower_is_better": true,
+        "metadata": {
+          "metric": "Observed inference time (s)",
+          "run_group": "WMT 2014"
+        }
+      }
+    ],
+    "rows": [
+      [
+        {
+          "value": "EleutherAI/pythia-2.8b",
+          "description": "",
+          "markdown": false
+        },
+        {
+          "markdown": false
+        },
+        {
+          "description": "No matching runs",
+          "markdown": false
+        },
+        {
+          "description": "No matching runs",
+          "markdown": false
+        },
+        {
+          "description": "No matching runs",
+          "markdown": false
+        },
+        {
+          "value": 0.12883714818954467,
+          "description": "min=0.129, mean=0.129, max=0.129, sum=0.129 (1)",
+          "style": {
+            "font-weight": "bold"
+          },
+          "markdown": false,
+          "run_spec_names": [
+            "commonsense:dataset=openbookqa,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b"
+          ]
+        },
+        {
+          "value": 0.22061018314696193,
+          "description": "min=0.173, mean=0.221, max=0.282, sum=1.103 (5)",
+          "style": {
+            "font-weight": "bold"
+          },
+          "markdown": false,
+          "run_spec_names": [
+            "mmlu:subject=abstract_algebra,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=college_chemistry,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=computer_security,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=econometrics,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=us_foreign_policy,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b"
+          ]
+        },
+        {
+          "description": "No matching runs",
+          "markdown": false
+        },
+        {
+          "value": 2.5314217054843904,
+          "description": "min=2.531, mean=2.531, max=2.531, sum=2.531 (1)",
+          "style": {
+            "font-weight": "bold"
+          },
+          "markdown": false,
+          "run_spec_names": [
+            "gsm:model=EleutherAI_pythia-2.8b"
+          ]
+        },
+        {
+          "value": 0.3094366045219278,
+          "description": "min=0.141, mean=0.309, max=0.795, sum=1.547 (5)",
+          "style": {
+            "font-weight": "bold"
+          },
+          "markdown": false,
+          "run_spec_names": [
+            "legalbench:subset=abercrombie,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=corporate_lobbying,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=function_of_decision_section,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=international_citizenship_questions,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=proa,model=EleutherAI_pythia-2.8b"
+          ]
+        },
+        {
+          "description": "No matching runs",
+          "markdown": false
+        },
+        {
+          "description": "No matching runs",
+          "markdown": false
+        }
+      ]
+    ],
+    "links": [
+      {
+        "text": "LaTeX",
+        "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/latex/core_scenarios_efficiency.tex"
+      },
+      {
+        "text": "JSON",
+        "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/json/core_scenarios_efficiency.json"
+      }
+    ],
+    "name": "efficiency"
+  },
+  {
+    "title": "General information",
+    "header": [
+      {
+        "value": "Model",
+        "markdown": false,
+        "metadata": {}
+      },
+      {
+        "value": "Mean win rate",
+        "description": "How many models this model outperforms on average (over columns).",
+        "markdown": false,
+        "lower_is_better": false,
+        "metadata": {}
+      },
+      {
+        "value": "NarrativeQA - # eval",
+        "description": "The NarrativeQA benchmark for reading comprehension over narratives [(Ko\u010disk\u00fd et al., 2017)](https://aclanthology.org/Q18-1023/).\n\n# eval: Number of evaluation instances.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# eval",
+          "run_group": "NarrativeQA"
+        }
+      },
+      {
+        "value": "NarrativeQA - # train",
+        "description": "The NarrativeQA benchmark for reading comprehension over narratives [(Ko\u010disk\u00fd et al., 2017)](https://aclanthology.org/Q18-1023/).\n\n# train: Number of training instances (e.g., in-context examples).",
+        "markdown": false,
+        "metadata": {
+          "metric": "# train",
+          "run_group": "NarrativeQA"
+        }
+      },
+      {
+        "value": "NarrativeQA - truncated",
+        "description": "The NarrativeQA benchmark for reading comprehension over narratives [(Ko\u010disk\u00fd et al., 2017)](https://aclanthology.org/Q18-1023/).\n\ntruncated: Fraction of instances where the prompt itself was truncated (implies that there were no in-context examples).",
+        "markdown": false,
+        "metadata": {
+          "metric": "truncated",
+          "run_group": "NarrativeQA"
+        }
+      },
+      {
+        "value": "NarrativeQA - # prompt tokens",
+        "description": "The NarrativeQA benchmark for reading comprehension over narratives [(Ko\u010disk\u00fd et al., 2017)](https://aclanthology.org/Q18-1023/).\n\n# prompt tokens: Number of tokens in the prompt.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# prompt tokens",
+          "run_group": "NarrativeQA"
+        }
+      },
+      {
+        "value": "NarrativeQA - # output tokens",
+        "description": "The NarrativeQA benchmark for reading comprehension over narratives [(Ko\u010disk\u00fd et al., 2017)](https://aclanthology.org/Q18-1023/).\n\n# output tokens: Actual number of output tokens.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# output tokens",
+          "run_group": "NarrativeQA"
+        }
+      },
+      {
+        "value": "NaturalQuestions (open-book) - # eval",
+        "description": "The NaturalQuestions [(Kwiatkowski et al., 2019)](https://aclanthology.org/Q19-1026/) benchmark for question answering based on naturally-occurring queries through Google Search. The input includes the Wikipedia page with the answer.\n\n# eval: Number of evaluation instances.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# eval",
+          "run_group": "NaturalQuestions (open-book)"
+        }
+      },
+      {
+        "value": "NaturalQuestions (open-book) - # train",
+        "description": "The NaturalQuestions [(Kwiatkowski et al., 2019)](https://aclanthology.org/Q19-1026/) benchmark for question answering based on naturally-occurring queries through Google Search. The input includes the Wikipedia page with the answer.\n\n# train: Number of training instances (e.g., in-context examples).",
+        "markdown": false,
+        "metadata": {
+          "metric": "# train",
+          "run_group": "NaturalQuestions (open-book)"
+        }
+      },
+      {
+        "value": "NaturalQuestions (open-book) - truncated",
+        "description": "The NaturalQuestions [(Kwiatkowski et al., 2019)](https://aclanthology.org/Q19-1026/) benchmark for question answering based on naturally-occurring queries through Google Search. The input includes the Wikipedia page with the answer.\n\ntruncated: Fraction of instances where the prompt itself was truncated (implies that there were no in-context examples).",
+        "markdown": false,
+        "metadata": {
+          "metric": "truncated",
+          "run_group": "NaturalQuestions (open-book)"
+        }
+      },
+      {
+        "value": "NaturalQuestions (open-book) - # prompt tokens",
+        "description": "The NaturalQuestions [(Kwiatkowski et al., 2019)](https://aclanthology.org/Q19-1026/) benchmark for question answering based on naturally-occurring queries through Google Search. The input includes the Wikipedia page with the answer.\n\n# prompt tokens: Number of tokens in the prompt.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# prompt tokens",
+          "run_group": "NaturalQuestions (open-book)"
+        }
+      },
+      {
+        "value": "NaturalQuestions (open-book) - # output tokens",
+        "description": "The NaturalQuestions [(Kwiatkowski et al., 2019)](https://aclanthology.org/Q19-1026/) benchmark for question answering based on naturally-occurring queries through Google Search. The input includes the Wikipedia page with the answer.\n\n# output tokens: Actual number of output tokens.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# output tokens",
+          "run_group": "NaturalQuestions (open-book)"
+        }
+      },
+      {
+        "value": "NaturalQuestions (closed-book) - # eval",
+        "description": "The NaturalQuestions [(Kwiatkowski et al., 2019)](https://aclanthology.org/Q19-1026/) benchmark for question answering based on naturally-occurring queries through Google Search. The input does not include the Wikipedia page with the answer.\n\n# eval: Number of evaluation instances.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# eval",
+          "run_group": "NaturalQuestions (closed-book)"
+        }
+      },
+      {
+        "value": "NaturalQuestions (closed-book) - # train",
+        "description": "The NaturalQuestions [(Kwiatkowski et al., 2019)](https://aclanthology.org/Q19-1026/) benchmark for question answering based on naturally-occurring queries through Google Search. The input does not include the Wikipedia page with the answer.\n\n# train: Number of training instances (e.g., in-context examples).",
+        "markdown": false,
+        "metadata": {
+          "metric": "# train",
+          "run_group": "NaturalQuestions (closed-book)"
+        }
+      },
+      {
+        "value": "NaturalQuestions (closed-book) - truncated",
+        "description": "The NaturalQuestions [(Kwiatkowski et al., 2019)](https://aclanthology.org/Q19-1026/) benchmark for question answering based on naturally-occurring queries through Google Search. The input does not include the Wikipedia page with the answer.\n\ntruncated: Fraction of instances where the prompt itself was truncated (implies that there were no in-context examples).",
+        "markdown": false,
+        "metadata": {
+          "metric": "truncated",
+          "run_group": "NaturalQuestions (closed-book)"
+        }
+      },
+      {
+        "value": "NaturalQuestions (closed-book) - # prompt tokens",
+        "description": "The NaturalQuestions [(Kwiatkowski et al., 2019)](https://aclanthology.org/Q19-1026/) benchmark for question answering based on naturally-occurring queries through Google Search. The input does not include the Wikipedia page with the answer.\n\n# prompt tokens: Number of tokens in the prompt.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# prompt tokens",
+          "run_group": "NaturalQuestions (closed-book)"
+        }
+      },
+      {
+        "value": "NaturalQuestions (closed-book) - # output tokens",
+        "description": "The NaturalQuestions [(Kwiatkowski et al., 2019)](https://aclanthology.org/Q19-1026/) benchmark for question answering based on naturally-occurring queries through Google Search. The input does not include the Wikipedia page with the answer.\n\n# output tokens: Actual number of output tokens.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# output tokens",
+          "run_group": "NaturalQuestions (closed-book)"
+        }
+      },
+      {
+        "value": "OpenbookQA - # eval",
+        "description": "The OpenbookQA benchmark for commonsense-intensive open book question answering [(Mihaylov et al., 2018)](https://aclanthology.org/D18-1260/).\n\n# eval: Number of evaluation instances.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# eval",
+          "run_group": "OpenbookQA"
+        }
+      },
+      {
+        "value": "OpenbookQA - # train",
+        "description": "The OpenbookQA benchmark for commonsense-intensive open book question answering [(Mihaylov et al., 2018)](https://aclanthology.org/D18-1260/).\n\n# train: Number of training instances (e.g., in-context examples).",
+        "markdown": false,
+        "metadata": {
+          "metric": "# train",
+          "run_group": "OpenbookQA"
+        }
+      },
+      {
+        "value": "OpenbookQA - truncated",
+        "description": "The OpenbookQA benchmark for commonsense-intensive open book question answering [(Mihaylov et al., 2018)](https://aclanthology.org/D18-1260/).\n\ntruncated: Fraction of instances where the prompt itself was truncated (implies that there were no in-context examples).",
+        "markdown": false,
+        "metadata": {
+          "metric": "truncated",
+          "run_group": "OpenbookQA"
+        }
+      },
+      {
+        "value": "OpenbookQA - # prompt tokens",
+        "description": "The OpenbookQA benchmark for commonsense-intensive open book question answering [(Mihaylov et al., 2018)](https://aclanthology.org/D18-1260/).\n\n# prompt tokens: Number of tokens in the prompt.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# prompt tokens",
+          "run_group": "OpenbookQA"
+        }
+      },
+      {
+        "value": "OpenbookQA - # output tokens",
+        "description": "The OpenbookQA benchmark for commonsense-intensive open book question answering [(Mihaylov et al., 2018)](https://aclanthology.org/D18-1260/).\n\n# output tokens: Actual number of output tokens.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# output tokens",
+          "run_group": "OpenbookQA"
+        }
+      },
+      {
+        "value": "MMLU - # eval",
+        "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# eval: Number of evaluation instances.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# eval",
+          "run_group": "MMLU"
+        }
+      },
+      {
+        "value": "MMLU - # train",
+        "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# train: Number of training instances (e.g., in-context examples).",
+        "markdown": false,
+        "metadata": {
+          "metric": "# train",
+          "run_group": "MMLU"
+        }
+      },
+      {
+        "value": "MMLU - truncated",
+        "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\ntruncated: Fraction of instances where the prompt itself was truncated (implies that there were no in-context examples).",
+        "markdown": false,
+        "metadata": {
+          "metric": "truncated",
+          "run_group": "MMLU"
+        }
+      },
+      {
+        "value": "MMLU - # prompt tokens",
+        "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# prompt tokens: Number of tokens in the prompt.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# prompt tokens",
+          "run_group": "MMLU"
+        }
+      },
+      {
+        "value": "MMLU - # output tokens",
+        "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# output tokens: Actual number of output tokens.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# output tokens",
+          "run_group": "MMLU"
+        }
+      },
+      {
+        "value": "MATH - # eval",
+        "description": "The MATH benchmark for measuring mathematical problem solving on competition math problems with chain-of-thought style reasoning [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2103.03874.pdf).\n\n# eval: Number of evaluation instances.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# eval",
+          "run_group": "MATH"
+        }
+      },
+      {
+        "value": "MATH - # train",
+        "description": "The MATH benchmark for measuring mathematical problem solving on competition math problems with chain-of-thought style reasoning [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2103.03874.pdf).\n\n# train: Number of training instances (e.g., in-context examples).",
+        "markdown": false,
+        "metadata": {
+          "metric": "# train",
+          "run_group": "MATH"
+        }
+      },
+      {
+        "value": "MATH - truncated",
+        "description": "The MATH benchmark for measuring mathematical problem solving on competition math problems with chain-of-thought style reasoning [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2103.03874.pdf).\n\ntruncated: Fraction of instances where the prompt itself was truncated (implies that there were no in-context examples).",
+        "markdown": false,
+        "metadata": {
+          "metric": "truncated",
+          "run_group": "MATH"
+        }
+      },
+      {
+        "value": "MATH - # prompt tokens",
+        "description": "The MATH benchmark for measuring mathematical problem solving on competition math problems with chain-of-thought style reasoning [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2103.03874.pdf).\n\n# prompt tokens: Number of tokens in the prompt.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# prompt tokens",
+          "run_group": "MATH"
+        }
+      },
+      {
+        "value": "MATH - # output tokens",
+        "description": "The MATH benchmark for measuring mathematical problem solving on competition math problems with chain-of-thought style reasoning [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2103.03874.pdf).\n\n# output tokens: Actual number of output tokens.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# output tokens",
+          "run_group": "MATH"
+        }
+      },
+      {
+        "value": "GSM8K - # eval",
+        "description": "The grade school math word problems dataset (GSM8K) for testing mathematical reasoning on grade-school math problems [(Cobbe et al., 2021)](https://arxiv.org/pdf/2110.14168.pdf).\n\n# eval: Number of evaluation instances.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# eval",
+          "run_group": "GSM8K"
+        }
+      },
+      {
+        "value": "GSM8K - # train",
+        "description": "The grade school math word problems dataset (GSM8K) for testing mathematical reasoning on grade-school math problems [(Cobbe et al., 2021)](https://arxiv.org/pdf/2110.14168.pdf).\n\n# train: Number of training instances (e.g., in-context examples).",
+        "markdown": false,
+        "metadata": {
+          "metric": "# train",
+          "run_group": "GSM8K"
+        }
+      },
+      {
+        "value": "GSM8K - truncated",
+        "description": "The grade school math word problems dataset (GSM8K) for testing mathematical reasoning on grade-school math problems [(Cobbe et al., 2021)](https://arxiv.org/pdf/2110.14168.pdf).\n\ntruncated: Fraction of instances where the prompt itself was truncated (implies that there were no in-context examples).",
+        "markdown": false,
+        "metadata": {
+          "metric": "truncated",
+          "run_group": "GSM8K"
+        }
+      },
+      {
+        "value": "GSM8K - # prompt tokens",
+        "description": "The grade school math word problems dataset (GSM8K) for testing mathematical reasoning on grade-school math problems [(Cobbe et al., 2021)](https://arxiv.org/pdf/2110.14168.pdf).\n\n# prompt tokens: Number of tokens in the prompt.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# prompt tokens",
+          "run_group": "GSM8K"
+        }
+      },
+      {
+        "value": "GSM8K - # output tokens",
+        "description": "The grade school math word problems dataset (GSM8K) for testing mathematical reasoning on grade-school math problems [(Cobbe et al., 2021)](https://arxiv.org/pdf/2110.14168.pdf).\n\n# output tokens: Actual number of output tokens.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# output tokens",
+          "run_group": "GSM8K"
+        }
+      },
+      {
+        "value": "LegalBench - # eval",
+        "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# eval: Number of evaluation instances.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# eval",
+          "run_group": "LegalBench"
+        }
+      },
+      {
+        "value": "LegalBench - # train",
+        "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# train: Number of training instances (e.g., in-context examples).",
+        "markdown": false,
+        "metadata": {
+          "metric": "# train",
+          "run_group": "LegalBench"
+        }
+      },
+      {
+        "value": "LegalBench - truncated",
+        "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\ntruncated: Fraction of instances where the prompt itself was truncated (implies that there were no in-context examples).",
+        "markdown": false,
+        "metadata": {
+          "metric": "truncated",
+          "run_group": "LegalBench"
+        }
+      },
+      {
+        "value": "LegalBench - # prompt tokens",
+        "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# prompt tokens: Number of tokens in the prompt.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# prompt tokens",
+          "run_group": "LegalBench"
+        }
+      },
+      {
+        "value": "LegalBench - # output tokens",
+        "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# output tokens: Actual number of output tokens.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# output tokens",
+          "run_group": "LegalBench"
+        }
+      },
+      {
+        "value": "MedQA - # eval",
+        "description": "MedQA is an open domain question answering dataset composed of questions from professional medical board exams ([Jin et al. 2020](https://arxiv.org/pdf/2009.13081.pdf)).\n\n# eval: Number of evaluation instances.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# eval",
+          "run_group": "MedQA"
+        }
+      },
+      {
+        "value": "MedQA - # train",
+        "description": "MedQA is an open domain question answering dataset composed of questions from professional medical board exams ([Jin et al. 2020](https://arxiv.org/pdf/2009.13081.pdf)).\n\n# train: Number of training instances (e.g., in-context examples).",
+        "markdown": false,
+        "metadata": {
+          "metric": "# train",
+          "run_group": "MedQA"
+        }
+      },
+      {
+        "value": "MedQA - truncated",
+        "description": "MedQA is an open domain question answering dataset composed of questions from professional medical board exams ([Jin et al. 2020](https://arxiv.org/pdf/2009.13081.pdf)).\n\ntruncated: Fraction of instances where the prompt itself was truncated (implies that there were no in-context examples).",
+        "markdown": false,
+        "metadata": {
+          "metric": "truncated",
+          "run_group": "MedQA"
+        }
+      },
+      {
+        "value": "MedQA - # prompt tokens",
+        "description": "MedQA is an open domain question answering dataset composed of questions from professional medical board exams ([Jin et al. 2020](https://arxiv.org/pdf/2009.13081.pdf)).\n\n# prompt tokens: Number of tokens in the prompt.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# prompt tokens",
+          "run_group": "MedQA"
+        }
+      },
+      {
+        "value": "MedQA - # output tokens",
+        "description": "MedQA is an open domain question answering dataset composed of questions from professional medical board exams ([Jin et al. 2020](https://arxiv.org/pdf/2009.13081.pdf)).\n\n# output tokens: Actual number of output tokens.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# output tokens",
+          "run_group": "MedQA"
+        }
+      },
+      {
+        "value": "WMT 2014 - # eval",
+        "description": "WMT 2014 is a collection of machine translation datasets [(website)](https://www.statmt.org/wmt14/index.html).\n\n# eval: Number of evaluation instances.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# eval",
+          "run_group": "WMT 2014"
+        }
+      },
+      {
+        "value": "WMT 2014 - # train",
+        "description": "WMT 2014 is a collection of machine translation datasets [(website)](https://www.statmt.org/wmt14/index.html).\n\n# train: Number of training instances (e.g., in-context examples).",
+        "markdown": false,
+        "metadata": {
+          "metric": "# train",
+          "run_group": "WMT 2014"
+        }
+      },
+      {
+        "value": "WMT 2014 - truncated",
+        "description": "WMT 2014 is a collection of machine translation datasets [(website)](https://www.statmt.org/wmt14/index.html).\n\ntruncated: Fraction of instances where the prompt itself was truncated (implies that there were no in-context examples).",
+        "markdown": false,
+        "metadata": {
+          "metric": "truncated",
+          "run_group": "WMT 2014"
+        }
+      },
+      {
+        "value": "WMT 2014 - # prompt tokens",
+        "description": "WMT 2014 is a collection of machine translation datasets [(website)](https://www.statmt.org/wmt14/index.html).\n\n# prompt tokens: Number of tokens in the prompt.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# prompt tokens",
+          "run_group": "WMT 2014"
+        }
+      },
+      {
+        "value": "WMT 2014 - # output tokens",
+        "description": "WMT 2014 is a collection of machine translation datasets [(website)](https://www.statmt.org/wmt14/index.html).\n\n# output tokens: Actual number of output tokens.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# output tokens",
+          "run_group": "WMT 2014"
+        }
+      }
+    ],
+    "rows": [
+      [
+        {
+          "value": "EleutherAI/pythia-2.8b",
+          "description": "",
+          "markdown": false
+        },
+        {
+          "markdown": false
+        },
+        {
+          "description": "No matching runs",
+          "markdown": false
+        },
+        {
+          "description": "No matching runs",
+          "markdown": false
+        },
+        {
+          "description": "No matching runs",
+          "markdown": false
+        },
+        {
+          "description": "No matching runs",
+          "markdown": false
+        },
+        {
+          "description": "No matching runs",
+          "markdown": false
+        },
+        {
+          "description": "No matching runs",
+          "markdown": false
+        },
+        {
+          "description": "No matching runs",
+          "markdown": false
+        },
+        {
+          "description": "No matching runs",
+          "markdown": false
+        },
+        {
+          "description": "No matching runs",
+          "markdown": false
+        },
+        {
+          "description": "No matching runs",
+          "markdown": false
+        },
+        {
+          "description": "No matching runs",
+          "markdown": false
+        },
+        {
+          "description": "No matching runs",
+          "markdown": false
+        },
+        {
+          "description": "No matching runs",
+          "markdown": false
+        },
+        {
+          "description": "No matching runs",
+          "markdown": false
+        },
+        {
+          "description": "No matching runs",
+          "markdown": false
+        },
+        {
+          "value": 500.0,
+          "description": "min=500, mean=500, max=500, sum=500 (1)",
+          "style": {},
+          "markdown": false,
+          "run_spec_names": [
+            "commonsense:dataset=openbookqa,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b"
+          ]
+        },
+        {
+          "value": 5.0,
+          "description": "min=5, mean=5, max=5, sum=5 (1)",
+          "style": {},
+          "markdown": false,
+          "run_spec_names": [
+            "commonsense:dataset=openbookqa,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b"
+          ]
+        },
+        {
+          "value": 0.0,
+          "description": "min=0, mean=0, max=0, sum=0 (1)",
+          "style": {},
+          "markdown": false,
+          "run_spec_names": [
+            "commonsense:dataset=openbookqa,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b"
+          ]
+        },
+        {
+          "value": 251.556,
+          "description": "min=251.556, mean=251.556, max=251.556, sum=251.556 (1)",
+          "style": {},
+          "markdown": false,
+          "run_spec_names": [
+            "commonsense:dataset=openbookqa,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b"
+          ]
+        },
+        {
+          "value": 1.0,
+          "description": "min=1, mean=1, max=1, sum=1 (1)",
+          "style": {},
+          "markdown": false,
+          "run_spec_names": [
+            "commonsense:dataset=openbookqa,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b"
+          ]
+        },
+        {
+          "value": 102.8,
+          "description": "min=100, mean=102.8, max=114, sum=514 (5)",
+          "style": {},
+          "markdown": false,
+          "run_spec_names": [
+            "mmlu:subject=abstract_algebra,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=college_chemistry,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=computer_security,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=econometrics,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=us_foreign_policy,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b"
+          ]
+        },
+        {
+          "value": 5.0,
+          "description": "min=5, mean=5, max=5, sum=25 (5)",
+          "style": {},
+          "markdown": false,
+          "run_spec_names": [
+            "mmlu:subject=abstract_algebra,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=college_chemistry,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=computer_security,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=econometrics,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=us_foreign_policy,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b"
+          ]
+        },
+        {
+          "value": 0.0,
+          "description": "min=0, mean=0, max=0, sum=0 (5)",
+          "style": {},
+          "markdown": false,
+          "run_spec_names": [
+            "mmlu:subject=abstract_algebra,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=college_chemistry,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=computer_security,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=econometrics,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=us_foreign_policy,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b"
+          ]
+        },
+        {
+          "value": 467.935649122807,
+          "description": "min=358.76, mean=467.936, max=612.798, sum=2339.678 (5)",
+          "style": {},
+          "markdown": false,
+          "run_spec_names": [
+            "mmlu:subject=abstract_algebra,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=college_chemistry,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=computer_security,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=econometrics,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=us_foreign_policy,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b"
+          ]
+        },
+        {
+          "value": 1.0,
+          "description": "min=1, mean=1, max=1, sum=5 (5)",
+          "style": {},
+          "markdown": false,
+          "run_spec_names": [
+            "mmlu:subject=abstract_algebra,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=college_chemistry,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=computer_security,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=econometrics,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=us_foreign_policy,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b"
+          ]
+        },
+        {
+          "description": "No matching runs",
+          "markdown": false
+        },
+        {
+          "description": "No matching runs",
+          "markdown": false
+        },
+        {
+          "description": "No matching runs",
+          "markdown": false
+        },
+        {
+          "description": "No matching runs",
+          "markdown": false
+        },
+        {
+          "description": "No matching runs",
+          "markdown": false
+        },
+        {
+          "value": 1000.0,
+          "description": "min=1000, mean=1000, max=1000, sum=1000 (1)",
+          "style": {},
+          "markdown": false,
+          "run_spec_names": [
+            "gsm:model=EleutherAI_pythia-2.8b"
+          ]
+        },
+        {
+          "value": 5.0,
+          "description": "min=5, mean=5, max=5, sum=5 (1)",
+          "style": {},
+          "markdown": false,
+          "run_spec_names": [
+            "gsm:model=EleutherAI_pythia-2.8b"
+          ]
+        },
+        {
+          "value": 0.0,
+          "description": "min=0, mean=0, max=0, sum=0 (1)",
+          "style": {},
+          "markdown": false,
+          "run_spec_names": [
+            "gsm:model=EleutherAI_pythia-2.8b"
+          ]
+        },
+        {
+          "value": 939.582,
+          "description": "min=939.582, mean=939.582, max=939.582, sum=939.582 (1)",
+          "style": {},
+          "markdown": false,
+          "run_spec_names": [
+            "gsm:model=EleutherAI_pythia-2.8b"
+          ]
+        },
+        {
+          "value": 168.459,
+          "description": "min=168.459, mean=168.459, max=168.459, sum=168.459 (1)",
+          "style": {},
+          "markdown": false,
+          "run_spec_names": [
+            "gsm:model=EleutherAI_pythia-2.8b"
+          ]
+        },
+        {
+          "value": 409.4,
+          "description": "min=95, mean=409.4, max=1000, sum=2047 (5)",
+          "style": {},
+          "markdown": false,
+          "run_spec_names": [
+            "legalbench:subset=abercrombie,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=corporate_lobbying,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=function_of_decision_section,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=international_citizenship_questions,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=proa,model=EleutherAI_pythia-2.8b"
+          ]
+        },
+        {
+          "value": 3.8604081632653062,
+          "description": "min=0.302, mean=3.86, max=5, sum=19.302 (5)",
+          "style": {},
+          "markdown": false,
+          "run_spec_names": [
+            "legalbench:subset=abercrombie,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=corporate_lobbying,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=function_of_decision_section,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=international_citizenship_questions,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=proa,model=EleutherAI_pythia-2.8b"
+          ]
+        },
+        {
+          "value": 0.002857142857142857,
+          "description": "min=0, mean=0.003, max=0.014, sum=0.014 (5)",
+          "style": {},
+          "markdown": false,
+          "run_spec_names": [
+            "legalbench:subset=abercrombie,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=corporate_lobbying,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=function_of_decision_section,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=international_citizenship_questions,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=proa,model=EleutherAI_pythia-2.8b"
+          ]
+        },
+        {
+          "value": 560.6440716343213,
+          "description": "min=206.779, mean=560.644, max=1497.455, sum=2803.22 (5)",
+          "style": {},
+          "markdown": false,
+          "run_spec_names": [
+            "legalbench:subset=abercrombie,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=corporate_lobbying,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=function_of_decision_section,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=international_citizenship_questions,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=proa,model=EleutherAI_pythia-2.8b"
+          ]
+        },
+        {
+          "value": 1.6417916921537006,
+          "description": "min=1, mean=1.642, max=4.124, sum=8.209 (5)",
+          "style": {},
+          "markdown": false,
+          "run_spec_names": [
+            "legalbench:subset=abercrombie,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=corporate_lobbying,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=function_of_decision_section,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=international_citizenship_questions,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=proa,model=EleutherAI_pythia-2.8b"
+          ]
+        },
+        {
+          "description": "No matching runs",
+          "markdown": false
+        },
+        {
+          "description": "No matching runs",
+          "markdown": false
+        },
+        {
+          "description": "No matching runs",
+          "markdown": false
+        },
+        {
+          "description": "No matching runs",
+          "markdown": false
+        },
+        {
+          "description": "No matching runs",
+          "markdown": false
+        },
+        {
+          "description": "No matching runs",
+          "markdown": false
+        },
+        {
+          "description": "No matching runs",
+          "markdown": false
+        },
+        {
+          "description": "No matching runs",
+          "markdown": false
+        },
+        {
+          "description": "No matching runs",
+          "markdown": false
+        },
+        {
+          "description": "No matching runs",
+          "markdown": false
+        }
+      ]
+    ],
+    "links": [
+      {
+        "text": "LaTeX",
+        "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/latex/core_scenarios_general_information.tex"
+      },
+      {
+        "text": "JSON",
+        "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/json/core_scenarios_general_information.json"
+      }
+    ],
+    "name": "general_information"
+  }
+]

lite_pythia-2.8b-step5000/groups/gsm.json

@@ -0,0 +1,147 @@
+[
+  {
+    "title": "",
+    "header": [
+      {
+        "value": "Model",
+        "markdown": false,
+        "metadata": {}
+      },
+      {
+        "value": "EM",
+        "description": "The grade school math word problems dataset (GSM8K) for testing mathematical reasoning on grade-school math problems [(Cobbe et al., 2021)](https://arxiv.org/pdf/2110.14168.pdf).\n\nExact match (final number): Fraction of instances that the predicted output matches a correct reference exactly, ignoring text preceding the specified indicator.",
+        "markdown": false,
+        "lower_is_better": false,
+        "metadata": {
+          "metric": "EM",
+          "run_group": "GSM8K"
+        }
+      },
+      {
+        "value": "Observed inference time (s)",
+        "description": "The grade school math word problems dataset (GSM8K) for testing mathematical reasoning on grade-school math problems [(Cobbe et al., 2021)](https://arxiv.org/pdf/2110.14168.pdf).\n\nObserved inference runtime (s): Average observed time to process a request to the model (via an API, and thus depends on particular deployment).",
+        "markdown": false,
+        "lower_is_better": true,
+        "metadata": {
+          "metric": "Observed inference time (s)",
+          "run_group": "GSM8K"
+        }
+      },
+      {
+        "value": "# eval",
+        "description": "The grade school math word problems dataset (GSM8K) for testing mathematical reasoning on grade-school math problems [(Cobbe et al., 2021)](https://arxiv.org/pdf/2110.14168.pdf).\n\n# eval: Number of evaluation instances.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# eval",
+          "run_group": "GSM8K"
+        }
+      },
+      {
+        "value": "# train",
+        "description": "The grade school math word problems dataset (GSM8K) for testing mathematical reasoning on grade-school math problems [(Cobbe et al., 2021)](https://arxiv.org/pdf/2110.14168.pdf).\n\n# train: Number of training instances (e.g., in-context examples).",
+        "markdown": false,
+        "metadata": {
+          "metric": "# train",
+          "run_group": "GSM8K"
+        }
+      },
+      {
+        "value": "truncated",
+        "description": "The grade school math word problems dataset (GSM8K) for testing mathematical reasoning on grade-school math problems [(Cobbe et al., 2021)](https://arxiv.org/pdf/2110.14168.pdf).\n\ntruncated: Fraction of instances where the prompt itself was truncated (implies that there were no in-context examples).",
+        "markdown": false,
+        "metadata": {
+          "metric": "truncated",
+          "run_group": "GSM8K"
+        }
+      },
+      {
+        "value": "# prompt tokens",
+        "description": "The grade school math word problems dataset (GSM8K) for testing mathematical reasoning on grade-school math problems [(Cobbe et al., 2021)](https://arxiv.org/pdf/2110.14168.pdf).\n\n# prompt tokens: Number of tokens in the prompt.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# prompt tokens",
+          "run_group": "GSM8K"
+        }
+      },
+      {
+        "value": "# output tokens",
+        "description": "The grade school math word problems dataset (GSM8K) for testing mathematical reasoning on grade-school math problems [(Cobbe et al., 2021)](https://arxiv.org/pdf/2110.14168.pdf).\n\n# output tokens: Actual number of output tokens.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# output tokens",
+          "run_group": "GSM8K"
+        }
+      }
+    ],
+    "rows": [
+      [
+        {
+          "value": "EleutherAI/pythia-2.8b",
+          "description": "",
+          "href": "?group=gsm&subgroup=&runSpecs=%5B%22gsm%3Amodel%3DEleutherAI_pythia-2.8b%22%5D",
+          "markdown": false,
+          "run_spec_names": [
+            "gsm:model=EleutherAI_pythia-2.8b"
+          ]
+        },
+        {
+          "value": 0.01,
+          "description": "min=0.01, mean=0.01, max=0.01, sum=0.01 (1)",
+          "style": {
+            "font-weight": "bold"
+          },
+          "markdown": false
+        },
+        {
+          "value": 2.5314217054843904,
+          "description": "min=2.531, mean=2.531, max=2.531, sum=2.531 (1)",
+          "style": {
+            "font-weight": "bold"
+          },
+          "markdown": false
+        },
+        {
+          "value": 1000.0,
+          "description": "min=1000, mean=1000, max=1000, sum=1000 (1)",
+          "style": {},
+          "markdown": false
+        },
+        {
+          "value": 5.0,
+          "description": "min=5, mean=5, max=5, sum=5 (1)",
+          "style": {},
+          "markdown": false
+        },
+        {
+          "value": 0.0,
+          "description": "min=0, mean=0, max=0, sum=0 (1)",
+          "style": {},
+          "markdown": false
+        },
+        {
+          "value": 939.582,
+          "description": "min=939.582, mean=939.582, max=939.582, sum=939.582 (1)",
+          "style": {},
+          "markdown": false
+        },
+        {
+          "value": 168.459,
+          "description": "min=168.459, mean=168.459, max=168.459, sum=168.459 (1)",
+          "style": {},
+          "markdown": false
+        }
+      ]
+    ],
+    "links": [
+      {
+        "text": "LaTeX",
+        "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/latex/gsm_gsm_.tex"
+      },
+      {
+        "text": "JSON",
+        "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/json/gsm_gsm_.json"
+      }
+    ],
+    "name": "gsm_"
+  }
+]

lite_pythia-2.8b-step5000/groups/json/core_scenarios_accuracy.json

@@ -0,0 +1,216 @@
+{
+  "title": "Accuracy",
+  "header": [
+    {
+      "value": "Model",
+      "markdown": false,
+      "metadata": {}
+    },
+    {
+      "value": "Mean win rate",
+      "description": "How many models this model outperforms on average (over columns).",
+      "markdown": false,
+      "lower_is_better": false,
+      "metadata": {}
+    },
+    {
+      "value": "NarrativeQA - F1",
+      "description": "The NarrativeQA benchmark for reading comprehension over narratives [(Ko\u010disk\u00fd et al., 2017)](https://aclanthology.org/Q18-1023/).\n\nF1: Average F1 score in terms of word overlap between the model output and correct reference.",
+      "markdown": false,
+      "lower_is_better": false,
+      "metadata": {
+        "metric": "F1",
+        "run_group": "NarrativeQA"
+      }
+    },
+    {
+      "value": "NaturalQuestions (open-book) - F1",
+      "description": "The NaturalQuestions [(Kwiatkowski et al., 2019)](https://aclanthology.org/Q19-1026/) benchmark for question answering based on naturally-occurring queries through Google Search. The input includes the Wikipedia page with the answer.\n\nF1: Average F1 score in terms of word overlap between the model output and correct reference.",
+      "markdown": false,
+      "lower_is_better": false,
+      "metadata": {
+        "metric": "F1",
+        "run_group": "NaturalQuestions (open-book)"
+      }
+    },
+    {
+      "value": "NaturalQuestions (closed-book) - F1",
+      "description": "The NaturalQuestions [(Kwiatkowski et al., 2019)](https://aclanthology.org/Q19-1026/) benchmark for question answering based on naturally-occurring queries through Google Search. The input does not include the Wikipedia page with the answer.\n\nF1: Average F1 score in terms of word overlap between the model output and correct reference.",
+      "markdown": false,
+      "lower_is_better": false,
+      "metadata": {
+        "metric": "F1",
+        "run_group": "NaturalQuestions (closed-book)"
+      }
+    },
+    {
+      "value": "OpenbookQA - EM",
+      "description": "The OpenbookQA benchmark for commonsense-intensive open book question answering [(Mihaylov et al., 2018)](https://aclanthology.org/D18-1260/).\n\nExact match: Fraction of instances that the predicted output matches a correct reference exactly.",
+      "markdown": false,
+      "lower_is_better": false,
+      "metadata": {
+        "metric": "EM",
+        "run_group": "OpenbookQA"
+      }
+    },
+    {
+      "value": "MMLU - EM",
+      "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\nExact match: Fraction of instances that the predicted output matches a correct reference exactly.",
+      "markdown": false,
+      "lower_is_better": false,
+      "metadata": {
+        "metric": "EM",
+        "run_group": "MMLU"
+      }
+    },
+    {
+      "value": "MATH - Equivalent (CoT)",
+      "description": "The MATH benchmark for measuring mathematical problem solving on competition math problems with chain-of-thought style reasoning [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2103.03874.pdf).\n\nEquivalent (CoT): Fraction of model outputs that are mathematically equivalent to the correct reference when using chain-of-thought prompting.",
+      "markdown": false,
+      "lower_is_better": false,
+      "metadata": {
+        "metric": "Equivalent (CoT)",
+        "run_group": "MATH"
+      }
+    },
+    {
+      "value": "GSM8K - EM",
+      "description": "The grade school math word problems dataset (GSM8K) for testing mathematical reasoning on grade-school math problems [(Cobbe et al., 2021)](https://arxiv.org/pdf/2110.14168.pdf).\n\nExact match (final number): Fraction of instances that the predicted output matches a correct reference exactly, ignoring text preceding the specified indicator.",
+      "markdown": false,
+      "lower_is_better": false,
+      "metadata": {
+        "metric": "EM",
+        "run_group": "GSM8K"
+      }
+    },
+    {
+      "value": "LegalBench - EM",
+      "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\nQuasi-exact match: Fraction of instances that the predicted output matches a correct reference up to light processing.",
+      "markdown": false,
+      "lower_is_better": false,
+      "metadata": {
+        "metric": "EM",
+        "run_group": "LegalBench"
+      }
+    },
+    {
+      "value": "MedQA - EM",
+      "description": "MedQA is an open domain question answering dataset composed of questions from professional medical board exams ([Jin et al. 2020](https://arxiv.org/pdf/2009.13081.pdf)).\n\nQuasi-exact match: Fraction of instances that the predicted output matches a correct reference up to light processing.",
+      "markdown": false,
+      "lower_is_better": false,
+      "metadata": {
+        "metric": "EM",
+        "run_group": "MedQA"
+      }
+    },
+    {
+      "value": "WMT 2014 - BLEU-4",
+      "description": "WMT 2014 is a collection of machine translation datasets [(website)](https://www.statmt.org/wmt14/index.html).\n\nBLEU-4: Average BLEU score [(Papineni et al., 2002)](https://aclanthology.org/P02-1040/) based on 4-gram overlap.",
+      "markdown": false,
+      "lower_is_better": false,
+      "metadata": {
+        "metric": "BLEU-4",
+        "run_group": "WMT 2014"
+      }
+    }
+  ],
+  "rows": [
+    [
+      {
+        "value": "EleutherAI/pythia-2.8b",
+        "description": "",
+        "markdown": false
+      },
+      {
+        "markdown": false
+      },
+      {
+        "description": "No matching runs",
+        "markdown": false
+      },
+      {
+        "description": "No matching runs",
+        "markdown": false
+      },
+      {
+        "description": "No matching runs",
+        "markdown": false
+      },
+      {
+        "value": 0.22,
+        "description": "min=0.22, mean=0.22, max=0.22, sum=0.22 (1)",
+        "style": {
+          "font-weight": "bold"
+        },
+        "markdown": false,
+        "run_spec_names": [
+          "commonsense:dataset=openbookqa,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b"
+        ]
+      },
+      {
+        "value": 0.27487719298245616,
+        "description": "min=0.25, mean=0.275, max=0.32, sum=1.374 (5)",
+        "style": {
+          "font-weight": "bold"
+        },
+        "markdown": false,
+        "run_spec_names": [
+          "mmlu:subject=abstract_algebra,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=college_chemistry,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=computer_security,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=econometrics,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=us_foreign_policy,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b"
+        ]
+      },
+      {
+        "description": "No matching runs",
+        "markdown": false
+      },
+      {
+        "value": 0.01,
+        "description": "min=0.01, mean=0.01, max=0.01, sum=0.01 (1)",
+        "style": {
+          "font-weight": "bold"
+        },
+        "markdown": false,
+        "run_spec_names": [
+          "gsm:model=EleutherAI_pythia-2.8b"
+        ]
+      },
+      {
+        "value": 0.3189986232611502,
+        "description": "min=0.144, mean=0.319, max=0.619, sum=1.595 (5)",
+        "style": {
+          "font-weight": "bold"
+        },
+        "markdown": false,
+        "run_spec_names": [
+          "legalbench:subset=abercrombie,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=corporate_lobbying,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=function_of_decision_section,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=international_citizenship_questions,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=proa,model=EleutherAI_pythia-2.8b"
+        ]
+      },
+      {
+        "description": "No matching runs",
+        "markdown": false
+      },
+      {
+        "description": "No matching runs",
+        "markdown": false
+      }
+    ]
+  ],
+  "links": [
+    {
+      "text": "LaTeX",
+      "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/latex/core_scenarios_accuracy.tex"
+    },
+    {
+      "text": "JSON",
+      "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/json/core_scenarios_accuracy.json"
+    }
+  ],
+  "name": "accuracy"
+}

lite_pythia-2.8b-step5000/groups/json/core_scenarios_efficiency.json

@@ -0,0 +1,216 @@
+{
+  "title": "Efficiency",
+  "header": [
+    {
+      "value": "Model",
+      "markdown": false,
+      "metadata": {}
+    },
+    {
+      "value": "Mean win rate",
+      "description": "How many models this model outperforms on average (over columns).",
+      "markdown": false,
+      "lower_is_better": false,
+      "metadata": {}
+    },
+    {
+      "value": "NarrativeQA - Observed inference time (s)",
+      "description": "The NarrativeQA benchmark for reading comprehension over narratives [(Ko\u010disk\u00fd et al., 2017)](https://aclanthology.org/Q18-1023/).\n\nObserved inference runtime (s): Average observed time to process a request to the model (via an API, and thus depends on particular deployment).",
+      "markdown": false,
+      "lower_is_better": true,
+      "metadata": {
+        "metric": "Observed inference time (s)",
+        "run_group": "NarrativeQA"
+      }
+    },
+    {
+      "value": "NaturalQuestions (open-book) - Observed inference time (s)",
+      "description": "The NaturalQuestions [(Kwiatkowski et al., 2019)](https://aclanthology.org/Q19-1026/) benchmark for question answering based on naturally-occurring queries through Google Search. The input includes the Wikipedia page with the answer.\n\nObserved inference runtime (s): Average observed time to process a request to the model (via an API, and thus depends on particular deployment).",
+      "markdown": false,
+      "lower_is_better": true,
+      "metadata": {
+        "metric": "Observed inference time (s)",
+        "run_group": "NaturalQuestions (open-book)"
+      }
+    },
+    {
+      "value": "NaturalQuestions (closed-book) - Observed inference time (s)",
+      "description": "The NaturalQuestions [(Kwiatkowski et al., 2019)](https://aclanthology.org/Q19-1026/) benchmark for question answering based on naturally-occurring queries through Google Search. The input does not include the Wikipedia page with the answer.\n\nObserved inference runtime (s): Average observed time to process a request to the model (via an API, and thus depends on particular deployment).",
+      "markdown": false,
+      "lower_is_better": true,
+      "metadata": {
+        "metric": "Observed inference time (s)",
+        "run_group": "NaturalQuestions (closed-book)"
+      }
+    },
+    {
+      "value": "OpenbookQA - Observed inference time (s)",
+      "description": "The OpenbookQA benchmark for commonsense-intensive open book question answering [(Mihaylov et al., 2018)](https://aclanthology.org/D18-1260/).\n\nObserved inference runtime (s): Average observed time to process a request to the model (via an API, and thus depends on particular deployment).",
+      "markdown": false,
+      "lower_is_better": true,
+      "metadata": {
+        "metric": "Observed inference time (s)",
+        "run_group": "OpenbookQA"
+      }
+    },
+    {
+      "value": "MMLU - Observed inference time (s)",
+      "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\nObserved inference runtime (s): Average observed time to process a request to the model (via an API, and thus depends on particular deployment).",
+      "markdown": false,
+      "lower_is_better": true,
+      "metadata": {
+        "metric": "Observed inference time (s)",
+        "run_group": "MMLU"
+      }
+    },
+    {
+      "value": "MATH - Observed inference time (s)",
+      "description": "The MATH benchmark for measuring mathematical problem solving on competition math problems with chain-of-thought style reasoning [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2103.03874.pdf).\n\nObserved inference runtime (s): Average observed time to process a request to the model (via an API, and thus depends on particular deployment).",
+      "markdown": false,
+      "lower_is_better": true,
+      "metadata": {
+        "metric": "Observed inference time (s)",
+        "run_group": "MATH"
+      }
+    },
+    {
+      "value": "GSM8K - Observed inference time (s)",
+      "description": "The grade school math word problems dataset (GSM8K) for testing mathematical reasoning on grade-school math problems [(Cobbe et al., 2021)](https://arxiv.org/pdf/2110.14168.pdf).\n\nObserved inference runtime (s): Average observed time to process a request to the model (via an API, and thus depends on particular deployment).",
+      "markdown": false,
+      "lower_is_better": true,
+      "metadata": {
+        "metric": "Observed inference time (s)",
+        "run_group": "GSM8K"
+      }
+    },
+    {
+      "value": "LegalBench - Observed inference time (s)",
+      "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\nObserved inference runtime (s): Average observed time to process a request to the model (via an API, and thus depends on particular deployment).",
+      "markdown": false,
+      "lower_is_better": true,
+      "metadata": {
+        "metric": "Observed inference time (s)",
+        "run_group": "LegalBench"
+      }
+    },
+    {
+      "value": "MedQA - Observed inference time (s)",
+      "description": "MedQA is an open domain question answering dataset composed of questions from professional medical board exams ([Jin et al. 2020](https://arxiv.org/pdf/2009.13081.pdf)).\n\nObserved inference runtime (s): Average observed time to process a request to the model (via an API, and thus depends on particular deployment).",
+      "markdown": false,
+      "lower_is_better": true,
+      "metadata": {
+        "metric": "Observed inference time (s)",
+        "run_group": "MedQA"
+      }
+    },
+    {
+      "value": "WMT 2014 - Observed inference time (s)",
+      "description": "WMT 2014 is a collection of machine translation datasets [(website)](https://www.statmt.org/wmt14/index.html).\n\nObserved inference runtime (s): Average observed time to process a request to the model (via an API, and thus depends on particular deployment).",
+      "markdown": false,
+      "lower_is_better": true,
+      "metadata": {
+        "metric": "Observed inference time (s)",
+        "run_group": "WMT 2014"
+      }
+    }
+  ],
+  "rows": [
+    [
+      {
+        "value": "EleutherAI/pythia-2.8b",
+        "description": "",
+        "markdown": false
+      },
+      {
+        "markdown": false
+      },
+      {
+        "description": "No matching runs",
+        "markdown": false
+      },
+      {
+        "description": "No matching runs",
+        "markdown": false
+      },
+      {
+        "description": "No matching runs",
+        "markdown": false
+      },
+      {
+        "value": 0.12883714818954467,
+        "description": "min=0.129, mean=0.129, max=0.129, sum=0.129 (1)",
+        "style": {
+          "font-weight": "bold"
+        },
+        "markdown": false,
+        "run_spec_names": [
+          "commonsense:dataset=openbookqa,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b"
+        ]
+      },
+      {
+        "value": 0.22061018314696193,
+        "description": "min=0.173, mean=0.221, max=0.282, sum=1.103 (5)",
+        "style": {
+          "font-weight": "bold"
+        },
+        "markdown": false,
+        "run_spec_names": [
+          "mmlu:subject=abstract_algebra,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=college_chemistry,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=computer_security,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=econometrics,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=us_foreign_policy,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b"
+        ]
+      },
+      {
+        "description": "No matching runs",
+        "markdown": false
+      },
+      {
+        "value": 2.5314217054843904,
+        "description": "min=2.531, mean=2.531, max=2.531, sum=2.531 (1)",
+        "style": {
+          "font-weight": "bold"
+        },
+        "markdown": false,
+        "run_spec_names": [
+          "gsm:model=EleutherAI_pythia-2.8b"
+        ]
+      },
+      {
+        "value": 0.3094366045219278,
+        "description": "min=0.141, mean=0.309, max=0.795, sum=1.547 (5)",
+        "style": {
+          "font-weight": "bold"
+        },
+        "markdown": false,
+        "run_spec_names": [
+          "legalbench:subset=abercrombie,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=corporate_lobbying,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=function_of_decision_section,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=international_citizenship_questions,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=proa,model=EleutherAI_pythia-2.8b"
+        ]
+      },
+      {
+        "description": "No matching runs",
+        "markdown": false
+      },
+      {
+        "description": "No matching runs",
+        "markdown": false
+      }
+    ]
+  ],
+  "links": [
+    {
+      "text": "LaTeX",
+      "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/latex/core_scenarios_efficiency.tex"
+    },
+    {
+      "text": "JSON",
+      "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/json/core_scenarios_efficiency.json"
+    }
+  ],
+  "name": "efficiency"
+}

lite_pythia-2.8b-step5000/groups/json/core_scenarios_general_information.json

@@ -0,0 +1,830 @@
+{
+  "title": "General information",
+  "header": [
+    {
+      "value": "Model",
+      "markdown": false,
+      "metadata": {}
+    },
+    {
+      "value": "Mean win rate",
+      "description": "How many models this model outperforms on average (over columns).",
+      "markdown": false,
+      "lower_is_better": false,
+      "metadata": {}
+    },
+    {
+      "value": "NarrativeQA - # eval",
+      "description": "The NarrativeQA benchmark for reading comprehension over narratives [(Ko\u010disk\u00fd et al., 2017)](https://aclanthology.org/Q18-1023/).\n\n# eval: Number of evaluation instances.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# eval",
+        "run_group": "NarrativeQA"
+      }
+    },
+    {
+      "value": "NarrativeQA - # train",
+      "description": "The NarrativeQA benchmark for reading comprehension over narratives [(Ko\u010disk\u00fd et al., 2017)](https://aclanthology.org/Q18-1023/).\n\n# train: Number of training instances (e.g., in-context examples).",
+      "markdown": false,
+      "metadata": {
+        "metric": "# train",
+        "run_group": "NarrativeQA"
+      }
+    },
+    {
+      "value": "NarrativeQA - truncated",
+      "description": "The NarrativeQA benchmark for reading comprehension over narratives [(Ko\u010disk\u00fd et al., 2017)](https://aclanthology.org/Q18-1023/).\n\ntruncated: Fraction of instances where the prompt itself was truncated (implies that there were no in-context examples).",
+      "markdown": false,
+      "metadata": {
+        "metric": "truncated",
+        "run_group": "NarrativeQA"
+      }
+    },
+    {
+      "value": "NarrativeQA - # prompt tokens",
+      "description": "The NarrativeQA benchmark for reading comprehension over narratives [(Ko\u010disk\u00fd et al., 2017)](https://aclanthology.org/Q18-1023/).\n\n# prompt tokens: Number of tokens in the prompt.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# prompt tokens",
+        "run_group": "NarrativeQA"
+      }
+    },
+    {
+      "value": "NarrativeQA - # output tokens",
+      "description": "The NarrativeQA benchmark for reading comprehension over narratives [(Ko\u010disk\u00fd et al., 2017)](https://aclanthology.org/Q18-1023/).\n\n# output tokens: Actual number of output tokens.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# output tokens",
+        "run_group": "NarrativeQA"
+      }
+    },
+    {
+      "value": "NaturalQuestions (open-book) - # eval",
+      "description": "The NaturalQuestions [(Kwiatkowski et al., 2019)](https://aclanthology.org/Q19-1026/) benchmark for question answering based on naturally-occurring queries through Google Search. The input includes the Wikipedia page with the answer.\n\n# eval: Number of evaluation instances.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# eval",
+        "run_group": "NaturalQuestions (open-book)"
+      }
+    },
+    {
+      "value": "NaturalQuestions (open-book) - # train",
+      "description": "The NaturalQuestions [(Kwiatkowski et al., 2019)](https://aclanthology.org/Q19-1026/) benchmark for question answering based on naturally-occurring queries through Google Search. The input includes the Wikipedia page with the answer.\n\n# train: Number of training instances (e.g., in-context examples).",
+      "markdown": false,
+      "metadata": {
+        "metric": "# train",
+        "run_group": "NaturalQuestions (open-book)"
+      }
+    },
+    {
+      "value": "NaturalQuestions (open-book) - truncated",
+      "description": "The NaturalQuestions [(Kwiatkowski et al., 2019)](https://aclanthology.org/Q19-1026/) benchmark for question answering based on naturally-occurring queries through Google Search. The input includes the Wikipedia page with the answer.\n\ntruncated: Fraction of instances where the prompt itself was truncated (implies that there were no in-context examples).",
+      "markdown": false,
+      "metadata": {
+        "metric": "truncated",
+        "run_group": "NaturalQuestions (open-book)"
+      }
+    },
+    {
+      "value": "NaturalQuestions (open-book) - # prompt tokens",
+      "description": "The NaturalQuestions [(Kwiatkowski et al., 2019)](https://aclanthology.org/Q19-1026/) benchmark for question answering based on naturally-occurring queries through Google Search. The input includes the Wikipedia page with the answer.\n\n# prompt tokens: Number of tokens in the prompt.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# prompt tokens",
+        "run_group": "NaturalQuestions (open-book)"
+      }
+    },
+    {
+      "value": "NaturalQuestions (open-book) - # output tokens",
+      "description": "The NaturalQuestions [(Kwiatkowski et al., 2019)](https://aclanthology.org/Q19-1026/) benchmark for question answering based on naturally-occurring queries through Google Search. The input includes the Wikipedia page with the answer.\n\n# output tokens: Actual number of output tokens.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# output tokens",
+        "run_group": "NaturalQuestions (open-book)"
+      }
+    },
+    {
+      "value": "NaturalQuestions (closed-book) - # eval",
+      "description": "The NaturalQuestions [(Kwiatkowski et al., 2019)](https://aclanthology.org/Q19-1026/) benchmark for question answering based on naturally-occurring queries through Google Search. The input does not include the Wikipedia page with the answer.\n\n# eval: Number of evaluation instances.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# eval",
+        "run_group": "NaturalQuestions (closed-book)"
+      }
+    },
+    {
+      "value": "NaturalQuestions (closed-book) - # train",
+      "description": "The NaturalQuestions [(Kwiatkowski et al., 2019)](https://aclanthology.org/Q19-1026/) benchmark for question answering based on naturally-occurring queries through Google Search. The input does not include the Wikipedia page with the answer.\n\n# train: Number of training instances (e.g., in-context examples).",
+      "markdown": false,
+      "metadata": {
+        "metric": "# train",
+        "run_group": "NaturalQuestions (closed-book)"
+      }
+    },
+    {
+      "value": "NaturalQuestions (closed-book) - truncated",
+      "description": "The NaturalQuestions [(Kwiatkowski et al., 2019)](https://aclanthology.org/Q19-1026/) benchmark for question answering based on naturally-occurring queries through Google Search. The input does not include the Wikipedia page with the answer.\n\ntruncated: Fraction of instances where the prompt itself was truncated (implies that there were no in-context examples).",
+      "markdown": false,
+      "metadata": {
+        "metric": "truncated",
+        "run_group": "NaturalQuestions (closed-book)"
+      }
+    },
+    {
+      "value": "NaturalQuestions (closed-book) - # prompt tokens",
+      "description": "The NaturalQuestions [(Kwiatkowski et al., 2019)](https://aclanthology.org/Q19-1026/) benchmark for question answering based on naturally-occurring queries through Google Search. The input does not include the Wikipedia page with the answer.\n\n# prompt tokens: Number of tokens in the prompt.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# prompt tokens",
+        "run_group": "NaturalQuestions (closed-book)"
+      }
+    },
+    {
+      "value": "NaturalQuestions (closed-book) - # output tokens",
+      "description": "The NaturalQuestions [(Kwiatkowski et al., 2019)](https://aclanthology.org/Q19-1026/) benchmark for question answering based on naturally-occurring queries through Google Search. The input does not include the Wikipedia page with the answer.\n\n# output tokens: Actual number of output tokens.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# output tokens",
+        "run_group": "NaturalQuestions (closed-book)"
+      }
+    },
+    {
+      "value": "OpenbookQA - # eval",
+      "description": "The OpenbookQA benchmark for commonsense-intensive open book question answering [(Mihaylov et al., 2018)](https://aclanthology.org/D18-1260/).\n\n# eval: Number of evaluation instances.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# eval",
+        "run_group": "OpenbookQA"
+      }
+    },
+    {
+      "value": "OpenbookQA - # train",
+      "description": "The OpenbookQA benchmark for commonsense-intensive open book question answering [(Mihaylov et al., 2018)](https://aclanthology.org/D18-1260/).\n\n# train: Number of training instances (e.g., in-context examples).",
+      "markdown": false,
+      "metadata": {
+        "metric": "# train",
+        "run_group": "OpenbookQA"
+      }
+    },
+    {
+      "value": "OpenbookQA - truncated",
+      "description": "The OpenbookQA benchmark for commonsense-intensive open book question answering [(Mihaylov et al., 2018)](https://aclanthology.org/D18-1260/).\n\ntruncated: Fraction of instances where the prompt itself was truncated (implies that there were no in-context examples).",
+      "markdown": false,
+      "metadata": {
+        "metric": "truncated",
+        "run_group": "OpenbookQA"
+      }
+    },
+    {
+      "value": "OpenbookQA - # prompt tokens",
+      "description": "The OpenbookQA benchmark for commonsense-intensive open book question answering [(Mihaylov et al., 2018)](https://aclanthology.org/D18-1260/).\n\n# prompt tokens: Number of tokens in the prompt.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# prompt tokens",
+        "run_group": "OpenbookQA"
+      }
+    },
+    {
+      "value": "OpenbookQA - # output tokens",
+      "description": "The OpenbookQA benchmark for commonsense-intensive open book question answering [(Mihaylov et al., 2018)](https://aclanthology.org/D18-1260/).\n\n# output tokens: Actual number of output tokens.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# output tokens",
+        "run_group": "OpenbookQA"
+      }
+    },
+    {
+      "value": "MMLU - # eval",
+      "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# eval: Number of evaluation instances.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# eval",
+        "run_group": "MMLU"
+      }
+    },
+    {
+      "value": "MMLU - # train",
+      "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# train: Number of training instances (e.g., in-context examples).",
+      "markdown": false,
+      "metadata": {
+        "metric": "# train",
+        "run_group": "MMLU"
+      }
+    },
+    {
+      "value": "MMLU - truncated",
+      "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\ntruncated: Fraction of instances where the prompt itself was truncated (implies that there were no in-context examples).",
+      "markdown": false,
+      "metadata": {
+        "metric": "truncated",
+        "run_group": "MMLU"
+      }
+    },
+    {
+      "value": "MMLU - # prompt tokens",
+      "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# prompt tokens: Number of tokens in the prompt.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# prompt tokens",
+        "run_group": "MMLU"
+      }
+    },
+    {
+      "value": "MMLU - # output tokens",
+      "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# output tokens: Actual number of output tokens.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# output tokens",
+        "run_group": "MMLU"
+      }
+    },
+    {
+      "value": "MATH - # eval",
+      "description": "The MATH benchmark for measuring mathematical problem solving on competition math problems with chain-of-thought style reasoning [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2103.03874.pdf).\n\n# eval: Number of evaluation instances.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# eval",
+        "run_group": "MATH"
+      }
+    },
+    {
+      "value": "MATH - # train",
+      "description": "The MATH benchmark for measuring mathematical problem solving on competition math problems with chain-of-thought style reasoning [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2103.03874.pdf).\n\n# train: Number of training instances (e.g., in-context examples).",
+      "markdown": false,
+      "metadata": {
+        "metric": "# train",
+        "run_group": "MATH"
+      }
+    },
+    {
+      "value": "MATH - truncated",
+      "description": "The MATH benchmark for measuring mathematical problem solving on competition math problems with chain-of-thought style reasoning [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2103.03874.pdf).\n\ntruncated: Fraction of instances where the prompt itself was truncated (implies that there were no in-context examples).",
+      "markdown": false,
+      "metadata": {
+        "metric": "truncated",
+        "run_group": "MATH"
+      }
+    },
+    {
+      "value": "MATH - # prompt tokens",
+      "description": "The MATH benchmark for measuring mathematical problem solving on competition math problems with chain-of-thought style reasoning [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2103.03874.pdf).\n\n# prompt tokens: Number of tokens in the prompt.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# prompt tokens",
+        "run_group": "MATH"
+      }
+    },
+    {
+      "value": "MATH - # output tokens",
+      "description": "The MATH benchmark for measuring mathematical problem solving on competition math problems with chain-of-thought style reasoning [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2103.03874.pdf).\n\n# output tokens: Actual number of output tokens.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# output tokens",
+        "run_group": "MATH"
+      }
+    },
+    {
+      "value": "GSM8K - # eval",
+      "description": "The grade school math word problems dataset (GSM8K) for testing mathematical reasoning on grade-school math problems [(Cobbe et al., 2021)](https://arxiv.org/pdf/2110.14168.pdf).\n\n# eval: Number of evaluation instances.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# eval",
+        "run_group": "GSM8K"
+      }
+    },
+    {
+      "value": "GSM8K - # train",
+      "description": "The grade school math word problems dataset (GSM8K) for testing mathematical reasoning on grade-school math problems [(Cobbe et al., 2021)](https://arxiv.org/pdf/2110.14168.pdf).\n\n# train: Number of training instances (e.g., in-context examples).",
+      "markdown": false,
+      "metadata": {
+        "metric": "# train",
+        "run_group": "GSM8K"
+      }
+    },
+    {
+      "value": "GSM8K - truncated",
+      "description": "The grade school math word problems dataset (GSM8K) for testing mathematical reasoning on grade-school math problems [(Cobbe et al., 2021)](https://arxiv.org/pdf/2110.14168.pdf).\n\ntruncated: Fraction of instances where the prompt itself was truncated (implies that there were no in-context examples).",
+      "markdown": false,
+      "metadata": {
+        "metric": "truncated",
+        "run_group": "GSM8K"
+      }
+    },
+    {
+      "value": "GSM8K - # prompt tokens",
+      "description": "The grade school math word problems dataset (GSM8K) for testing mathematical reasoning on grade-school math problems [(Cobbe et al., 2021)](https://arxiv.org/pdf/2110.14168.pdf).\n\n# prompt tokens: Number of tokens in the prompt.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# prompt tokens",
+        "run_group": "GSM8K"
+      }
+    },
+    {
+      "value": "GSM8K - # output tokens",
+      "description": "The grade school math word problems dataset (GSM8K) for testing mathematical reasoning on grade-school math problems [(Cobbe et al., 2021)](https://arxiv.org/pdf/2110.14168.pdf).\n\n# output tokens: Actual number of output tokens.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# output tokens",
+        "run_group": "GSM8K"
+      }
+    },
+    {
+      "value": "LegalBench - # eval",
+      "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# eval: Number of evaluation instances.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# eval",
+        "run_group": "LegalBench"
+      }
+    },
+    {
+      "value": "LegalBench - # train",
+      "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# train: Number of training instances (e.g., in-context examples).",
+      "markdown": false,
+      "metadata": {
+        "metric": "# train",
+        "run_group": "LegalBench"
+      }
+    },
+    {
+      "value": "LegalBench - truncated",
+      "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\ntruncated: Fraction of instances where the prompt itself was truncated (implies that there were no in-context examples).",
+      "markdown": false,
+      "metadata": {
+        "metric": "truncated",
+        "run_group": "LegalBench"
+      }
+    },
+    {
+      "value": "LegalBench - # prompt tokens",
+      "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# prompt tokens: Number of tokens in the prompt.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# prompt tokens",
+        "run_group": "LegalBench"
+      }
+    },
+    {
+      "value": "LegalBench - # output tokens",
+      "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# output tokens: Actual number of output tokens.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# output tokens",
+        "run_group": "LegalBench"
+      }
+    },
+    {
+      "value": "MedQA - # eval",
+      "description": "MedQA is an open domain question answering dataset composed of questions from professional medical board exams ([Jin et al. 2020](https://arxiv.org/pdf/2009.13081.pdf)).\n\n# eval: Number of evaluation instances.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# eval",
+        "run_group": "MedQA"
+      }
+    },
+    {
+      "value": "MedQA - # train",
+      "description": "MedQA is an open domain question answering dataset composed of questions from professional medical board exams ([Jin et al. 2020](https://arxiv.org/pdf/2009.13081.pdf)).\n\n# train: Number of training instances (e.g., in-context examples).",
+      "markdown": false,
+      "metadata": {
+        "metric": "# train",
+        "run_group": "MedQA"
+      }
+    },
+    {
+      "value": "MedQA - truncated",
+      "description": "MedQA is an open domain question answering dataset composed of questions from professional medical board exams ([Jin et al. 2020](https://arxiv.org/pdf/2009.13081.pdf)).\n\ntruncated: Fraction of instances where the prompt itself was truncated (implies that there were no in-context examples).",
+      "markdown": false,
+      "metadata": {
+        "metric": "truncated",
+        "run_group": "MedQA"
+      }
+    },
+    {
+      "value": "MedQA - # prompt tokens",
+      "description": "MedQA is an open domain question answering dataset composed of questions from professional medical board exams ([Jin et al. 2020](https://arxiv.org/pdf/2009.13081.pdf)).\n\n# prompt tokens: Number of tokens in the prompt.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# prompt tokens",
+        "run_group": "MedQA"
+      }
+    },
+    {
+      "value": "MedQA - # output tokens",
+      "description": "MedQA is an open domain question answering dataset composed of questions from professional medical board exams ([Jin et al. 2020](https://arxiv.org/pdf/2009.13081.pdf)).\n\n# output tokens: Actual number of output tokens.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# output tokens",
+        "run_group": "MedQA"
+      }
+    },
+    {
+      "value": "WMT 2014 - # eval",
+      "description": "WMT 2014 is a collection of machine translation datasets [(website)](https://www.statmt.org/wmt14/index.html).\n\n# eval: Number of evaluation instances.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# eval",
+        "run_group": "WMT 2014"
+      }
+    },
+    {
+      "value": "WMT 2014 - # train",
+      "description": "WMT 2014 is a collection of machine translation datasets [(website)](https://www.statmt.org/wmt14/index.html).\n\n# train: Number of training instances (e.g., in-context examples).",
+      "markdown": false,
+      "metadata": {
+        "metric": "# train",
+        "run_group": "WMT 2014"
+      }
+    },
+    {
+      "value": "WMT 2014 - truncated",
+      "description": "WMT 2014 is a collection of machine translation datasets [(website)](https://www.statmt.org/wmt14/index.html).\n\ntruncated: Fraction of instances where the prompt itself was truncated (implies that there were no in-context examples).",
+      "markdown": false,
+      "metadata": {
+        "metric": "truncated",
+        "run_group": "WMT 2014"
+      }
+    },
+    {
+      "value": "WMT 2014 - # prompt tokens",
+      "description": "WMT 2014 is a collection of machine translation datasets [(website)](https://www.statmt.org/wmt14/index.html).\n\n# prompt tokens: Number of tokens in the prompt.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# prompt tokens",
+        "run_group": "WMT 2014"
+      }
+    },
+    {
+      "value": "WMT 2014 - # output tokens",
+      "description": "WMT 2014 is a collection of machine translation datasets [(website)](https://www.statmt.org/wmt14/index.html).\n\n# output tokens: Actual number of output tokens.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# output tokens",
+        "run_group": "WMT 2014"
+      }
+    }
+  ],
+  "rows": [
+    [
+      {
+        "value": "EleutherAI/pythia-2.8b",
+        "description": "",
+        "markdown": false
+      },
+      {
+        "markdown": false
+      },
+      {
+        "description": "No matching runs",
+        "markdown": false
+      },
+      {
+        "description": "No matching runs",
+        "markdown": false
+      },
+      {
+        "description": "No matching runs",
+        "markdown": false
+      },
+      {
+        "description": "No matching runs",
+        "markdown": false
+      },
+      {
+        "description": "No matching runs",
+        "markdown": false
+      },
+      {
+        "description": "No matching runs",
+        "markdown": false
+      },
+      {
+        "description": "No matching runs",
+        "markdown": false
+      },
+      {
+        "description": "No matching runs",
+        "markdown": false
+      },
+      {
+        "description": "No matching runs",
+        "markdown": false
+      },
+      {
+        "description": "No matching runs",
+        "markdown": false
+      },
+      {
+        "description": "No matching runs",
+        "markdown": false
+      },
+      {
+        "description": "No matching runs",
+        "markdown": false
+      },
+      {
+        "description": "No matching runs",
+        "markdown": false
+      },
+      {
+        "description": "No matching runs",
+        "markdown": false
+      },
+      {
+        "description": "No matching runs",
+        "markdown": false
+      },
+      {
+        "value": 500.0,
+        "description": "min=500, mean=500, max=500, sum=500 (1)",
+        "style": {},
+        "markdown": false,
+        "run_spec_names": [
+          "commonsense:dataset=openbookqa,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b"
+        ]
+      },
+      {
+        "value": 5.0,
+        "description": "min=5, mean=5, max=5, sum=5 (1)",
+        "style": {},
+        "markdown": false,
+        "run_spec_names": [
+          "commonsense:dataset=openbookqa,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b"
+        ]
+      },
+      {
+        "value": 0.0,
+        "description": "min=0, mean=0, max=0, sum=0 (1)",
+        "style": {},
+        "markdown": false,
+        "run_spec_names": [
+          "commonsense:dataset=openbookqa,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b"
+        ]
+      },
+      {
+        "value": 251.556,
+        "description": "min=251.556, mean=251.556, max=251.556, sum=251.556 (1)",
+        "style": {},
+        "markdown": false,
+        "run_spec_names": [
+          "commonsense:dataset=openbookqa,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b"
+        ]
+      },
+      {
+        "value": 1.0,
+        "description": "min=1, mean=1, max=1, sum=1 (1)",
+        "style": {},
+        "markdown": false,
+        "run_spec_names": [
+          "commonsense:dataset=openbookqa,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b"
+        ]
+      },
+      {
+        "value": 102.8,
+        "description": "min=100, mean=102.8, max=114, sum=514 (5)",
+        "style": {},
+        "markdown": false,
+        "run_spec_names": [
+          "mmlu:subject=abstract_algebra,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=college_chemistry,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=computer_security,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=econometrics,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=us_foreign_policy,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b"
+        ]
+      },
+      {
+        "value": 5.0,
+        "description": "min=5, mean=5, max=5, sum=25 (5)",
+        "style": {},
+        "markdown": false,
+        "run_spec_names": [
+          "mmlu:subject=abstract_algebra,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=college_chemistry,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=computer_security,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=econometrics,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=us_foreign_policy,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b"
+        ]
+      },
+      {
+        "value": 0.0,
+        "description": "min=0, mean=0, max=0, sum=0 (5)",
+        "style": {},
+        "markdown": false,
+        "run_spec_names": [
+          "mmlu:subject=abstract_algebra,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=college_chemistry,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=computer_security,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=econometrics,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=us_foreign_policy,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b"
+        ]
+      },
+      {
+        "value": 467.935649122807,
+        "description": "min=358.76, mean=467.936, max=612.798, sum=2339.678 (5)",
+        "style": {},
+        "markdown": false,
+        "run_spec_names": [
+          "mmlu:subject=abstract_algebra,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=college_chemistry,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=computer_security,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=econometrics,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=us_foreign_policy,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b"
+        ]
+      },
+      {
+        "value": 1.0,
+        "description": "min=1, mean=1, max=1, sum=5 (5)",
+        "style": {},
+        "markdown": false,
+        "run_spec_names": [
+          "mmlu:subject=abstract_algebra,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=college_chemistry,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=computer_security,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=econometrics,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=us_foreign_policy,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b"
+        ]
+      },
+      {
+        "description": "No matching runs",
+        "markdown": false
+      },
+      {
+        "description": "No matching runs",
+        "markdown": false
+      },
+      {
+        "description": "No matching runs",
+        "markdown": false
+      },
+      {
+        "description": "No matching runs",
+        "markdown": false
+      },
+      {
+        "description": "No matching runs",
+        "markdown": false
+      },
+      {
+        "value": 1000.0,
+        "description": "min=1000, mean=1000, max=1000, sum=1000 (1)",
+        "style": {},
+        "markdown": false,
+        "run_spec_names": [
+          "gsm:model=EleutherAI_pythia-2.8b"
+        ]
+      },
+      {
+        "value": 5.0,
+        "description": "min=5, mean=5, max=5, sum=5 (1)",
+        "style": {},
+        "markdown": false,
+        "run_spec_names": [
+          "gsm:model=EleutherAI_pythia-2.8b"
+        ]
+      },
+      {
+        "value": 0.0,
+        "description": "min=0, mean=0, max=0, sum=0 (1)",
+        "style": {},
+        "markdown": false,
+        "run_spec_names": [
+          "gsm:model=EleutherAI_pythia-2.8b"
+        ]
+      },
+      {
+        "value": 939.582,
+        "description": "min=939.582, mean=939.582, max=939.582, sum=939.582 (1)",
+        "style": {},
+        "markdown": false,
+        "run_spec_names": [
+          "gsm:model=EleutherAI_pythia-2.8b"
+        ]
+      },
+      {
+        "value": 168.459,
+        "description": "min=168.459, mean=168.459, max=168.459, sum=168.459 (1)",
+        "style": {},
+        "markdown": false,
+        "run_spec_names": [
+          "gsm:model=EleutherAI_pythia-2.8b"
+        ]
+      },
+      {
+        "value": 409.4,
+        "description": "min=95, mean=409.4, max=1000, sum=2047 (5)",
+        "style": {},
+        "markdown": false,
+        "run_spec_names": [
+          "legalbench:subset=abercrombie,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=corporate_lobbying,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=function_of_decision_section,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=international_citizenship_questions,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=proa,model=EleutherAI_pythia-2.8b"
+        ]
+      },
+      {
+        "value": 3.8604081632653062,
+        "description": "min=0.302, mean=3.86, max=5, sum=19.302 (5)",
+        "style": {},
+        "markdown": false,
+        "run_spec_names": [
+          "legalbench:subset=abercrombie,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=corporate_lobbying,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=function_of_decision_section,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=international_citizenship_questions,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=proa,model=EleutherAI_pythia-2.8b"
+        ]
+      },
+      {
+        "value": 0.002857142857142857,
+        "description": "min=0, mean=0.003, max=0.014, sum=0.014 (5)",
+        "style": {},
+        "markdown": false,
+        "run_spec_names": [
+          "legalbench:subset=abercrombie,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=corporate_lobbying,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=function_of_decision_section,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=international_citizenship_questions,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=proa,model=EleutherAI_pythia-2.8b"
+        ]
+      },
+      {
+        "value": 560.6440716343213,
+        "description": "min=206.779, mean=560.644, max=1497.455, sum=2803.22 (5)",
+        "style": {},
+        "markdown": false,
+        "run_spec_names": [
+          "legalbench:subset=abercrombie,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=corporate_lobbying,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=function_of_decision_section,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=international_citizenship_questions,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=proa,model=EleutherAI_pythia-2.8b"
+        ]
+      },
+      {
+        "value": 1.6417916921537006,
+        "description": "min=1, mean=1.642, max=4.124, sum=8.209 (5)",
+        "style": {},
+        "markdown": false,
+        "run_spec_names": [
+          "legalbench:subset=abercrombie,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=corporate_lobbying,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=function_of_decision_section,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=international_citizenship_questions,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=proa,model=EleutherAI_pythia-2.8b"
+        ]
+      },
+      {
+        "description": "No matching runs",
+        "markdown": false
+      },
+      {
+        "description": "No matching runs",
+        "markdown": false
+      },
+      {
+        "description": "No matching runs",
+        "markdown": false
+      },
+      {
+        "description": "No matching runs",
+        "markdown": false
+      },
+      {
+        "description": "No matching runs",
+        "markdown": false
+      },
+      {
+        "description": "No matching runs",
+        "markdown": false
+      },
+      {
+        "description": "No matching runs",
+        "markdown": false
+      },
+      {
+        "description": "No matching runs",
+        "markdown": false
+      },
+      {
+        "description": "No matching runs",
+        "markdown": false
+      },
+      {
+        "description": "No matching runs",
+        "markdown": false
+      }
+    ]
+  ],
+  "links": [
+    {
+      "text": "LaTeX",
+      "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/latex/core_scenarios_general_information.tex"
+    },
+    {
+      "text": "JSON",
+      "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/json/core_scenarios_general_information.json"
+    }
+  ],
+  "name": "general_information"
+}

lite_pythia-2.8b-step5000/groups/json/gsm_gsm_.json

@@ -0,0 +1,145 @@
+{
+  "title": "",
+  "header": [
+    {
+      "value": "Model",
+      "markdown": false,
+      "metadata": {}
+    },
+    {
+      "value": "EM",
+      "description": "The grade school math word problems dataset (GSM8K) for testing mathematical reasoning on grade-school math problems [(Cobbe et al., 2021)](https://arxiv.org/pdf/2110.14168.pdf).\n\nExact match (final number): Fraction of instances that the predicted output matches a correct reference exactly, ignoring text preceding the specified indicator.",
+      "markdown": false,
+      "lower_is_better": false,
+      "metadata": {
+        "metric": "EM",
+        "run_group": "GSM8K"
+      }
+    },
+    {
+      "value": "Observed inference time (s)",
+      "description": "The grade school math word problems dataset (GSM8K) for testing mathematical reasoning on grade-school math problems [(Cobbe et al., 2021)](https://arxiv.org/pdf/2110.14168.pdf).\n\nObserved inference runtime (s): Average observed time to process a request to the model (via an API, and thus depends on particular deployment).",
+      "markdown": false,
+      "lower_is_better": true,
+      "metadata": {
+        "metric": "Observed inference time (s)",
+        "run_group": "GSM8K"
+      }
+    },
+    {
+      "value": "# eval",
+      "description": "The grade school math word problems dataset (GSM8K) for testing mathematical reasoning on grade-school math problems [(Cobbe et al., 2021)](https://arxiv.org/pdf/2110.14168.pdf).\n\n# eval: Number of evaluation instances.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# eval",
+        "run_group": "GSM8K"
+      }
+    },
+    {
+      "value": "# train",
+      "description": "The grade school math word problems dataset (GSM8K) for testing mathematical reasoning on grade-school math problems [(Cobbe et al., 2021)](https://arxiv.org/pdf/2110.14168.pdf).\n\n# train: Number of training instances (e.g., in-context examples).",
+      "markdown": false,
+      "metadata": {
+        "metric": "# train",
+        "run_group": "GSM8K"
+      }
+    },
+    {
+      "value": "truncated",
+      "description": "The grade school math word problems dataset (GSM8K) for testing mathematical reasoning on grade-school math problems [(Cobbe et al., 2021)](https://arxiv.org/pdf/2110.14168.pdf).\n\ntruncated: Fraction of instances where the prompt itself was truncated (implies that there were no in-context examples).",
+      "markdown": false,
+      "metadata": {
+        "metric": "truncated",
+        "run_group": "GSM8K"
+      }
+    },
+    {
+      "value": "# prompt tokens",
+      "description": "The grade school math word problems dataset (GSM8K) for testing mathematical reasoning on grade-school math problems [(Cobbe et al., 2021)](https://arxiv.org/pdf/2110.14168.pdf).\n\n# prompt tokens: Number of tokens in the prompt.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# prompt tokens",
+        "run_group": "GSM8K"
+      }
+    },
+    {
+      "value": "# output tokens",
+      "description": "The grade school math word problems dataset (GSM8K) for testing mathematical reasoning on grade-school math problems [(Cobbe et al., 2021)](https://arxiv.org/pdf/2110.14168.pdf).\n\n# output tokens: Actual number of output tokens.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# output tokens",
+        "run_group": "GSM8K"
+      }
+    }
+  ],
+  "rows": [
+    [
+      {
+        "value": "EleutherAI/pythia-2.8b",
+        "description": "",
+        "href": "?group=gsm&subgroup=&runSpecs=%5B%22gsm%3Amodel%3DEleutherAI_pythia-2.8b%22%5D",
+        "markdown": false,
+        "run_spec_names": [
+          "gsm:model=EleutherAI_pythia-2.8b"
+        ]
+      },
+      {
+        "value": 0.01,
+        "description": "min=0.01, mean=0.01, max=0.01, sum=0.01 (1)",
+        "style": {
+          "font-weight": "bold"
+        },
+        "markdown": false
+      },
+      {
+        "value": 2.5314217054843904,
+        "description": "min=2.531, mean=2.531, max=2.531, sum=2.531 (1)",
+        "style": {
+          "font-weight": "bold"
+        },
+        "markdown": false
+      },
+      {
+        "value": 1000.0,
+        "description": "min=1000, mean=1000, max=1000, sum=1000 (1)",
+        "style": {},
+        "markdown": false
+      },
+      {
+        "value": 5.0,
+        "description": "min=5, mean=5, max=5, sum=5 (1)",
+        "style": {},
+        "markdown": false
+      },
+      {
+        "value": 0.0,
+        "description": "min=0, mean=0, max=0, sum=0 (1)",
+        "style": {},
+        "markdown": false
+      },
+      {
+        "value": 939.582,
+        "description": "min=939.582, mean=939.582, max=939.582, sum=939.582 (1)",
+        "style": {},
+        "markdown": false
+      },
+      {
+        "value": 168.459,
+        "description": "min=168.459, mean=168.459, max=168.459, sum=168.459 (1)",
+        "style": {},
+        "markdown": false
+      }
+    ]
+  ],
+  "links": [
+    {
+      "text": "LaTeX",
+      "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/latex/gsm_gsm_.tex"
+    },
+    {
+      "text": "JSON",
+      "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/json/gsm_gsm_.json"
+    }
+  ],
+  "name": "gsm_"
+}

lite_pythia-2.8b-step5000/groups/json/legalbench_legalbench.json

@@ -0,0 +1,190 @@
+{
+  "title": "LegalBench",
+  "header": [
+    {
+      "value": "Model",
+      "markdown": false,
+      "metadata": {}
+    },
+    {
+      "value": "EM",
+      "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\nQuasi-exact match: Fraction of instances that the predicted output matches a correct reference up to light processing.",
+      "markdown": false,
+      "lower_is_better": false,
+      "metadata": {
+        "metric": "EM",
+        "run_group": "LegalBench"
+      }
+    },
+    {
+      "value": "Observed inference time (s)",
+      "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\nObserved inference runtime (s): Average observed time to process a request to the model (via an API, and thus depends on particular deployment).",
+      "markdown": false,
+      "lower_is_better": true,
+      "metadata": {
+        "metric": "Observed inference time (s)",
+        "run_group": "LegalBench"
+      }
+    },
+    {
+      "value": "# eval",
+      "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# eval: Number of evaluation instances.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# eval",
+        "run_group": "LegalBench"
+      }
+    },
+    {
+      "value": "# train",
+      "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# train: Number of training instances (e.g., in-context examples).",
+      "markdown": false,
+      "metadata": {
+        "metric": "# train",
+        "run_group": "LegalBench"
+      }
+    },
+    {
+      "value": "truncated",
+      "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\ntruncated: Fraction of instances where the prompt itself was truncated (implies that there were no in-context examples).",
+      "markdown": false,
+      "metadata": {
+        "metric": "truncated",
+        "run_group": "LegalBench"
+      }
+    },
+    {
+      "value": "# prompt tokens",
+      "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# prompt tokens: Number of tokens in the prompt.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# prompt tokens",
+        "run_group": "LegalBench"
+      }
+    },
+    {
+      "value": "# output tokens",
+      "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# output tokens: Actual number of output tokens.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# output tokens",
+        "run_group": "LegalBench"
+      }
+    }
+  ],
+  "rows": [
+    [
+      {
+        "value": "EleutherAI/pythia-2.8b",
+        "description": "",
+        "markdown": false
+      },
+      {
+        "value": 0.3189986232611502,
+        "description": "min=0.144, mean=0.319, max=0.619, sum=1.595 (5)",
+        "style": {
+          "font-weight": "bold"
+        },
+        "markdown": false,
+        "run_spec_names": [
+          "legalbench:subset=abercrombie,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=corporate_lobbying,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=function_of_decision_section,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=international_citizenship_questions,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=proa,model=EleutherAI_pythia-2.8b"
+        ]
+      },
+      {
+        "value": 0.3094366045219278,
+        "description": "min=0.141, mean=0.309, max=0.795, sum=1.547 (5)",
+        "style": {
+          "font-weight": "bold"
+        },
+        "markdown": false,
+        "run_spec_names": [
+          "legalbench:subset=abercrombie,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=corporate_lobbying,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=function_of_decision_section,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=international_citizenship_questions,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=proa,model=EleutherAI_pythia-2.8b"
+        ]
+      },
+      {
+        "value": 409.4,
+        "description": "min=95, mean=409.4, max=1000, sum=2047 (5)",
+        "style": {},
+        "markdown": false,
+        "run_spec_names": [
+          "legalbench:subset=abercrombie,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=corporate_lobbying,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=function_of_decision_section,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=international_citizenship_questions,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=proa,model=EleutherAI_pythia-2.8b"
+        ]
+      },
+      {
+        "value": 3.8604081632653062,
+        "description": "min=0.302, mean=3.86, max=5, sum=19.302 (5)",
+        "style": {},
+        "markdown": false,
+        "run_spec_names": [
+          "legalbench:subset=abercrombie,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=corporate_lobbying,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=function_of_decision_section,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=international_citizenship_questions,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=proa,model=EleutherAI_pythia-2.8b"
+        ]
+      },
+      {
+        "value": 0.002857142857142857,
+        "description": "min=0, mean=0.003, max=0.014, sum=0.014 (5)",
+        "style": {},
+        "markdown": false,
+        "run_spec_names": [
+          "legalbench:subset=abercrombie,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=corporate_lobbying,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=function_of_decision_section,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=international_citizenship_questions,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=proa,model=EleutherAI_pythia-2.8b"
+        ]
+      },
+      {
+        "value": 560.6440716343213,
+        "description": "min=206.779, mean=560.644, max=1497.455, sum=2803.22 (5)",
+        "style": {},
+        "markdown": false,
+        "run_spec_names": [
+          "legalbench:subset=abercrombie,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=corporate_lobbying,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=function_of_decision_section,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=international_citizenship_questions,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=proa,model=EleutherAI_pythia-2.8b"
+        ]
+      },
+      {
+        "value": 1.6417916921537006,
+        "description": "min=1, mean=1.642, max=4.124, sum=8.209 (5)",
+        "style": {},
+        "markdown": false,
+        "run_spec_names": [
+          "legalbench:subset=abercrombie,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=corporate_lobbying,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=function_of_decision_section,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=international_citizenship_questions,model=EleutherAI_pythia-2.8b",
+          "legalbench:subset=proa,model=EleutherAI_pythia-2.8b"
+        ]
+      }
+    ]
+  ],
+  "links": [
+    {
+      "text": "LaTeX",
+      "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/latex/legalbench_legalbench.tex"
+    },
+    {
+      "text": "JSON",
+      "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/json/legalbench_legalbench.json"
+    }
+  ],
+  "name": "legalbench"
+}

lite_pythia-2.8b-step5000/groups/json/legalbench_legalbench_subset:abercrombie.json

@@ -0,0 +1,145 @@
+{
+  "title": "subset: abercrombie",
+  "header": [
+    {
+      "value": "Model",
+      "markdown": false,
+      "metadata": {}
+    },
+    {
+      "value": "EM",
+      "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\nQuasi-exact match: Fraction of instances that the predicted output matches a correct reference up to light processing.",
+      "markdown": false,
+      "lower_is_better": false,
+      "metadata": {
+        "metric": "EM",
+        "run_group": "LegalBench"
+      }
+    },
+    {
+      "value": "Observed inference time (s)",
+      "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\nObserved inference runtime (s): Average observed time to process a request to the model (via an API, and thus depends on particular deployment).",
+      "markdown": false,
+      "lower_is_better": true,
+      "metadata": {
+        "metric": "Observed inference time (s)",
+        "run_group": "LegalBench"
+      }
+    },
+    {
+      "value": "# eval",
+      "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# eval: Number of evaluation instances.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# eval",
+        "run_group": "LegalBench"
+      }
+    },
+    {
+      "value": "# train",
+      "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# train: Number of training instances (e.g., in-context examples).",
+      "markdown": false,
+      "metadata": {
+        "metric": "# train",
+        "run_group": "LegalBench"
+      }
+    },
+    {
+      "value": "truncated",
+      "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\ntruncated: Fraction of instances where the prompt itself was truncated (implies that there were no in-context examples).",
+      "markdown": false,
+      "metadata": {
+        "metric": "truncated",
+        "run_group": "LegalBench"
+      }
+    },
+    {
+      "value": "# prompt tokens",
+      "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# prompt tokens: Number of tokens in the prompt.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# prompt tokens",
+        "run_group": "LegalBench"
+      }
+    },
+    {
+      "value": "# output tokens",
+      "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# output tokens: Actual number of output tokens.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# output tokens",
+        "run_group": "LegalBench"
+      }
+    }
+  ],
+  "rows": [
+    [
+      {
+        "value": "EleutherAI/pythia-2.8b",
+        "description": "",
+        "href": "?group=legalbench&subgroup=subset%3A%20abercrombie&runSpecs=%5B%22legalbench%3Asubset%3Dabercrombie%2Cmodel%3DEleutherAI_pythia-2.8b%22%5D",
+        "markdown": false,
+        "run_spec_names": [
+          "legalbench:subset=abercrombie,model=EleutherAI_pythia-2.8b"
+        ]
+      },
+      {
+        "value": 0.2,
+        "description": "min=0.2, mean=0.2, max=0.2, sum=0.2 (1)",
+        "style": {
+          "font-weight": "bold"
+        },
+        "markdown": false
+      },
+      {
+        "value": 0.1405831638135408,
+        "description": "min=0.141, mean=0.141, max=0.141, sum=0.141 (1)",
+        "style": {
+          "font-weight": "bold"
+        },
+        "markdown": false
+      },
+      {
+        "value": 95.0,
+        "description": "min=95, mean=95, max=95, sum=95 (1)",
+        "style": {},
+        "markdown": false
+      },
+      {
+        "value": 5.0,
+        "description": "min=5, mean=5, max=5, sum=5 (1)",
+        "style": {},
+        "markdown": false
+      },
+      {
+        "value": 0.0,
+        "description": "min=0, mean=0, max=0, sum=0 (1)",
+        "style": {},
+        "markdown": false
+      },
+      {
+        "value": 206.77894736842106,
+        "description": "min=206.779, mean=206.779, max=206.779, sum=206.779 (1)",
+        "style": {},
+        "markdown": false
+      },
+      {
+        "value": 1.0,
+        "description": "min=1, mean=1, max=1, sum=1 (1)",
+        "style": {},
+        "markdown": false
+      }
+    ]
+  ],
+  "links": [
+    {
+      "text": "LaTeX",
+      "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/latex/legalbench_legalbench_subset:abercrombie.tex"
+    },
+    {
+      "text": "JSON",
+      "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/json/legalbench_legalbench_subset:abercrombie.json"
+    }
+  ],
+  "name": "legalbench_subset:abercrombie"
+}

lite_pythia-2.8b-step5000/groups/json/legalbench_legalbench_subset:corporate_lobbying.json

@@ -0,0 +1,145 @@
+{
+  "title": "subset: corporate_lobbying",
+  "header": [
+    {
+      "value": "Model",
+      "markdown": false,
+      "metadata": {}
+    },
+    {
+      "value": "EM",
+      "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\nQuasi-exact match: Fraction of instances that the predicted output matches a correct reference up to light processing.",
+      "markdown": false,
+      "lower_is_better": false,
+      "metadata": {
+        "metric": "EM",
+        "run_group": "LegalBench"
+      }
+    },
+    {
+      "value": "Observed inference time (s)",
+      "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\nObserved inference runtime (s): Average observed time to process a request to the model (via an API, and thus depends on particular deployment).",
+      "markdown": false,
+      "lower_is_better": true,
+      "metadata": {
+        "metric": "Observed inference time (s)",
+        "run_group": "LegalBench"
+      }
+    },
+    {
+      "value": "# eval",
+      "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# eval: Number of evaluation instances.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# eval",
+        "run_group": "LegalBench"
+      }
+    },
+    {
+      "value": "# train",
+      "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# train: Number of training instances (e.g., in-context examples).",
+      "markdown": false,
+      "metadata": {
+        "metric": "# train",
+        "run_group": "LegalBench"
+      }
+    },
+    {
+      "value": "truncated",
+      "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\ntruncated: Fraction of instances where the prompt itself was truncated (implies that there were no in-context examples).",
+      "markdown": false,
+      "metadata": {
+        "metric": "truncated",
+        "run_group": "LegalBench"
+      }
+    },
+    {
+      "value": "# prompt tokens",
+      "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# prompt tokens: Number of tokens in the prompt.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# prompt tokens",
+        "run_group": "LegalBench"
+      }
+    },
+    {
+      "value": "# output tokens",
+      "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# output tokens: Actual number of output tokens.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# output tokens",
+        "run_group": "LegalBench"
+      }
+    }
+  ],
+  "rows": [
+    [
+      {
+        "value": "EleutherAI/pythia-2.8b",
+        "description": "",
+        "href": "?group=legalbench&subgroup=subset%3A%20corporate_lobbying&runSpecs=%5B%22legalbench%3Asubset%3Dcorporate_lobbying%2Cmodel%3DEleutherAI_pythia-2.8b%22%5D",
+        "markdown": false,
+        "run_spec_names": [
+          "legalbench:subset=corporate_lobbying,model=EleutherAI_pythia-2.8b"
+        ]
+      },
+      {
+        "value": 0.2,
+        "description": "min=0.2, mean=0.2, max=0.2, sum=0.2 (1)",
+        "style": {
+          "font-weight": "bold"
+        },
+        "markdown": false
+      },
+      {
+        "value": 0.7949594166814065,
+        "description": "min=0.795, mean=0.795, max=0.795, sum=0.795 (1)",
+        "style": {
+          "font-weight": "bold"
+        },
+        "markdown": false
+      },
+      {
+        "value": 490.0,
+        "description": "min=490, mean=490, max=490, sum=490 (1)",
+        "style": {},
+        "markdown": false
+      },
+      {
+        "value": 0.3020408163265306,
+        "description": "min=0.302, mean=0.302, max=0.302, sum=0.302 (1)",
+        "style": {},
+        "markdown": false
+      },
+      {
+        "value": 0.014285714285714285,
+        "description": "min=0.014, mean=0.014, max=0.014, sum=0.014 (1)",
+        "style": {},
+        "markdown": false
+      },
+      {
+        "value": 1497.4551020408164,
+        "description": "min=1497.455, mean=1497.455, max=1497.455, sum=1497.455 (1)",
+        "style": {},
+        "markdown": false
+      },
+      {
+        "value": 4.124489795918367,
+        "description": "min=4.124, mean=4.124, max=4.124, sum=4.124 (1)",
+        "style": {},
+        "markdown": false
+      }
+    ]
+  ],
+  "links": [
+    {
+      "text": "LaTeX",
+      "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/latex/legalbench_legalbench_subset:corporate_lobbying.tex"
+    },
+    {
+      "text": "JSON",
+      "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/json/legalbench_legalbench_subset:corporate_lobbying.json"
+    }
+  ],
+  "name": "legalbench_subset:corporate_lobbying"
+}

lite_pythia-2.8b-step5000/groups/json/legalbench_legalbench_subset:function_of_decision_section.json

@@ -0,0 +1,145 @@
+{
+  "title": "subset: function_of_decision_section",
+  "header": [
+    {
+      "value": "Model",
+      "markdown": false,
+      "metadata": {}
+    },
+    {
+      "value": "EM",
+      "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\nQuasi-exact match: Fraction of instances that the predicted output matches a correct reference up to light processing.",
+      "markdown": false,
+      "lower_is_better": false,
+      "metadata": {
+        "metric": "EM",
+        "run_group": "LegalBench"
+      }
+    },
+    {
+      "value": "Observed inference time (s)",
+      "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\nObserved inference runtime (s): Average observed time to process a request to the model (via an API, and thus depends on particular deployment).",
+      "markdown": false,
+      "lower_is_better": true,
+      "metadata": {
+        "metric": "Observed inference time (s)",
+        "run_group": "LegalBench"
+      }
+    },
+    {
+      "value": "# eval",
+      "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# eval: Number of evaluation instances.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# eval",
+        "run_group": "LegalBench"
+      }
+    },
+    {
+      "value": "# train",
+      "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# train: Number of training instances (e.g., in-context examples).",
+      "markdown": false,
+      "metadata": {
+        "metric": "# train",
+        "run_group": "LegalBench"
+      }
+    },
+    {
+      "value": "truncated",
+      "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\ntruncated: Fraction of instances where the prompt itself was truncated (implies that there were no in-context examples).",
+      "markdown": false,
+      "metadata": {
+        "metric": "truncated",
+        "run_group": "LegalBench"
+      }
+    },
+    {
+      "value": "# prompt tokens",
+      "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# prompt tokens: Number of tokens in the prompt.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# prompt tokens",
+        "run_group": "LegalBench"
+      }
+    },
+    {
+      "value": "# output tokens",
+      "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# output tokens: Actual number of output tokens.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# output tokens",
+        "run_group": "LegalBench"
+      }
+    }
+  ],
+  "rows": [
+    [
+      {
+        "value": "EleutherAI/pythia-2.8b",
+        "description": "",
+        "href": "?group=legalbench&subgroup=subset%3A%20function_of_decision_section&runSpecs=%5B%22legalbench%3Asubset%3Dfunction_of_decision_section%2Cmodel%3DEleutherAI_pythia-2.8b%22%5D",
+        "markdown": false,
+        "run_spec_names": [
+          "legalbench:subset=function_of_decision_section,model=EleutherAI_pythia-2.8b"
+        ]
+      },
+      {
+        "value": 0.1444141689373297,
+        "description": "min=0.144, mean=0.144, max=0.144, sum=0.144 (1)",
+        "style": {
+          "font-weight": "bold"
+        },
+        "markdown": false
+      },
+      {
+        "value": 0.2725935534495424,
+        "description": "min=0.273, mean=0.273, max=0.273, sum=0.273 (1)",
+        "style": {
+          "font-weight": "bold"
+        },
+        "markdown": false
+      },
+      {
+        "value": 367.0,
+        "description": "min=367, mean=367, max=367, sum=367 (1)",
+        "style": {},
+        "markdown": false
+      },
+      {
+        "value": 5.0,
+        "description": "min=5, mean=5, max=5, sum=5 (1)",
+        "style": {},
+        "markdown": false
+      },
+      {
+        "value": 0.0,
+        "description": "min=0, mean=0, max=0, sum=0 (1)",
+        "style": {},
+        "markdown": false
+      },
+      {
+        "value": 514.9182561307902,
+        "description": "min=514.918, mean=514.918, max=514.918, sum=514.918 (1)",
+        "style": {},
+        "markdown": false
+      },
+      {
+        "value": 1.0844686648501363,
+        "description": "min=1.084, mean=1.084, max=1.084, sum=1.084 (1)",
+        "style": {},
+        "markdown": false
+      }
+    ]
+  ],
+  "links": [
+    {
+      "text": "LaTeX",
+      "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/latex/legalbench_legalbench_subset:function_of_decision_section.tex"
+    },
+    {
+      "text": "JSON",
+      "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/json/legalbench_legalbench_subset:function_of_decision_section.json"
+    }
+  ],
+  "name": "legalbench_subset:function_of_decision_section"
+}

lite_pythia-2.8b-step5000/groups/json/legalbench_legalbench_subset:international_citizenship_questions.json

@@ -0,0 +1,145 @@
+{
+  "title": "subset: international_citizenship_questions",
+  "header": [
+    {
+      "value": "Model",
+      "markdown": false,
+      "metadata": {}
+    },
+    {
+      "value": "EM",
+      "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\nQuasi-exact match: Fraction of instances that the predicted output matches a correct reference up to light processing.",
+      "markdown": false,
+      "lower_is_better": false,
+      "metadata": {
+        "metric": "EM",
+        "run_group": "LegalBench"
+      }
+    },
+    {
+      "value": "Observed inference time (s)",
+      "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\nObserved inference runtime (s): Average observed time to process a request to the model (via an API, and thus depends on particular deployment).",
+      "markdown": false,
+      "lower_is_better": true,
+      "metadata": {
+        "metric": "Observed inference time (s)",
+        "run_group": "LegalBench"
+      }
+    },
+    {
+      "value": "# eval",
+      "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# eval: Number of evaluation instances.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# eval",
+        "run_group": "LegalBench"
+      }
+    },
+    {
+      "value": "# train",
+      "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# train: Number of training instances (e.g., in-context examples).",
+      "markdown": false,
+      "metadata": {
+        "metric": "# train",
+        "run_group": "LegalBench"
+      }
+    },
+    {
+      "value": "truncated",
+      "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\ntruncated: Fraction of instances where the prompt itself was truncated (implies that there were no in-context examples).",
+      "markdown": false,
+      "metadata": {
+        "metric": "truncated",
+        "run_group": "LegalBench"
+      }
+    },
+    {
+      "value": "# prompt tokens",
+      "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# prompt tokens: Number of tokens in the prompt.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# prompt tokens",
+        "run_group": "LegalBench"
+      }
+    },
+    {
+      "value": "# output tokens",
+      "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# output tokens: Actual number of output tokens.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# output tokens",
+        "run_group": "LegalBench"
+      }
+    }
+  ],
+  "rows": [
+    [
+      {
+        "value": "EleutherAI/pythia-2.8b",
+        "description": "",
+        "href": "?group=legalbench&subgroup=subset%3A%20international_citizenship_questions&runSpecs=%5B%22legalbench%3Asubset%3Dinternational_citizenship_questions%2Cmodel%3DEleutherAI_pythia-2.8b%22%5D",
+        "markdown": false,
+        "run_spec_names": [
+          "legalbench:subset=international_citizenship_questions,model=EleutherAI_pythia-2.8b"
+        ]
+      },
+      {
+        "value": 0.619,
+        "description": "min=0.619, mean=0.619, max=0.619, sum=0.619 (1)",
+        "style": {
+          "font-weight": "bold"
+        },
+        "markdown": false
+      },
+      {
+        "value": 0.15224194407463074,
+        "description": "min=0.152, mean=0.152, max=0.152, sum=0.152 (1)",
+        "style": {
+          "font-weight": "bold"
+        },
+        "markdown": false
+      },
+      {
+        "value": 1000.0,
+        "description": "min=1000, mean=1000, max=1000, sum=1000 (1)",
+        "style": {},
+        "markdown": false
+      },
+      {
+        "value": 4.0,
+        "description": "min=4, mean=4, max=4, sum=4 (1)",
+        "style": {},
+        "markdown": false
+      },
+      {
+        "value": 0.0,
+        "description": "min=0, mean=0, max=0, sum=0 (1)",
+        "style": {},
+        "markdown": false
+      },
+      {
+        "value": 250.447,
+        "description": "min=250.447, mean=250.447, max=250.447, sum=250.447 (1)",
+        "style": {},
+        "markdown": false
+      },
+      {
+        "value": 1.0,
+        "description": "min=1, mean=1, max=1, sum=1 (1)",
+        "style": {},
+        "markdown": false
+      }
+    ]
+  ],
+  "links": [
+    {
+      "text": "LaTeX",
+      "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/latex/legalbench_legalbench_subset:international_citizenship_questions.tex"
+    },
+    {
+      "text": "JSON",
+      "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/json/legalbench_legalbench_subset:international_citizenship_questions.json"
+    }
+  ],
+  "name": "legalbench_subset:international_citizenship_questions"
+}

lite_pythia-2.8b-step5000/groups/json/legalbench_legalbench_subset:proa.json

@@ -0,0 +1,145 @@
+{
+  "title": "subset: proa",
+  "header": [
+    {
+      "value": "Model",
+      "markdown": false,
+      "metadata": {}
+    },
+    {
+      "value": "EM",
+      "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\nQuasi-exact match: Fraction of instances that the predicted output matches a correct reference up to light processing.",
+      "markdown": false,
+      "lower_is_better": false,
+      "metadata": {
+        "metric": "EM",
+        "run_group": "LegalBench"
+      }
+    },
+    {
+      "value": "Observed inference time (s)",
+      "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\nObserved inference runtime (s): Average observed time to process a request to the model (via an API, and thus depends on particular deployment).",
+      "markdown": false,
+      "lower_is_better": true,
+      "metadata": {
+        "metric": "Observed inference time (s)",
+        "run_group": "LegalBench"
+      }
+    },
+    {
+      "value": "# eval",
+      "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# eval: Number of evaluation instances.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# eval",
+        "run_group": "LegalBench"
+      }
+    },
+    {
+      "value": "# train",
+      "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# train: Number of training instances (e.g., in-context examples).",
+      "markdown": false,
+      "metadata": {
+        "metric": "# train",
+        "run_group": "LegalBench"
+      }
+    },
+    {
+      "value": "truncated",
+      "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\ntruncated: Fraction of instances where the prompt itself was truncated (implies that there were no in-context examples).",
+      "markdown": false,
+      "metadata": {
+        "metric": "truncated",
+        "run_group": "LegalBench"
+      }
+    },
+    {
+      "value": "# prompt tokens",
+      "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# prompt tokens: Number of tokens in the prompt.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# prompt tokens",
+        "run_group": "LegalBench"
+      }
+    },
+    {
+      "value": "# output tokens",
+      "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# output tokens: Actual number of output tokens.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# output tokens",
+        "run_group": "LegalBench"
+      }
+    }
+  ],
+  "rows": [
+    [
+      {
+        "value": "EleutherAI/pythia-2.8b",
+        "description": "",
+        "href": "?group=legalbench&subgroup=subset%3A%20proa&runSpecs=%5B%22legalbench%3Asubset%3Dproa%2Cmodel%3DEleutherAI_pythia-2.8b%22%5D",
+        "markdown": false,
+        "run_spec_names": [
+          "legalbench:subset=proa,model=EleutherAI_pythia-2.8b"
+        ]
+      },
+      {
+        "value": 0.43157894736842106,
+        "description": "min=0.432, mean=0.432, max=0.432, sum=0.432 (1)",
+        "style": {
+          "font-weight": "bold"
+        },
+        "markdown": false
+      },
+      {
+        "value": 0.18680494459051836,
+        "description": "min=0.187, mean=0.187, max=0.187, sum=0.187 (1)",
+        "style": {
+          "font-weight": "bold"
+        },
+        "markdown": false
+      },
+      {
+        "value": 95.0,
+        "description": "min=95, mean=95, max=95, sum=95 (1)",
+        "style": {},
+        "markdown": false
+      },
+      {
+        "value": 5.0,
+        "description": "min=5, mean=5, max=5, sum=5 (1)",
+        "style": {},
+        "markdown": false
+      },
+      {
+        "value": 0.0,
+        "description": "min=0, mean=0, max=0, sum=0 (1)",
+        "style": {},
+        "markdown": false
+      },
+      {
+        "value": 333.62105263157895,
+        "description": "min=333.621, mean=333.621, max=333.621, sum=333.621 (1)",
+        "style": {},
+        "markdown": false
+      },
+      {
+        "value": 1.0,
+        "description": "min=1, mean=1, max=1, sum=1 (1)",
+        "style": {},
+        "markdown": false
+      }
+    ]
+  ],
+  "links": [
+    {
+      "text": "LaTeX",
+      "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/latex/legalbench_legalbench_subset:proa.tex"
+    },
+    {
+      "text": "JSON",
+      "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/json/legalbench_legalbench_subset:proa.json"
+    }
+  ],
+  "name": "legalbench_subset:proa"
+}

lite_pythia-2.8b-step5000/groups/json/mmlu_mmlu.json

@@ -0,0 +1,190 @@
+{
+  "title": "MMLU (Massive Multitask Language Understanding)",
+  "header": [
+    {
+      "value": "Model",
+      "markdown": false,
+      "metadata": {}
+    },
+    {
+      "value": "EM",
+      "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\nExact match: Fraction of instances that the predicted output matches a correct reference exactly.",
+      "markdown": false,
+      "lower_is_better": false,
+      "metadata": {
+        "metric": "EM",
+        "run_group": "MMLU"
+      }
+    },
+    {
+      "value": "Observed inference time (s)",
+      "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\nObserved inference runtime (s): Average observed time to process a request to the model (via an API, and thus depends on particular deployment).",
+      "markdown": false,
+      "lower_is_better": true,
+      "metadata": {
+        "metric": "Observed inference time (s)",
+        "run_group": "MMLU"
+      }
+    },
+    {
+      "value": "# eval",
+      "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# eval: Number of evaluation instances.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# eval",
+        "run_group": "MMLU"
+      }
+    },
+    {
+      "value": "# train",
+      "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# train: Number of training instances (e.g., in-context examples).",
+      "markdown": false,
+      "metadata": {
+        "metric": "# train",
+        "run_group": "MMLU"
+      }
+    },
+    {
+      "value": "truncated",
+      "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\ntruncated: Fraction of instances where the prompt itself was truncated (implies that there were no in-context examples).",
+      "markdown": false,
+      "metadata": {
+        "metric": "truncated",
+        "run_group": "MMLU"
+      }
+    },
+    {
+      "value": "# prompt tokens",
+      "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# prompt tokens: Number of tokens in the prompt.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# prompt tokens",
+        "run_group": "MMLU"
+      }
+    },
+    {
+      "value": "# output tokens",
+      "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# output tokens: Actual number of output tokens.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# output tokens",
+        "run_group": "MMLU"
+      }
+    }
+  ],
+  "rows": [
+    [
+      {
+        "value": "EleutherAI/pythia-2.8b",
+        "description": "",
+        "markdown": false
+      },
+      {
+        "value": 0.27487719298245616,
+        "description": "min=0.25, mean=0.275, max=0.32, sum=1.374 (5)",
+        "style": {
+          "font-weight": "bold"
+        },
+        "markdown": false,
+        "run_spec_names": [
+          "mmlu:subject=abstract_algebra,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=college_chemistry,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=computer_security,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=econometrics,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=us_foreign_policy,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b"
+        ]
+      },
+      {
+        "value": 0.22061018314696193,
+        "description": "min=0.173, mean=0.221, max=0.282, sum=1.103 (5)",
+        "style": {
+          "font-weight": "bold"
+        },
+        "markdown": false,
+        "run_spec_names": [
+          "mmlu:subject=abstract_algebra,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=college_chemistry,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=computer_security,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=econometrics,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=us_foreign_policy,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b"
+        ]
+      },
+      {
+        "value": 102.8,
+        "description": "min=100, mean=102.8, max=114, sum=514 (5)",
+        "style": {},
+        "markdown": false,
+        "run_spec_names": [
+          "mmlu:subject=abstract_algebra,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=college_chemistry,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=computer_security,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=econometrics,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=us_foreign_policy,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b"
+        ]
+      },
+      {
+        "value": 5.0,
+        "description": "min=5, mean=5, max=5, sum=25 (5)",
+        "style": {},
+        "markdown": false,
+        "run_spec_names": [
+          "mmlu:subject=abstract_algebra,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=college_chemistry,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=computer_security,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=econometrics,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=us_foreign_policy,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b"
+        ]
+      },
+      {
+        "value": 0.0,
+        "description": "min=0, mean=0, max=0, sum=0 (5)",
+        "style": {},
+        "markdown": false,
+        "run_spec_names": [
+          "mmlu:subject=abstract_algebra,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=college_chemistry,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=computer_security,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=econometrics,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=us_foreign_policy,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b"
+        ]
+      },
+      {
+        "value": 467.935649122807,
+        "description": "min=358.76, mean=467.936, max=612.798, sum=2339.678 (5)",
+        "style": {},
+        "markdown": false,
+        "run_spec_names": [
+          "mmlu:subject=abstract_algebra,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=college_chemistry,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=computer_security,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=econometrics,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=us_foreign_policy,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b"
+        ]
+      },
+      {
+        "value": 1.0,
+        "description": "min=1, mean=1, max=1, sum=5 (5)",
+        "style": {},
+        "markdown": false,
+        "run_spec_names": [
+          "mmlu:subject=abstract_algebra,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=college_chemistry,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=computer_security,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=econometrics,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+          "mmlu:subject=us_foreign_policy,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b"
+        ]
+      }
+    ]
+  ],
+  "links": [
+    {
+      "text": "LaTeX",
+      "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/latex/mmlu_mmlu.tex"
+    },
+    {
+      "text": "JSON",
+      "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/json/mmlu_mmlu.json"
+    }
+  ],
+  "name": "mmlu"
+}

lite_pythia-2.8b-step5000/groups/json/mmlu_mmlu_subject:abstract_algebra.json

@@ -0,0 +1,145 @@
+{
+  "title": "subject: abstract_algebra",
+  "header": [
+    {
+      "value": "Model",
+      "markdown": false,
+      "metadata": {}
+    },
+    {
+      "value": "EM",
+      "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\nExact match: Fraction of instances that the predicted output matches a correct reference exactly.",
+      "markdown": false,
+      "lower_is_better": false,
+      "metadata": {
+        "metric": "EM",
+        "run_group": "MMLU"
+      }
+    },
+    {
+      "value": "Observed inference time (s)",
+      "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\nObserved inference runtime (s): Average observed time to process a request to the model (via an API, and thus depends on particular deployment).",
+      "markdown": false,
+      "lower_is_better": true,
+      "metadata": {
+        "metric": "Observed inference time (s)",
+        "run_group": "MMLU"
+      }
+    },
+    {
+      "value": "# eval",
+      "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# eval: Number of evaluation instances.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# eval",
+        "run_group": "MMLU"
+      }
+    },
+    {
+      "value": "# train",
+      "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# train: Number of training instances (e.g., in-context examples).",
+      "markdown": false,
+      "metadata": {
+        "metric": "# train",
+        "run_group": "MMLU"
+      }
+    },
+    {
+      "value": "truncated",
+      "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\ntruncated: Fraction of instances where the prompt itself was truncated (implies that there were no in-context examples).",
+      "markdown": false,
+      "metadata": {
+        "metric": "truncated",
+        "run_group": "MMLU"
+      }
+    },
+    {
+      "value": "# prompt tokens",
+      "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# prompt tokens: Number of tokens in the prompt.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# prompt tokens",
+        "run_group": "MMLU"
+      }
+    },
+    {
+      "value": "# output tokens",
+      "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# output tokens: Actual number of output tokens.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# output tokens",
+        "run_group": "MMLU"
+      }
+    }
+  ],
+  "rows": [
+    [
+      {
+        "value": "EleutherAI/pythia-2.8b",
+        "description": "",
+        "href": "?group=mmlu&subgroup=subject%3A%20abstract_algebra&runSpecs=%5B%22mmlu%3Asubject%3Dabstract_algebra%2Cmethod%3Dmultiple_choice_joint%2Cmodel%3DEleutherAI_pythia-2.8b%22%5D",
+        "markdown": false,
+        "run_spec_names": [
+          "mmlu:subject=abstract_algebra,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b"
+        ]
+      },
+      {
+        "value": 0.27,
+        "description": "min=0.27, mean=0.27, max=0.27, sum=0.27 (1)",
+        "style": {
+          "font-weight": "bold"
+        },
+        "markdown": false
+      },
+      {
+        "value": 0.17254346132278442,
+        "description": "min=0.173, mean=0.173, max=0.173, sum=0.173 (1)",
+        "style": {
+          "font-weight": "bold"
+        },
+        "markdown": false
+      },
+      {
+        "value": 100.0,
+        "description": "min=100, mean=100, max=100, sum=100 (1)",
+        "style": {},
+        "markdown": false
+      },
+      {
+        "value": 5.0,
+        "description": "min=5, mean=5, max=5, sum=5 (1)",
+        "style": {},
+        "markdown": false
+      },
+      {
+        "value": 0.0,
+        "description": "min=0, mean=0, max=0, sum=0 (1)",
+        "style": {},
+        "markdown": false
+      },
+      {
+        "value": 358.76,
+        "description": "min=358.76, mean=358.76, max=358.76, sum=358.76 (1)",
+        "style": {},
+        "markdown": false
+      },
+      {
+        "value": 1.0,
+        "description": "min=1, mean=1, max=1, sum=1 (1)",
+        "style": {},
+        "markdown": false
+      }
+    ]
+  ],
+  "links": [
+    {
+      "text": "LaTeX",
+      "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/latex/mmlu_mmlu_subject:abstract_algebra.tex"
+    },
+    {
+      "text": "JSON",
+      "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/json/mmlu_mmlu_subject:abstract_algebra.json"
+    }
+  ],
+  "name": "mmlu_subject:abstract_algebra"
+}

lite_pythia-2.8b-step5000/groups/json/mmlu_mmlu_subject:college_chemistry.json

@@ -0,0 +1,145 @@
+{
+  "title": "subject: college_chemistry",
+  "header": [
+    {
+      "value": "Model",
+      "markdown": false,
+      "metadata": {}
+    },
+    {
+      "value": "EM",
+      "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\nExact match: Fraction of instances that the predicted output matches a correct reference exactly.",
+      "markdown": false,
+      "lower_is_better": false,
+      "metadata": {
+        "metric": "EM",
+        "run_group": "MMLU"
+      }
+    },
+    {
+      "value": "Observed inference time (s)",
+      "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\nObserved inference runtime (s): Average observed time to process a request to the model (via an API, and thus depends on particular deployment).",
+      "markdown": false,
+      "lower_is_better": true,
+      "metadata": {
+        "metric": "Observed inference time (s)",
+        "run_group": "MMLU"
+      }
+    },
+    {
+      "value": "# eval",
+      "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# eval: Number of evaluation instances.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# eval",
+        "run_group": "MMLU"
+      }
+    },
+    {
+      "value": "# train",
+      "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# train: Number of training instances (e.g., in-context examples).",
+      "markdown": false,
+      "metadata": {
+        "metric": "# train",
+        "run_group": "MMLU"
+      }
+    },
+    {
+      "value": "truncated",
+      "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\ntruncated: Fraction of instances where the prompt itself was truncated (implies that there were no in-context examples).",
+      "markdown": false,
+      "metadata": {
+        "metric": "truncated",
+        "run_group": "MMLU"
+      }
+    },
+    {
+      "value": "# prompt tokens",
+      "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# prompt tokens: Number of tokens in the prompt.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# prompt tokens",
+        "run_group": "MMLU"
+      }
+    },
+    {
+      "value": "# output tokens",
+      "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# output tokens: Actual number of output tokens.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# output tokens",
+        "run_group": "MMLU"
+      }
+    }
+  ],
+  "rows": [
+    [
+      {
+        "value": "EleutherAI/pythia-2.8b",
+        "description": "",
+        "href": "?group=mmlu&subgroup=subject%3A%20college_chemistry&runSpecs=%5B%22mmlu%3Asubject%3Dcollege_chemistry%2Cmethod%3Dmultiple_choice_joint%2Cmodel%3DEleutherAI_pythia-2.8b%22%5D",
+        "markdown": false,
+        "run_spec_names": [
+          "mmlu:subject=college_chemistry,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b"
+        ]
+      },
+      {
+        "value": 0.32,
+        "description": "min=0.32, mean=0.32, max=0.32, sum=0.32 (1)",
+        "style": {
+          "font-weight": "bold"
+        },
+        "markdown": false
+      },
+      {
+        "value": 0.2506118679046631,
+        "description": "min=0.251, mean=0.251, max=0.251, sum=0.251 (1)",
+        "style": {
+          "font-weight": "bold"
+        },
+        "markdown": false
+      },
+      {
+        "value": 100.0,
+        "description": "min=100, mean=100, max=100, sum=100 (1)",
+        "style": {},
+        "markdown": false
+      },
+      {
+        "value": 5.0,
+        "description": "min=5, mean=5, max=5, sum=5 (1)",
+        "style": {},
+        "markdown": false
+      },
+      {
+        "value": 0.0,
+        "description": "min=0, mean=0, max=0, sum=0 (1)",
+        "style": {},
+        "markdown": false
+      },
+      {
+        "value": 535.85,
+        "description": "min=535.85, mean=535.85, max=535.85, sum=535.85 (1)",
+        "style": {},
+        "markdown": false
+      },
+      {
+        "value": 1.0,
+        "description": "min=1, mean=1, max=1, sum=1 (1)",
+        "style": {},
+        "markdown": false
+      }
+    ]
+  ],
+  "links": [
+    {
+      "text": "LaTeX",
+      "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/latex/mmlu_mmlu_subject:college_chemistry.tex"
+    },
+    {
+      "text": "JSON",
+      "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/json/mmlu_mmlu_subject:college_chemistry.json"
+    }
+  ],
+  "name": "mmlu_subject:college_chemistry"
+}

lite_pythia-2.8b-step5000/groups/json/mmlu_mmlu_subject:computer_security.json

@@ -0,0 +1,145 @@
+{
+  "title": "subject: computer_security",
+  "header": [
+    {
+      "value": "Model",
+      "markdown": false,
+      "metadata": {}
+    },
+    {
+      "value": "EM",
+      "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\nExact match: Fraction of instances that the predicted output matches a correct reference exactly.",
+      "markdown": false,
+      "lower_is_better": false,
+      "metadata": {
+        "metric": "EM",
+        "run_group": "MMLU"
+      }
+    },
+    {
+      "value": "Observed inference time (s)",
+      "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\nObserved inference runtime (s): Average observed time to process a request to the model (via an API, and thus depends on particular deployment).",
+      "markdown": false,
+      "lower_is_better": true,
+      "metadata": {
+        "metric": "Observed inference time (s)",
+        "run_group": "MMLU"
+      }
+    },
+    {
+      "value": "# eval",
+      "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# eval: Number of evaluation instances.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# eval",
+        "run_group": "MMLU"
+      }
+    },
+    {
+      "value": "# train",
+      "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# train: Number of training instances (e.g., in-context examples).",
+      "markdown": false,
+      "metadata": {
+        "metric": "# train",
+        "run_group": "MMLU"
+      }
+    },
+    {
+      "value": "truncated",
+      "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\ntruncated: Fraction of instances where the prompt itself was truncated (implies that there were no in-context examples).",
+      "markdown": false,
+      "metadata": {
+        "metric": "truncated",
+        "run_group": "MMLU"
+      }
+    },
+    {
+      "value": "# prompt tokens",
+      "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# prompt tokens: Number of tokens in the prompt.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# prompt tokens",
+        "run_group": "MMLU"
+      }
+    },
+    {
+      "value": "# output tokens",
+      "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# output tokens: Actual number of output tokens.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# output tokens",
+        "run_group": "MMLU"
+      }
+    }
+  ],
+  "rows": [
+    [
+      {
+        "value": "EleutherAI/pythia-2.8b",
+        "description": "",
+        "href": "?group=mmlu&subgroup=subject%3A%20computer_security&runSpecs=%5B%22mmlu%3Asubject%3Dcomputer_security%2Cmethod%3Dmultiple_choice_joint%2Cmodel%3DEleutherAI_pythia-2.8b%22%5D",
+        "markdown": false,
+        "run_spec_names": [
+          "mmlu:subject=computer_security,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b"
+        ]
+      },
+      {
+        "value": 0.25,
+        "description": "min=0.25, mean=0.25, max=0.25, sum=0.25 (1)",
+        "style": {
+          "font-weight": "bold"
+        },
+        "markdown": false
+      },
+      {
+        "value": 0.18395774602890014,
+        "description": "min=0.184, mean=0.184, max=0.184, sum=0.184 (1)",
+        "style": {
+          "font-weight": "bold"
+        },
+        "markdown": false
+      },
+      {
+        "value": 100.0,
+        "description": "min=100, mean=100, max=100, sum=100 (1)",
+        "style": {},
+        "markdown": false
+      },
+      {
+        "value": 5.0,
+        "description": "min=5, mean=5, max=5, sum=5 (1)",
+        "style": {},
+        "markdown": false
+      },
+      {
+        "value": 0.0,
+        "description": "min=0, mean=0, max=0, sum=0 (1)",
+        "style": {},
+        "markdown": false
+      },
+      {
+        "value": 388.19,
+        "description": "min=388.19, mean=388.19, max=388.19, sum=388.19 (1)",
+        "style": {},
+        "markdown": false
+      },
+      {
+        "value": 1.0,
+        "description": "min=1, mean=1, max=1, sum=1 (1)",
+        "style": {},
+        "markdown": false
+      }
+    ]
+  ],
+  "links": [
+    {
+      "text": "LaTeX",
+      "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/latex/mmlu_mmlu_subject:computer_security.tex"
+    },
+    {
+      "text": "JSON",
+      "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/json/mmlu_mmlu_subject:computer_security.json"
+    }
+  ],
+  "name": "mmlu_subject:computer_security"
+}

lite_pythia-2.8b-step5000/groups/json/mmlu_mmlu_subject:econometrics.json

@@ -0,0 +1,145 @@
+{
+  "title": "subject: econometrics",
+  "header": [
+    {
+      "value": "Model",
+      "markdown": false,
+      "metadata": {}
+    },
+    {
+      "value": "EM",
+      "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\nExact match: Fraction of instances that the predicted output matches a correct reference exactly.",
+      "markdown": false,
+      "lower_is_better": false,
+      "metadata": {
+        "metric": "EM",
+        "run_group": "MMLU"
+      }
+    },
+    {
+      "value": "Observed inference time (s)",
+      "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\nObserved inference runtime (s): Average observed time to process a request to the model (via an API, and thus depends on particular deployment).",
+      "markdown": false,
+      "lower_is_better": true,
+      "metadata": {
+        "metric": "Observed inference time (s)",
+        "run_group": "MMLU"
+      }
+    },
+    {
+      "value": "# eval",
+      "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# eval: Number of evaluation instances.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# eval",
+        "run_group": "MMLU"
+      }
+    },
+    {
+      "value": "# train",
+      "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# train: Number of training instances (e.g., in-context examples).",
+      "markdown": false,
+      "metadata": {
+        "metric": "# train",
+        "run_group": "MMLU"
+      }
+    },
+    {
+      "value": "truncated",
+      "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\ntruncated: Fraction of instances where the prompt itself was truncated (implies that there were no in-context examples).",
+      "markdown": false,
+      "metadata": {
+        "metric": "truncated",
+        "run_group": "MMLU"
+      }
+    },
+    {
+      "value": "# prompt tokens",
+      "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# prompt tokens: Number of tokens in the prompt.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# prompt tokens",
+        "run_group": "MMLU"
+      }
+    },
+    {
+      "value": "# output tokens",
+      "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# output tokens: Actual number of output tokens.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# output tokens",
+        "run_group": "MMLU"
+      }
+    }
+  ],
+  "rows": [
+    [
+      {
+        "value": "EleutherAI/pythia-2.8b",
+        "description": "",
+        "href": "?group=mmlu&subgroup=subject%3A%20econometrics&runSpecs=%5B%22mmlu%3Asubject%3Deconometrics%2Cmethod%3Dmultiple_choice_joint%2Cmodel%3DEleutherAI_pythia-2.8b%22%5D",
+        "markdown": false,
+        "run_spec_names": [
+          "mmlu:subject=econometrics,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b"
+        ]
+      },
+      {
+        "value": 0.2543859649122807,
+        "description": "min=0.254, mean=0.254, max=0.254, sum=0.254 (1)",
+        "style": {
+          "font-weight": "bold"
+        },
+        "markdown": false
+      },
+      {
+        "value": 0.28234320356134784,
+        "description": "min=0.282, mean=0.282, max=0.282, sum=0.282 (1)",
+        "style": {
+          "font-weight": "bold"
+        },
+        "markdown": false
+      },
+      {
+        "value": 114.0,
+        "description": "min=114, mean=114, max=114, sum=114 (1)",
+        "style": {},
+        "markdown": false
+      },
+      {
+        "value": 5.0,
+        "description": "min=5, mean=5, max=5, sum=5 (1)",
+        "style": {},
+        "markdown": false
+      },
+      {
+        "value": 0.0,
+        "description": "min=0, mean=0, max=0, sum=0 (1)",
+        "style": {},
+        "markdown": false
+      },
+      {
+        "value": 612.7982456140351,
+        "description": "min=612.798, mean=612.798, max=612.798, sum=612.798 (1)",
+        "style": {},
+        "markdown": false
+      },
+      {
+        "value": 1.0,
+        "description": "min=1, mean=1, max=1, sum=1 (1)",
+        "style": {},
+        "markdown": false
+      }
+    ]
+  ],
+  "links": [
+    {
+      "text": "LaTeX",
+      "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/latex/mmlu_mmlu_subject:econometrics.tex"
+    },
+    {
+      "text": "JSON",
+      "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/json/mmlu_mmlu_subject:econometrics.json"
+    }
+  ],
+  "name": "mmlu_subject:econometrics"
+}

lite_pythia-2.8b-step5000/groups/json/mmlu_mmlu_subject:us_foreign_policy.json

@@ -0,0 +1,145 @@
+{
+  "title": "subject: us_foreign_policy",
+  "header": [
+    {
+      "value": "Model",
+      "markdown": false,
+      "metadata": {}
+    },
+    {
+      "value": "EM",
+      "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\nExact match: Fraction of instances that the predicted output matches a correct reference exactly.",
+      "markdown": false,
+      "lower_is_better": false,
+      "metadata": {
+        "metric": "EM",
+        "run_group": "MMLU"
+      }
+    },
+    {
+      "value": "Observed inference time (s)",
+      "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\nObserved inference runtime (s): Average observed time to process a request to the model (via an API, and thus depends on particular deployment).",
+      "markdown": false,
+      "lower_is_better": true,
+      "metadata": {
+        "metric": "Observed inference time (s)",
+        "run_group": "MMLU"
+      }
+    },
+    {
+      "value": "# eval",
+      "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# eval: Number of evaluation instances.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# eval",
+        "run_group": "MMLU"
+      }
+    },
+    {
+      "value": "# train",
+      "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# train: Number of training instances (e.g., in-context examples).",
+      "markdown": false,
+      "metadata": {
+        "metric": "# train",
+        "run_group": "MMLU"
+      }
+    },
+    {
+      "value": "truncated",
+      "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\ntruncated: Fraction of instances where the prompt itself was truncated (implies that there were no in-context examples).",
+      "markdown": false,
+      "metadata": {
+        "metric": "truncated",
+        "run_group": "MMLU"
+      }
+    },
+    {
+      "value": "# prompt tokens",
+      "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# prompt tokens: Number of tokens in the prompt.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# prompt tokens",
+        "run_group": "MMLU"
+      }
+    },
+    {
+      "value": "# output tokens",
+      "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# output tokens: Actual number of output tokens.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# output tokens",
+        "run_group": "MMLU"
+      }
+    }
+  ],
+  "rows": [
+    [
+      {
+        "value": "EleutherAI/pythia-2.8b",
+        "description": "",
+        "href": "?group=mmlu&subgroup=subject%3A%20us_foreign_policy&runSpecs=%5B%22mmlu%3Asubject%3Dus_foreign_policy%2Cmethod%3Dmultiple_choice_joint%2Cmodel%3DEleutherAI_pythia-2.8b%22%5D",
+        "markdown": false,
+        "run_spec_names": [
+          "mmlu:subject=us_foreign_policy,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b"
+        ]
+      },
+      {
+        "value": 0.28,
+        "description": "min=0.28, mean=0.28, max=0.28, sum=0.28 (1)",
+        "style": {
+          "font-weight": "bold"
+        },
+        "markdown": false
+      },
+      {
+        "value": 0.21359463691711425,
+        "description": "min=0.214, mean=0.214, max=0.214, sum=0.214 (1)",
+        "style": {
+          "font-weight": "bold"
+        },
+        "markdown": false
+      },
+      {
+        "value": 100.0,
+        "description": "min=100, mean=100, max=100, sum=100 (1)",
+        "style": {},
+        "markdown": false
+      },
+      {
+        "value": 5.0,
+        "description": "min=5, mean=5, max=5, sum=5 (1)",
+        "style": {},
+        "markdown": false
+      },
+      {
+        "value": 0.0,
+        "description": "min=0, mean=0, max=0, sum=0 (1)",
+        "style": {},
+        "markdown": false
+      },
+      {
+        "value": 444.08,
+        "description": "min=444.08, mean=444.08, max=444.08, sum=444.08 (1)",
+        "style": {},
+        "markdown": false
+      },
+      {
+        "value": 1.0,
+        "description": "min=1, mean=1, max=1, sum=1 (1)",
+        "style": {},
+        "markdown": false
+      }
+    ]
+  ],
+  "links": [
+    {
+      "text": "LaTeX",
+      "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/latex/mmlu_mmlu_subject:us_foreign_policy.tex"
+    },
+    {
+      "text": "JSON",
+      "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/json/mmlu_mmlu_subject:us_foreign_policy.json"
+    }
+  ],
+  "name": "mmlu_subject:us_foreign_policy"
+}

lite_pythia-2.8b-step5000/groups/json/openbookqa_openbookqa_.json

@@ -0,0 +1,145 @@
+{
+  "title": "",
+  "header": [
+    {
+      "value": "Model",
+      "markdown": false,
+      "metadata": {}
+    },
+    {
+      "value": "EM",
+      "description": "The OpenbookQA benchmark for commonsense-intensive open book question answering [(Mihaylov et al., 2018)](https://aclanthology.org/D18-1260/).\n\nExact match: Fraction of instances that the predicted output matches a correct reference exactly.",
+      "markdown": false,
+      "lower_is_better": false,
+      "metadata": {
+        "metric": "EM",
+        "run_group": "OpenbookQA"
+      }
+    },
+    {
+      "value": "Observed inference time (s)",
+      "description": "The OpenbookQA benchmark for commonsense-intensive open book question answering [(Mihaylov et al., 2018)](https://aclanthology.org/D18-1260/).\n\nObserved inference runtime (s): Average observed time to process a request to the model (via an API, and thus depends on particular deployment).",
+      "markdown": false,
+      "lower_is_better": true,
+      "metadata": {
+        "metric": "Observed inference time (s)",
+        "run_group": "OpenbookQA"
+      }
+    },
+    {
+      "value": "# eval",
+      "description": "The OpenbookQA benchmark for commonsense-intensive open book question answering [(Mihaylov et al., 2018)](https://aclanthology.org/D18-1260/).\n\n# eval: Number of evaluation instances.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# eval",
+        "run_group": "OpenbookQA"
+      }
+    },
+    {
+      "value": "# train",
+      "description": "The OpenbookQA benchmark for commonsense-intensive open book question answering [(Mihaylov et al., 2018)](https://aclanthology.org/D18-1260/).\n\n# train: Number of training instances (e.g., in-context examples).",
+      "markdown": false,
+      "metadata": {
+        "metric": "# train",
+        "run_group": "OpenbookQA"
+      }
+    },
+    {
+      "value": "truncated",
+      "description": "The OpenbookQA benchmark for commonsense-intensive open book question answering [(Mihaylov et al., 2018)](https://aclanthology.org/D18-1260/).\n\ntruncated: Fraction of instances where the prompt itself was truncated (implies that there were no in-context examples).",
+      "markdown": false,
+      "metadata": {
+        "metric": "truncated",
+        "run_group": "OpenbookQA"
+      }
+    },
+    {
+      "value": "# prompt tokens",
+      "description": "The OpenbookQA benchmark for commonsense-intensive open book question answering [(Mihaylov et al., 2018)](https://aclanthology.org/D18-1260/).\n\n# prompt tokens: Number of tokens in the prompt.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# prompt tokens",
+        "run_group": "OpenbookQA"
+      }
+    },
+    {
+      "value": "# output tokens",
+      "description": "The OpenbookQA benchmark for commonsense-intensive open book question answering [(Mihaylov et al., 2018)](https://aclanthology.org/D18-1260/).\n\n# output tokens: Actual number of output tokens.",
+      "markdown": false,
+      "metadata": {
+        "metric": "# output tokens",
+        "run_group": "OpenbookQA"
+      }
+    }
+  ],
+  "rows": [
+    [
+      {
+        "value": "EleutherAI/pythia-2.8b",
+        "description": "",
+        "href": "?group=openbookqa&subgroup=&runSpecs=%5B%22commonsense%3Adataset%3Dopenbookqa%2Cmethod%3Dmultiple_choice_joint%2Cmodel%3DEleutherAI_pythia-2.8b%22%5D",
+        "markdown": false,
+        "run_spec_names": [
+          "commonsense:dataset=openbookqa,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b"
+        ]
+      },
+      {
+        "value": 0.22,
+        "description": "min=0.22, mean=0.22, max=0.22, sum=0.22 (1)",
+        "style": {
+          "font-weight": "bold"
+        },
+        "markdown": false
+      },
+      {
+        "value": 0.12883714818954467,
+        "description": "min=0.129, mean=0.129, max=0.129, sum=0.129 (1)",
+        "style": {
+          "font-weight": "bold"
+        },
+        "markdown": false
+      },
+      {
+        "value": 500.0,
+        "description": "min=500, mean=500, max=500, sum=500 (1)",
+        "style": {},
+        "markdown": false
+      },
+      {
+        "value": 5.0,
+        "description": "min=5, mean=5, max=5, sum=5 (1)",
+        "style": {},
+        "markdown": false
+      },
+      {
+        "value": 0.0,
+        "description": "min=0, mean=0, max=0, sum=0 (1)",
+        "style": {},
+        "markdown": false
+      },
+      {
+        "value": 251.556,
+        "description": "min=251.556, mean=251.556, max=251.556, sum=251.556 (1)",
+        "style": {},
+        "markdown": false
+      },
+      {
+        "value": 1.0,
+        "description": "min=1, mean=1, max=1, sum=1 (1)",
+        "style": {},
+        "markdown": false
+      }
+    ]
+  ],
+  "links": [
+    {
+      "text": "LaTeX",
+      "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/latex/openbookqa_openbookqa_.tex"
+    },
+    {
+      "text": "JSON",
+      "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/json/openbookqa_openbookqa_.json"
+    }
+  ],
+  "name": "openbookqa_"
+}

lite_pythia-2.8b-step5000/groups/latex/core_scenarios_accuracy.tex

@@ -0,0 +1,12 @@
+\begin{table*}[htp]
+\resizebox{\textwidth}{!}{
+\begin{tabular}{lrrrr}
+\toprule
+Model & OpenbookQA - EM & MMLU - EM & GSM8K - EM & LegalBench - EM \\
+\midrule
+EleutherAI/pythia-2.8b & 0.22 & 0.27487719298245616 & 0.01 & 0.3189986232611502 \\
+\bottomrule
+\end{tabular}}
+\caption{Results for accuracy (core_scenarios)}
+\label{fig:accuracy (core_scenarios)}
+\end{table*}

lite_pythia-2.8b-step5000/groups/latex/core_scenarios_efficiency.tex

@@ -0,0 +1,12 @@
+\begin{table*}[htp]
+\resizebox{\textwidth}{!}{
+\begin{tabular}{lrrrr}
+\toprule
+Model & OpenbookQA - Observed inference time (s) & MMLU - Observed inference time (s) & GSM8K - Observed inference time (s) & LegalBench - Observed inference time (s) \\
+\midrule
+EleutherAI/pythia-2.8b & 0.12883714818954467 & 0.22061018314696193 & 2.5314217054843904 & 0.3094366045219278 \\
+\bottomrule
+\end{tabular}}
+\caption{Results for efficiency (core_scenarios)}
+\label{fig:efficiency (core_scenarios)}
+\end{table*}

lite_pythia-2.8b-step5000/groups/latex/core_scenarios_general_information.tex

@@ -0,0 +1,12 @@
+\begin{table*}[htp]
+\resizebox{\textwidth}{!}{
+\begin{tabular}{lrrrrrrrrrrrrrrrrrrrr}
+\toprule
+Model & OpenbookQA - # eval & OpenbookQA - # train & OpenbookQA - truncated & OpenbookQA - # prompt tokens & OpenbookQA - # output tokens & MMLU - # eval & MMLU - # train & MMLU - truncated & MMLU - # prompt tokens & MMLU - # output tokens & GSM8K - # eval & GSM8K - # train & GSM8K - truncated & GSM8K - # prompt tokens & GSM8K - # output tokens & LegalBench - # eval & LegalBench - # train & LegalBench - truncated & LegalBench - # prompt tokens & LegalBench - # output tokens \\
+\midrule
+EleutherAI/pythia-2.8b & 500.0 & 5.0 &  & 251.556 & 1.0 & 102.8 & 5.0 &  & 467.935649122807 & 1.0 & 1000.0 & 5.0 &  & 939.582 & 168.459 & 409.4 & 3.8604081632653062 & 0.002857142857142857 & 560.6440716343213 & 1.6417916921537006 \\
+\bottomrule
+\end{tabular}}
+\caption{Results for general_information (core_scenarios)}
+\label{fig:general_information (core_scenarios)}
+\end{table*}

lite_pythia-2.8b-step5000/groups/latex/gsm_gsm_.tex

@@ -0,0 +1,12 @@
+\begin{table*}[htp]
+\resizebox{\textwidth}{!}{
+\begin{tabular}{lrrrrrrr}
+\toprule
+Model & EM & Observed inference time (s) & # eval & # train & truncated & # prompt tokens & # output tokens \\
+\midrule
+EleutherAI/pythia-2.8b & 0.01 & 2.5314217054843904 & 1000.0 & 5.0 &  & 939.582 & 168.459 \\
+\bottomrule
+\end{tabular}}
+\caption{Results for gsm_ (gsm)}
+\label{fig:gsm_ (gsm)}
+\end{table*}

lite_pythia-2.8b-step5000/groups/latex/legalbench_legalbench.tex

@@ -0,0 +1,12 @@
+\begin{table*}[htp]
+\resizebox{\textwidth}{!}{
+\begin{tabular}{lrrrrrrr}
+\toprule
+Model & EM & Observed inference time (s) & # eval & # train & truncated & # prompt tokens & # output tokens \\
+\midrule
+EleutherAI/pythia-2.8b & 0.3189986232611502 & 0.3094366045219278 & 409.4 & 3.8604081632653062 & 0.002857142857142857 & 560.6440716343213 & 1.6417916921537006 \\
+\bottomrule
+\end{tabular}}
+\caption{Results for legalbench (legalbench)}
+\label{fig:legalbench (legalbench)}
+\end{table*}

lite_pythia-2.8b-step5000/groups/latex/legalbench_legalbench_subset:abercrombie.tex

@@ -0,0 +1,12 @@
+\begin{table*}[htp]
+\resizebox{\textwidth}{!}{
+\begin{tabular}{lrrrrrrr}
+\toprule
+Model & EM & Observed inference time (s) & # eval & # train & truncated & # prompt tokens & # output tokens \\
+\midrule
+EleutherAI/pythia-2.8b & 0.2 & 0.1405831638135408 & 95.0 & 5.0 &  & 206.77894736842106 & 1.0 \\
+\bottomrule
+\end{tabular}}
+\caption{Results for legalbench_subset:abercrombie (legalbench)}
+\label{fig:legalbench_subset:abercrombie (legalbench)}
+\end{table*}

lite_pythia-2.8b-step5000/groups/latex/legalbench_legalbench_subset:corporate_lobbying.tex

@@ -0,0 +1,12 @@
+\begin{table*}[htp]
+\resizebox{\textwidth}{!}{
+\begin{tabular}{lrrrrrrr}
+\toprule
+Model & EM & Observed inference time (s) & # eval & # train & truncated & # prompt tokens & # output tokens \\
+\midrule
+EleutherAI/pythia-2.8b & 0.2 & 0.7949594166814065 & 490.0 & 0.3020408163265306 & 0.014285714285714285 & 1497.4551020408164 & 4.124489795918367 \\
+\bottomrule
+\end{tabular}}
+\caption{Results for legalbench_subset:corporate_lobbying (legalbench)}
+\label{fig:legalbench_subset:corporate_lobbying (legalbench)}
+\end{table*}

lite_pythia-2.8b-step5000/groups/latex/legalbench_legalbench_subset:function_of_decision_section.tex

@@ -0,0 +1,12 @@
+\begin{table*}[htp]
+\resizebox{\textwidth}{!}{
+\begin{tabular}{lrrrrrrr}
+\toprule
+Model & EM & Observed inference time (s) & # eval & # train & truncated & # prompt tokens & # output tokens \\
+\midrule
+EleutherAI/pythia-2.8b & 0.1444141689373297 & 0.2725935534495424 & 367.0 & 5.0 &  & 514.9182561307902 & 1.0844686648501363 \\
+\bottomrule
+\end{tabular}}
+\caption{Results for legalbench_subset:function_of_decision_section (legalbench)}
+\label{fig:legalbench_subset:function_of_decision_section (legalbench)}
+\end{table*}

lite_pythia-2.8b-step5000/groups/latex/legalbench_legalbench_subset:international_citizenship_questions.tex

@@ -0,0 +1,12 @@
+\begin{table*}[htp]
+\resizebox{\textwidth}{!}{
+\begin{tabular}{lrrrrrrr}
+\toprule
+Model & EM & Observed inference time (s) & # eval & # train & truncated & # prompt tokens & # output tokens \\
+\midrule
+EleutherAI/pythia-2.8b & 0.619 & 0.15224194407463074 & 1000.0 & 4.0 &  & 250.447 & 1.0 \\
+\bottomrule
+\end{tabular}}
+\caption{Results for legalbench_subset:international_citizenship_questions (legalbench)}
+\label{fig:legalbench_subset:international_citizenship_questions (legalbench)}
+\end{table*}

lite_pythia-2.8b-step5000/groups/latex/legalbench_legalbench_subset:proa.tex

@@ -0,0 +1,12 @@
+\begin{table*}[htp]
+\resizebox{\textwidth}{!}{
+\begin{tabular}{lrrrrrrr}
+\toprule
+Model & EM & Observed inference time (s) & # eval & # train & truncated & # prompt tokens & # output tokens \\
+\midrule
+EleutherAI/pythia-2.8b & 0.43157894736842106 & 0.18680494459051836 & 95.0 & 5.0 &  & 333.62105263157895 & 1.0 \\
+\bottomrule
+\end{tabular}}
+\caption{Results for legalbench_subset:proa (legalbench)}
+\label{fig:legalbench_subset:proa (legalbench)}
+\end{table*}

lite_pythia-2.8b-step5000/groups/latex/mmlu_mmlu.tex

@@ -0,0 +1,12 @@
+\begin{table*}[htp]
+\resizebox{\textwidth}{!}{
+\begin{tabular}{lrrrrrrr}
+\toprule
+Model & EM & Observed inference time (s) & # eval & # train & truncated & # prompt tokens & # output tokens \\
+\midrule
+EleutherAI/pythia-2.8b & 0.27487719298245616 & 0.22061018314696193 & 102.8 & 5.0 &  & 467.935649122807 & 1.0 \\
+\bottomrule
+\end{tabular}}
+\caption{Results for mmlu (mmlu)}
+\label{fig:mmlu (mmlu)}
+\end{table*}

lite_pythia-2.8b-step5000/groups/latex/mmlu_mmlu_subject:abstract_algebra.tex

@@ -0,0 +1,12 @@
+\begin{table*}[htp]
+\resizebox{\textwidth}{!}{
+\begin{tabular}{lrrrrrrr}
+\toprule
+Model & EM & Observed inference time (s) & # eval & # train & truncated & # prompt tokens & # output tokens \\
+\midrule
+EleutherAI/pythia-2.8b & 0.27 & 0.17254346132278442 & 100.0 & 5.0 &  & 358.76 & 1.0 \\
+\bottomrule
+\end{tabular}}
+\caption{Results for mmlu_subject:abstract_algebra (mmlu)}
+\label{fig:mmlu_subject:abstract_algebra (mmlu)}
+\end{table*}

lite_pythia-2.8b-step5000/groups/latex/mmlu_mmlu_subject:college_chemistry.tex

@@ -0,0 +1,12 @@
+\begin{table*}[htp]
+\resizebox{\textwidth}{!}{
+\begin{tabular}{lrrrrrrr}
+\toprule
+Model & EM & Observed inference time (s) & # eval & # train & truncated & # prompt tokens & # output tokens \\
+\midrule
+EleutherAI/pythia-2.8b & 0.32 & 0.2506118679046631 & 100.0 & 5.0 &  & 535.85 & 1.0 \\
+\bottomrule
+\end{tabular}}
+\caption{Results for mmlu_subject:college_chemistry (mmlu)}
+\label{fig:mmlu_subject:college_chemistry (mmlu)}
+\end{table*}

lite_pythia-2.8b-step5000/groups/latex/mmlu_mmlu_subject:computer_security.tex

@@ -0,0 +1,12 @@
+\begin{table*}[htp]
+\resizebox{\textwidth}{!}{
+\begin{tabular}{lrrrrrrr}
+\toprule
+Model & EM & Observed inference time (s) & # eval & # train & truncated & # prompt tokens & # output tokens \\
+\midrule
+EleutherAI/pythia-2.8b & 0.25 & 0.18395774602890014 & 100.0 & 5.0 &  & 388.19 & 1.0 \\
+\bottomrule
+\end{tabular}}
+\caption{Results for mmlu_subject:computer_security (mmlu)}
+\label{fig:mmlu_subject:computer_security (mmlu)}
+\end{table*}

lite_pythia-2.8b-step5000/groups/latex/mmlu_mmlu_subject:econometrics.tex

@@ -0,0 +1,12 @@
+\begin{table*}[htp]
+\resizebox{\textwidth}{!}{
+\begin{tabular}{lrrrrrrr}
+\toprule
+Model & EM & Observed inference time (s) & # eval & # train & truncated & # prompt tokens & # output tokens \\
+\midrule
+EleutherAI/pythia-2.8b & 0.2543859649122807 & 0.28234320356134784 & 114.0 & 5.0 &  & 612.7982456140351 & 1.0 \\
+\bottomrule
+\end{tabular}}
+\caption{Results for mmlu_subject:econometrics (mmlu)}
+\label{fig:mmlu_subject:econometrics (mmlu)}
+\end{table*}

lite_pythia-2.8b-step5000/groups/latex/mmlu_mmlu_subject:us_foreign_policy.tex

@@ -0,0 +1,12 @@
+\begin{table*}[htp]
+\resizebox{\textwidth}{!}{
+\begin{tabular}{lrrrrrrr}
+\toprule
+Model & EM & Observed inference time (s) & # eval & # train & truncated & # prompt tokens & # output tokens \\
+\midrule
+EleutherAI/pythia-2.8b & 0.28 & 0.21359463691711425 & 100.0 & 5.0 &  & 444.08 & 1.0 \\
+\bottomrule
+\end{tabular}}
+\caption{Results for mmlu_subject:us_foreign_policy (mmlu)}
+\label{fig:mmlu_subject:us_foreign_policy (mmlu)}
+\end{table*}

lite_pythia-2.8b-step5000/groups/latex/openbookqa_openbookqa_.tex

@@ -0,0 +1,12 @@
+\begin{table*}[htp]
+\resizebox{\textwidth}{!}{
+\begin{tabular}{lrrrrrrr}
+\toprule
+Model & EM & Observed inference time (s) & # eval & # train & truncated & # prompt tokens & # output tokens \\
+\midrule
+EleutherAI/pythia-2.8b & 0.22 & 0.12883714818954467 & 500.0 & 5.0 &  & 251.556 & 1.0 \\
+\bottomrule
+\end{tabular}}
+\caption{Results for openbookqa_ (openbookqa)}
+\label{fig:openbookqa_ (openbookqa)}
+\end{table*}

lite_pythia-2.8b-step5000/groups/legalbench.json

@@ -0,0 +1,917 @@
+[
+  {
+    "title": "LegalBench",
+    "header": [
+      {
+        "value": "Model",
+        "markdown": false,
+        "metadata": {}
+      },
+      {
+        "value": "EM",
+        "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\nQuasi-exact match: Fraction of instances that the predicted output matches a correct reference up to light processing.",
+        "markdown": false,
+        "lower_is_better": false,
+        "metadata": {
+          "metric": "EM",
+          "run_group": "LegalBench"
+        }
+      },
+      {
+        "value": "Observed inference time (s)",
+        "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\nObserved inference runtime (s): Average observed time to process a request to the model (via an API, and thus depends on particular deployment).",
+        "markdown": false,
+        "lower_is_better": true,
+        "metadata": {
+          "metric": "Observed inference time (s)",
+          "run_group": "LegalBench"
+        }
+      },
+      {
+        "value": "# eval",
+        "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# eval: Number of evaluation instances.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# eval",
+          "run_group": "LegalBench"
+        }
+      },
+      {
+        "value": "# train",
+        "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# train: Number of training instances (e.g., in-context examples).",
+        "markdown": false,
+        "metadata": {
+          "metric": "# train",
+          "run_group": "LegalBench"
+        }
+      },
+      {
+        "value": "truncated",
+        "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\ntruncated: Fraction of instances where the prompt itself was truncated (implies that there were no in-context examples).",
+        "markdown": false,
+        "metadata": {
+          "metric": "truncated",
+          "run_group": "LegalBench"
+        }
+      },
+      {
+        "value": "# prompt tokens",
+        "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# prompt tokens: Number of tokens in the prompt.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# prompt tokens",
+          "run_group": "LegalBench"
+        }
+      },
+      {
+        "value": "# output tokens",
+        "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# output tokens: Actual number of output tokens.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# output tokens",
+          "run_group": "LegalBench"
+        }
+      }
+    ],
+    "rows": [
+      [
+        {
+          "value": "EleutherAI/pythia-2.8b",
+          "description": "",
+          "markdown": false
+        },
+        {
+          "value": 0.3189986232611502,
+          "description": "min=0.144, mean=0.319, max=0.619, sum=1.595 (5)",
+          "style": {
+            "font-weight": "bold"
+          },
+          "markdown": false,
+          "run_spec_names": [
+            "legalbench:subset=abercrombie,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=corporate_lobbying,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=function_of_decision_section,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=international_citizenship_questions,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=proa,model=EleutherAI_pythia-2.8b"
+          ]
+        },
+        {
+          "value": 0.3094366045219278,
+          "description": "min=0.141, mean=0.309, max=0.795, sum=1.547 (5)",
+          "style": {
+            "font-weight": "bold"
+          },
+          "markdown": false,
+          "run_spec_names": [
+            "legalbench:subset=abercrombie,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=corporate_lobbying,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=function_of_decision_section,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=international_citizenship_questions,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=proa,model=EleutherAI_pythia-2.8b"
+          ]
+        },
+        {
+          "value": 409.4,
+          "description": "min=95, mean=409.4, max=1000, sum=2047 (5)",
+          "style": {},
+          "markdown": false,
+          "run_spec_names": [
+            "legalbench:subset=abercrombie,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=corporate_lobbying,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=function_of_decision_section,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=international_citizenship_questions,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=proa,model=EleutherAI_pythia-2.8b"
+          ]
+        },
+        {
+          "value": 3.8604081632653062,
+          "description": "min=0.302, mean=3.86, max=5, sum=19.302 (5)",
+          "style": {},
+          "markdown": false,
+          "run_spec_names": [
+            "legalbench:subset=abercrombie,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=corporate_lobbying,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=function_of_decision_section,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=international_citizenship_questions,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=proa,model=EleutherAI_pythia-2.8b"
+          ]
+        },
+        {
+          "value": 0.002857142857142857,
+          "description": "min=0, mean=0.003, max=0.014, sum=0.014 (5)",
+          "style": {},
+          "markdown": false,
+          "run_spec_names": [
+            "legalbench:subset=abercrombie,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=corporate_lobbying,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=function_of_decision_section,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=international_citizenship_questions,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=proa,model=EleutherAI_pythia-2.8b"
+          ]
+        },
+        {
+          "value": 560.6440716343213,
+          "description": "min=206.779, mean=560.644, max=1497.455, sum=2803.22 (5)",
+          "style": {},
+          "markdown": false,
+          "run_spec_names": [
+            "legalbench:subset=abercrombie,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=corporate_lobbying,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=function_of_decision_section,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=international_citizenship_questions,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=proa,model=EleutherAI_pythia-2.8b"
+          ]
+        },
+        {
+          "value": 1.6417916921537006,
+          "description": "min=1, mean=1.642, max=4.124, sum=8.209 (5)",
+          "style": {},
+          "markdown": false,
+          "run_spec_names": [
+            "legalbench:subset=abercrombie,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=corporate_lobbying,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=function_of_decision_section,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=international_citizenship_questions,model=EleutherAI_pythia-2.8b",
+            "legalbench:subset=proa,model=EleutherAI_pythia-2.8b"
+          ]
+        }
+      ]
+    ],
+    "links": [
+      {
+        "text": "LaTeX",
+        "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/latex/legalbench_legalbench.tex"
+      },
+      {
+        "text": "JSON",
+        "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/json/legalbench_legalbench.json"
+      }
+    ],
+    "name": "legalbench"
+  },
+  {
+    "title": "subset: abercrombie",
+    "header": [
+      {
+        "value": "Model",
+        "markdown": false,
+        "metadata": {}
+      },
+      {
+        "value": "EM",
+        "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\nQuasi-exact match: Fraction of instances that the predicted output matches a correct reference up to light processing.",
+        "markdown": false,
+        "lower_is_better": false,
+        "metadata": {
+          "metric": "EM",
+          "run_group": "LegalBench"
+        }
+      },
+      {
+        "value": "Observed inference time (s)",
+        "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\nObserved inference runtime (s): Average observed time to process a request to the model (via an API, and thus depends on particular deployment).",
+        "markdown": false,
+        "lower_is_better": true,
+        "metadata": {
+          "metric": "Observed inference time (s)",
+          "run_group": "LegalBench"
+        }
+      },
+      {
+        "value": "# eval",
+        "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# eval: Number of evaluation instances.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# eval",
+          "run_group": "LegalBench"
+        }
+      },
+      {
+        "value": "# train",
+        "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# train: Number of training instances (e.g., in-context examples).",
+        "markdown": false,
+        "metadata": {
+          "metric": "# train",
+          "run_group": "LegalBench"
+        }
+      },
+      {
+        "value": "truncated",
+        "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\ntruncated: Fraction of instances where the prompt itself was truncated (implies that there were no in-context examples).",
+        "markdown": false,
+        "metadata": {
+          "metric": "truncated",
+          "run_group": "LegalBench"
+        }
+      },
+      {
+        "value": "# prompt tokens",
+        "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# prompt tokens: Number of tokens in the prompt.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# prompt tokens",
+          "run_group": "LegalBench"
+        }
+      },
+      {
+        "value": "# output tokens",
+        "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# output tokens: Actual number of output tokens.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# output tokens",
+          "run_group": "LegalBench"
+        }
+      }
+    ],
+    "rows": [
+      [
+        {
+          "value": "EleutherAI/pythia-2.8b",
+          "description": "",
+          "href": "?group=legalbench&subgroup=subset%3A%20abercrombie&runSpecs=%5B%22legalbench%3Asubset%3Dabercrombie%2Cmodel%3DEleutherAI_pythia-2.8b%22%5D",
+          "markdown": false,
+          "run_spec_names": [
+            "legalbench:subset=abercrombie,model=EleutherAI_pythia-2.8b"
+          ]
+        },
+        {
+          "value": 0.2,
+          "description": "min=0.2, mean=0.2, max=0.2, sum=0.2 (1)",
+          "style": {
+            "font-weight": "bold"
+          },
+          "markdown": false
+        },
+        {
+          "value": 0.1405831638135408,
+          "description": "min=0.141, mean=0.141, max=0.141, sum=0.141 (1)",
+          "style": {
+            "font-weight": "bold"
+          },
+          "markdown": false
+        },
+        {
+          "value": 95.0,
+          "description": "min=95, mean=95, max=95, sum=95 (1)",
+          "style": {},
+          "markdown": false
+        },
+        {
+          "value": 5.0,
+          "description": "min=5, mean=5, max=5, sum=5 (1)",
+          "style": {},
+          "markdown": false
+        },
+        {
+          "value": 0.0,
+          "description": "min=0, mean=0, max=0, sum=0 (1)",
+          "style": {},
+          "markdown": false
+        },
+        {
+          "value": 206.77894736842106,
+          "description": "min=206.779, mean=206.779, max=206.779, sum=206.779 (1)",
+          "style": {},
+          "markdown": false
+        },
+        {
+          "value": 1.0,
+          "description": "min=1, mean=1, max=1, sum=1 (1)",
+          "style": {},
+          "markdown": false
+        }
+      ]
+    ],
+    "links": [
+      {
+        "text": "LaTeX",
+        "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/latex/legalbench_legalbench_subset:abercrombie.tex"
+      },
+      {
+        "text": "JSON",
+        "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/json/legalbench_legalbench_subset:abercrombie.json"
+      }
+    ],
+    "name": "legalbench_subset:abercrombie"
+  },
+  {
+    "title": "subset: corporate_lobbying",
+    "header": [
+      {
+        "value": "Model",
+        "markdown": false,
+        "metadata": {}
+      },
+      {
+        "value": "EM",
+        "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\nQuasi-exact match: Fraction of instances that the predicted output matches a correct reference up to light processing.",
+        "markdown": false,
+        "lower_is_better": false,
+        "metadata": {
+          "metric": "EM",
+          "run_group": "LegalBench"
+        }
+      },
+      {
+        "value": "Observed inference time (s)",
+        "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\nObserved inference runtime (s): Average observed time to process a request to the model (via an API, and thus depends on particular deployment).",
+        "markdown": false,
+        "lower_is_better": true,
+        "metadata": {
+          "metric": "Observed inference time (s)",
+          "run_group": "LegalBench"
+        }
+      },
+      {
+        "value": "# eval",
+        "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# eval: Number of evaluation instances.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# eval",
+          "run_group": "LegalBench"
+        }
+      },
+      {
+        "value": "# train",
+        "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# train: Number of training instances (e.g., in-context examples).",
+        "markdown": false,
+        "metadata": {
+          "metric": "# train",
+          "run_group": "LegalBench"
+        }
+      },
+      {
+        "value": "truncated",
+        "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\ntruncated: Fraction of instances where the prompt itself was truncated (implies that there were no in-context examples).",
+        "markdown": false,
+        "metadata": {
+          "metric": "truncated",
+          "run_group": "LegalBench"
+        }
+      },
+      {
+        "value": "# prompt tokens",
+        "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# prompt tokens: Number of tokens in the prompt.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# prompt tokens",
+          "run_group": "LegalBench"
+        }
+      },
+      {
+        "value": "# output tokens",
+        "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# output tokens: Actual number of output tokens.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# output tokens",
+          "run_group": "LegalBench"
+        }
+      }
+    ],
+    "rows": [
+      [
+        {
+          "value": "EleutherAI/pythia-2.8b",
+          "description": "",
+          "href": "?group=legalbench&subgroup=subset%3A%20corporate_lobbying&runSpecs=%5B%22legalbench%3Asubset%3Dcorporate_lobbying%2Cmodel%3DEleutherAI_pythia-2.8b%22%5D",
+          "markdown": false,
+          "run_spec_names": [
+            "legalbench:subset=corporate_lobbying,model=EleutherAI_pythia-2.8b"
+          ]
+        },
+        {
+          "value": 0.2,
+          "description": "min=0.2, mean=0.2, max=0.2, sum=0.2 (1)",
+          "style": {
+            "font-weight": "bold"
+          },
+          "markdown": false
+        },
+        {
+          "value": 0.7949594166814065,
+          "description": "min=0.795, mean=0.795, max=0.795, sum=0.795 (1)",
+          "style": {
+            "font-weight": "bold"
+          },
+          "markdown": false
+        },
+        {
+          "value": 490.0,
+          "description": "min=490, mean=490, max=490, sum=490 (1)",
+          "style": {},
+          "markdown": false
+        },
+        {
+          "value": 0.3020408163265306,
+          "description": "min=0.302, mean=0.302, max=0.302, sum=0.302 (1)",
+          "style": {},
+          "markdown": false
+        },
+        {
+          "value": 0.014285714285714285,
+          "description": "min=0.014, mean=0.014, max=0.014, sum=0.014 (1)",
+          "style": {},
+          "markdown": false
+        },
+        {
+          "value": 1497.4551020408164,
+          "description": "min=1497.455, mean=1497.455, max=1497.455, sum=1497.455 (1)",
+          "style": {},
+          "markdown": false
+        },
+        {
+          "value": 4.124489795918367,
+          "description": "min=4.124, mean=4.124, max=4.124, sum=4.124 (1)",
+          "style": {},
+          "markdown": false
+        }
+      ]
+    ],
+    "links": [
+      {
+        "text": "LaTeX",
+        "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/latex/legalbench_legalbench_subset:corporate_lobbying.tex"
+      },
+      {
+        "text": "JSON",
+        "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/json/legalbench_legalbench_subset:corporate_lobbying.json"
+      }
+    ],
+    "name": "legalbench_subset:corporate_lobbying"
+  },
+  {
+    "title": "subset: function_of_decision_section",
+    "header": [
+      {
+        "value": "Model",
+        "markdown": false,
+        "metadata": {}
+      },
+      {
+        "value": "EM",
+        "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\nQuasi-exact match: Fraction of instances that the predicted output matches a correct reference up to light processing.",
+        "markdown": false,
+        "lower_is_better": false,
+        "metadata": {
+          "metric": "EM",
+          "run_group": "LegalBench"
+        }
+      },
+      {
+        "value": "Observed inference time (s)",
+        "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\nObserved inference runtime (s): Average observed time to process a request to the model (via an API, and thus depends on particular deployment).",
+        "markdown": false,
+        "lower_is_better": true,
+        "metadata": {
+          "metric": "Observed inference time (s)",
+          "run_group": "LegalBench"
+        }
+      },
+      {
+        "value": "# eval",
+        "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# eval: Number of evaluation instances.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# eval",
+          "run_group": "LegalBench"
+        }
+      },
+      {
+        "value": "# train",
+        "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# train: Number of training instances (e.g., in-context examples).",
+        "markdown": false,
+        "metadata": {
+          "metric": "# train",
+          "run_group": "LegalBench"
+        }
+      },
+      {
+        "value": "truncated",
+        "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\ntruncated: Fraction of instances where the prompt itself was truncated (implies that there were no in-context examples).",
+        "markdown": false,
+        "metadata": {
+          "metric": "truncated",
+          "run_group": "LegalBench"
+        }
+      },
+      {
+        "value": "# prompt tokens",
+        "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# prompt tokens: Number of tokens in the prompt.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# prompt tokens",
+          "run_group": "LegalBench"
+        }
+      },
+      {
+        "value": "# output tokens",
+        "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# output tokens: Actual number of output tokens.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# output tokens",
+          "run_group": "LegalBench"
+        }
+      }
+    ],
+    "rows": [
+      [
+        {
+          "value": "EleutherAI/pythia-2.8b",
+          "description": "",
+          "href": "?group=legalbench&subgroup=subset%3A%20function_of_decision_section&runSpecs=%5B%22legalbench%3Asubset%3Dfunction_of_decision_section%2Cmodel%3DEleutherAI_pythia-2.8b%22%5D",
+          "markdown": false,
+          "run_spec_names": [
+            "legalbench:subset=function_of_decision_section,model=EleutherAI_pythia-2.8b"
+          ]
+        },
+        {
+          "value": 0.1444141689373297,
+          "description": "min=0.144, mean=0.144, max=0.144, sum=0.144 (1)",
+          "style": {
+            "font-weight": "bold"
+          },
+          "markdown": false
+        },
+        {
+          "value": 0.2725935534495424,
+          "description": "min=0.273, mean=0.273, max=0.273, sum=0.273 (1)",
+          "style": {
+            "font-weight": "bold"
+          },
+          "markdown": false
+        },
+        {
+          "value": 367.0,
+          "description": "min=367, mean=367, max=367, sum=367 (1)",
+          "style": {},
+          "markdown": false
+        },
+        {
+          "value": 5.0,
+          "description": "min=5, mean=5, max=5, sum=5 (1)",
+          "style": {},
+          "markdown": false
+        },
+        {
+          "value": 0.0,
+          "description": "min=0, mean=0, max=0, sum=0 (1)",
+          "style": {},
+          "markdown": false
+        },
+        {
+          "value": 514.9182561307902,
+          "description": "min=514.918, mean=514.918, max=514.918, sum=514.918 (1)",
+          "style": {},
+          "markdown": false
+        },
+        {
+          "value": 1.0844686648501363,
+          "description": "min=1.084, mean=1.084, max=1.084, sum=1.084 (1)",
+          "style": {},
+          "markdown": false
+        }
+      ]
+    ],
+    "links": [
+      {
+        "text": "LaTeX",
+        "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/latex/legalbench_legalbench_subset:function_of_decision_section.tex"
+      },
+      {
+        "text": "JSON",
+        "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/json/legalbench_legalbench_subset:function_of_decision_section.json"
+      }
+    ],
+    "name": "legalbench_subset:function_of_decision_section"
+  },
+  {
+    "title": "subset: international_citizenship_questions",
+    "header": [
+      {
+        "value": "Model",
+        "markdown": false,
+        "metadata": {}
+      },
+      {
+        "value": "EM",
+        "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\nQuasi-exact match: Fraction of instances that the predicted output matches a correct reference up to light processing.",
+        "markdown": false,
+        "lower_is_better": false,
+        "metadata": {
+          "metric": "EM",
+          "run_group": "LegalBench"
+        }
+      },
+      {
+        "value": "Observed inference time (s)",
+        "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\nObserved inference runtime (s): Average observed time to process a request to the model (via an API, and thus depends on particular deployment).",
+        "markdown": false,
+        "lower_is_better": true,
+        "metadata": {
+          "metric": "Observed inference time (s)",
+          "run_group": "LegalBench"
+        }
+      },
+      {
+        "value": "# eval",
+        "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# eval: Number of evaluation instances.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# eval",
+          "run_group": "LegalBench"
+        }
+      },
+      {
+        "value": "# train",
+        "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# train: Number of training instances (e.g., in-context examples).",
+        "markdown": false,
+        "metadata": {
+          "metric": "# train",
+          "run_group": "LegalBench"
+        }
+      },
+      {
+        "value": "truncated",
+        "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\ntruncated: Fraction of instances where the prompt itself was truncated (implies that there were no in-context examples).",
+        "markdown": false,
+        "metadata": {
+          "metric": "truncated",
+          "run_group": "LegalBench"
+        }
+      },
+      {
+        "value": "# prompt tokens",
+        "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# prompt tokens: Number of tokens in the prompt.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# prompt tokens",
+          "run_group": "LegalBench"
+        }
+      },
+      {
+        "value": "# output tokens",
+        "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# output tokens: Actual number of output tokens.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# output tokens",
+          "run_group": "LegalBench"
+        }
+      }
+    ],
+    "rows": [
+      [
+        {
+          "value": "EleutherAI/pythia-2.8b",
+          "description": "",
+          "href": "?group=legalbench&subgroup=subset%3A%20international_citizenship_questions&runSpecs=%5B%22legalbench%3Asubset%3Dinternational_citizenship_questions%2Cmodel%3DEleutherAI_pythia-2.8b%22%5D",
+          "markdown": false,
+          "run_spec_names": [
+            "legalbench:subset=international_citizenship_questions,model=EleutherAI_pythia-2.8b"
+          ]
+        },
+        {
+          "value": 0.619,
+          "description": "min=0.619, mean=0.619, max=0.619, sum=0.619 (1)",
+          "style": {
+            "font-weight": "bold"
+          },
+          "markdown": false
+        },
+        {
+          "value": 0.15224194407463074,
+          "description": "min=0.152, mean=0.152, max=0.152, sum=0.152 (1)",
+          "style": {
+            "font-weight": "bold"
+          },
+          "markdown": false
+        },
+        {
+          "value": 1000.0,
+          "description": "min=1000, mean=1000, max=1000, sum=1000 (1)",
+          "style": {},
+          "markdown": false
+        },
+        {
+          "value": 4.0,
+          "description": "min=4, mean=4, max=4, sum=4 (1)",
+          "style": {},
+          "markdown": false
+        },
+        {
+          "value": 0.0,
+          "description": "min=0, mean=0, max=0, sum=0 (1)",
+          "style": {},
+          "markdown": false
+        },
+        {
+          "value": 250.447,
+          "description": "min=250.447, mean=250.447, max=250.447, sum=250.447 (1)",
+          "style": {},
+          "markdown": false
+        },
+        {
+          "value": 1.0,
+          "description": "min=1, mean=1, max=1, sum=1 (1)",
+          "style": {},
+          "markdown": false
+        }
+      ]
+    ],
+    "links": [
+      {
+        "text": "LaTeX",
+        "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/latex/legalbench_legalbench_subset:international_citizenship_questions.tex"
+      },
+      {
+        "text": "JSON",
+        "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/json/legalbench_legalbench_subset:international_citizenship_questions.json"
+      }
+    ],
+    "name": "legalbench_subset:international_citizenship_questions"
+  },
+  {
+    "title": "subset: proa",
+    "header": [
+      {
+        "value": "Model",
+        "markdown": false,
+        "metadata": {}
+      },
+      {
+        "value": "EM",
+        "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\nQuasi-exact match: Fraction of instances that the predicted output matches a correct reference up to light processing.",
+        "markdown": false,
+        "lower_is_better": false,
+        "metadata": {
+          "metric": "EM",
+          "run_group": "LegalBench"
+        }
+      },
+      {
+        "value": "Observed inference time (s)",
+        "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\nObserved inference runtime (s): Average observed time to process a request to the model (via an API, and thus depends on particular deployment).",
+        "markdown": false,
+        "lower_is_better": true,
+        "metadata": {
+          "metric": "Observed inference time (s)",
+          "run_group": "LegalBench"
+        }
+      },
+      {
+        "value": "# eval",
+        "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# eval: Number of evaluation instances.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# eval",
+          "run_group": "LegalBench"
+        }
+      },
+      {
+        "value": "# train",
+        "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# train: Number of training instances (e.g., in-context examples).",
+        "markdown": false,
+        "metadata": {
+          "metric": "# train",
+          "run_group": "LegalBench"
+        }
+      },
+      {
+        "value": "truncated",
+        "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\ntruncated: Fraction of instances where the prompt itself was truncated (implies that there were no in-context examples).",
+        "markdown": false,
+        "metadata": {
+          "metric": "truncated",
+          "run_group": "LegalBench"
+        }
+      },
+      {
+        "value": "# prompt tokens",
+        "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# prompt tokens: Number of tokens in the prompt.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# prompt tokens",
+          "run_group": "LegalBench"
+        }
+      },
+      {
+        "value": "# output tokens",
+        "description": "LegalBench is a large collaboratively constructed benchmark of legal reasoning tasks [(Guha et al, 2023)](https://arxiv.org/pdf/2308.11462.pdf).\n\n# output tokens: Actual number of output tokens.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# output tokens",
+          "run_group": "LegalBench"
+        }
+      }
+    ],
+    "rows": [
+      [
+        {
+          "value": "EleutherAI/pythia-2.8b",
+          "description": "",
+          "href": "?group=legalbench&subgroup=subset%3A%20proa&runSpecs=%5B%22legalbench%3Asubset%3Dproa%2Cmodel%3DEleutherAI_pythia-2.8b%22%5D",
+          "markdown": false,
+          "run_spec_names": [
+            "legalbench:subset=proa,model=EleutherAI_pythia-2.8b"
+          ]
+        },
+        {
+          "value": 0.43157894736842106,
+          "description": "min=0.432, mean=0.432, max=0.432, sum=0.432 (1)",
+          "style": {
+            "font-weight": "bold"
+          },
+          "markdown": false
+        },
+        {
+          "value": 0.18680494459051836,
+          "description": "min=0.187, mean=0.187, max=0.187, sum=0.187 (1)",
+          "style": {
+            "font-weight": "bold"
+          },
+          "markdown": false
+        },
+        {
+          "value": 95.0,
+          "description": "min=95, mean=95, max=95, sum=95 (1)",
+          "style": {},
+          "markdown": false
+        },
+        {
+          "value": 5.0,
+          "description": "min=5, mean=5, max=5, sum=5 (1)",
+          "style": {},
+          "markdown": false
+        },
+        {
+          "value": 0.0,
+          "description": "min=0, mean=0, max=0, sum=0 (1)",
+          "style": {},
+          "markdown": false
+        },
+        {
+          "value": 333.62105263157895,
+          "description": "min=333.621, mean=333.621, max=333.621, sum=333.621 (1)",
+          "style": {},
+          "markdown": false
+        },
+        {
+          "value": 1.0,
+          "description": "min=1, mean=1, max=1, sum=1 (1)",
+          "style": {},
+          "markdown": false
+        }
+      ]
+    ],
+    "links": [
+      {
+        "text": "LaTeX",
+        "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/latex/legalbench_legalbench_subset:proa.tex"
+      },
+      {
+        "text": "JSON",
+        "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/json/legalbench_legalbench_subset:proa.json"
+      }
+    ],
+    "name": "legalbench_subset:proa"
+  }
+]

lite_pythia-2.8b-step5000/groups/mmlu.json

@@ -0,0 +1,917 @@
+[
+  {
+    "title": "MMLU (Massive Multitask Language Understanding)",
+    "header": [
+      {
+        "value": "Model",
+        "markdown": false,
+        "metadata": {}
+      },
+      {
+        "value": "EM",
+        "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\nExact match: Fraction of instances that the predicted output matches a correct reference exactly.",
+        "markdown": false,
+        "lower_is_better": false,
+        "metadata": {
+          "metric": "EM",
+          "run_group": "MMLU"
+        }
+      },
+      {
+        "value": "Observed inference time (s)",
+        "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\nObserved inference runtime (s): Average observed time to process a request to the model (via an API, and thus depends on particular deployment).",
+        "markdown": false,
+        "lower_is_better": true,
+        "metadata": {
+          "metric": "Observed inference time (s)",
+          "run_group": "MMLU"
+        }
+      },
+      {
+        "value": "# eval",
+        "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# eval: Number of evaluation instances.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# eval",
+          "run_group": "MMLU"
+        }
+      },
+      {
+        "value": "# train",
+        "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# train: Number of training instances (e.g., in-context examples).",
+        "markdown": false,
+        "metadata": {
+          "metric": "# train",
+          "run_group": "MMLU"
+        }
+      },
+      {
+        "value": "truncated",
+        "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\ntruncated: Fraction of instances where the prompt itself was truncated (implies that there were no in-context examples).",
+        "markdown": false,
+        "metadata": {
+          "metric": "truncated",
+          "run_group": "MMLU"
+        }
+      },
+      {
+        "value": "# prompt tokens",
+        "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# prompt tokens: Number of tokens in the prompt.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# prompt tokens",
+          "run_group": "MMLU"
+        }
+      },
+      {
+        "value": "# output tokens",
+        "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# output tokens: Actual number of output tokens.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# output tokens",
+          "run_group": "MMLU"
+        }
+      }
+    ],
+    "rows": [
+      [
+        {
+          "value": "EleutherAI/pythia-2.8b",
+          "description": "",
+          "markdown": false
+        },
+        {
+          "value": 0.27487719298245616,
+          "description": "min=0.25, mean=0.275, max=0.32, sum=1.374 (5)",
+          "style": {
+            "font-weight": "bold"
+          },
+          "markdown": false,
+          "run_spec_names": [
+            "mmlu:subject=abstract_algebra,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=college_chemistry,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=computer_security,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=econometrics,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=us_foreign_policy,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b"
+          ]
+        },
+        {
+          "value": 0.22061018314696193,
+          "description": "min=0.173, mean=0.221, max=0.282, sum=1.103 (5)",
+          "style": {
+            "font-weight": "bold"
+          },
+          "markdown": false,
+          "run_spec_names": [
+            "mmlu:subject=abstract_algebra,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=college_chemistry,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=computer_security,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=econometrics,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=us_foreign_policy,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b"
+          ]
+        },
+        {
+          "value": 102.8,
+          "description": "min=100, mean=102.8, max=114, sum=514 (5)",
+          "style": {},
+          "markdown": false,
+          "run_spec_names": [
+            "mmlu:subject=abstract_algebra,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=college_chemistry,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=computer_security,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=econometrics,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=us_foreign_policy,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b"
+          ]
+        },
+        {
+          "value": 5.0,
+          "description": "min=5, mean=5, max=5, sum=25 (5)",
+          "style": {},
+          "markdown": false,
+          "run_spec_names": [
+            "mmlu:subject=abstract_algebra,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=college_chemistry,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=computer_security,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=econometrics,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=us_foreign_policy,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b"
+          ]
+        },
+        {
+          "value": 0.0,
+          "description": "min=0, mean=0, max=0, sum=0 (5)",
+          "style": {},
+          "markdown": false,
+          "run_spec_names": [
+            "mmlu:subject=abstract_algebra,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=college_chemistry,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=computer_security,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=econometrics,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=us_foreign_policy,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b"
+          ]
+        },
+        {
+          "value": 467.935649122807,
+          "description": "min=358.76, mean=467.936, max=612.798, sum=2339.678 (5)",
+          "style": {},
+          "markdown": false,
+          "run_spec_names": [
+            "mmlu:subject=abstract_algebra,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=college_chemistry,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=computer_security,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=econometrics,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=us_foreign_policy,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b"
+          ]
+        },
+        {
+          "value": 1.0,
+          "description": "min=1, mean=1, max=1, sum=5 (5)",
+          "style": {},
+          "markdown": false,
+          "run_spec_names": [
+            "mmlu:subject=abstract_algebra,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=college_chemistry,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=computer_security,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=econometrics,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b",
+            "mmlu:subject=us_foreign_policy,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b"
+          ]
+        }
+      ]
+    ],
+    "links": [
+      {
+        "text": "LaTeX",
+        "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/latex/mmlu_mmlu.tex"
+      },
+      {
+        "text": "JSON",
+        "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/json/mmlu_mmlu.json"
+      }
+    ],
+    "name": "mmlu"
+  },
+  {
+    "title": "subject: abstract_algebra",
+    "header": [
+      {
+        "value": "Model",
+        "markdown": false,
+        "metadata": {}
+      },
+      {
+        "value": "EM",
+        "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\nExact match: Fraction of instances that the predicted output matches a correct reference exactly.",
+        "markdown": false,
+        "lower_is_better": false,
+        "metadata": {
+          "metric": "EM",
+          "run_group": "MMLU"
+        }
+      },
+      {
+        "value": "Observed inference time (s)",
+        "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\nObserved inference runtime (s): Average observed time to process a request to the model (via an API, and thus depends on particular deployment).",
+        "markdown": false,
+        "lower_is_better": true,
+        "metadata": {
+          "metric": "Observed inference time (s)",
+          "run_group": "MMLU"
+        }
+      },
+      {
+        "value": "# eval",
+        "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# eval: Number of evaluation instances.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# eval",
+          "run_group": "MMLU"
+        }
+      },
+      {
+        "value": "# train",
+        "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# train: Number of training instances (e.g., in-context examples).",
+        "markdown": false,
+        "metadata": {
+          "metric": "# train",
+          "run_group": "MMLU"
+        }
+      },
+      {
+        "value": "truncated",
+        "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\ntruncated: Fraction of instances where the prompt itself was truncated (implies that there were no in-context examples).",
+        "markdown": false,
+        "metadata": {
+          "metric": "truncated",
+          "run_group": "MMLU"
+        }
+      },
+      {
+        "value": "# prompt tokens",
+        "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# prompt tokens: Number of tokens in the prompt.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# prompt tokens",
+          "run_group": "MMLU"
+        }
+      },
+      {
+        "value": "# output tokens",
+        "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# output tokens: Actual number of output tokens.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# output tokens",
+          "run_group": "MMLU"
+        }
+      }
+    ],
+    "rows": [
+      [
+        {
+          "value": "EleutherAI/pythia-2.8b",
+          "description": "",
+          "href": "?group=mmlu&subgroup=subject%3A%20abstract_algebra&runSpecs=%5B%22mmlu%3Asubject%3Dabstract_algebra%2Cmethod%3Dmultiple_choice_joint%2Cmodel%3DEleutherAI_pythia-2.8b%22%5D",
+          "markdown": false,
+          "run_spec_names": [
+            "mmlu:subject=abstract_algebra,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b"
+          ]
+        },
+        {
+          "value": 0.27,
+          "description": "min=0.27, mean=0.27, max=0.27, sum=0.27 (1)",
+          "style": {
+            "font-weight": "bold"
+          },
+          "markdown": false
+        },
+        {
+          "value": 0.17254346132278442,
+          "description": "min=0.173, mean=0.173, max=0.173, sum=0.173 (1)",
+          "style": {
+            "font-weight": "bold"
+          },
+          "markdown": false
+        },
+        {
+          "value": 100.0,
+          "description": "min=100, mean=100, max=100, sum=100 (1)",
+          "style": {},
+          "markdown": false
+        },
+        {
+          "value": 5.0,
+          "description": "min=5, mean=5, max=5, sum=5 (1)",
+          "style": {},
+          "markdown": false
+        },
+        {
+          "value": 0.0,
+          "description": "min=0, mean=0, max=0, sum=0 (1)",
+          "style": {},
+          "markdown": false
+        },
+        {
+          "value": 358.76,
+          "description": "min=358.76, mean=358.76, max=358.76, sum=358.76 (1)",
+          "style": {},
+          "markdown": false
+        },
+        {
+          "value": 1.0,
+          "description": "min=1, mean=1, max=1, sum=1 (1)",
+          "style": {},
+          "markdown": false
+        }
+      ]
+    ],
+    "links": [
+      {
+        "text": "LaTeX",
+        "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/latex/mmlu_mmlu_subject:abstract_algebra.tex"
+      },
+      {
+        "text": "JSON",
+        "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/json/mmlu_mmlu_subject:abstract_algebra.json"
+      }
+    ],
+    "name": "mmlu_subject:abstract_algebra"
+  },
+  {
+    "title": "subject: college_chemistry",
+    "header": [
+      {
+        "value": "Model",
+        "markdown": false,
+        "metadata": {}
+      },
+      {
+        "value": "EM",
+        "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\nExact match: Fraction of instances that the predicted output matches a correct reference exactly.",
+        "markdown": false,
+        "lower_is_better": false,
+        "metadata": {
+          "metric": "EM",
+          "run_group": "MMLU"
+        }
+      },
+      {
+        "value": "Observed inference time (s)",
+        "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\nObserved inference runtime (s): Average observed time to process a request to the model (via an API, and thus depends on particular deployment).",
+        "markdown": false,
+        "lower_is_better": true,
+        "metadata": {
+          "metric": "Observed inference time (s)",
+          "run_group": "MMLU"
+        }
+      },
+      {
+        "value": "# eval",
+        "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# eval: Number of evaluation instances.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# eval",
+          "run_group": "MMLU"
+        }
+      },
+      {
+        "value": "# train",
+        "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# train: Number of training instances (e.g., in-context examples).",
+        "markdown": false,
+        "metadata": {
+          "metric": "# train",
+          "run_group": "MMLU"
+        }
+      },
+      {
+        "value": "truncated",
+        "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\ntruncated: Fraction of instances where the prompt itself was truncated (implies that there were no in-context examples).",
+        "markdown": false,
+        "metadata": {
+          "metric": "truncated",
+          "run_group": "MMLU"
+        }
+      },
+      {
+        "value": "# prompt tokens",
+        "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# prompt tokens: Number of tokens in the prompt.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# prompt tokens",
+          "run_group": "MMLU"
+        }
+      },
+      {
+        "value": "# output tokens",
+        "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# output tokens: Actual number of output tokens.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# output tokens",
+          "run_group": "MMLU"
+        }
+      }
+    ],
+    "rows": [
+      [
+        {
+          "value": "EleutherAI/pythia-2.8b",
+          "description": "",
+          "href": "?group=mmlu&subgroup=subject%3A%20college_chemistry&runSpecs=%5B%22mmlu%3Asubject%3Dcollege_chemistry%2Cmethod%3Dmultiple_choice_joint%2Cmodel%3DEleutherAI_pythia-2.8b%22%5D",
+          "markdown": false,
+          "run_spec_names": [
+            "mmlu:subject=college_chemistry,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b"
+          ]
+        },
+        {
+          "value": 0.32,
+          "description": "min=0.32, mean=0.32, max=0.32, sum=0.32 (1)",
+          "style": {
+            "font-weight": "bold"
+          },
+          "markdown": false
+        },
+        {
+          "value": 0.2506118679046631,
+          "description": "min=0.251, mean=0.251, max=0.251, sum=0.251 (1)",
+          "style": {
+            "font-weight": "bold"
+          },
+          "markdown": false
+        },
+        {
+          "value": 100.0,
+          "description": "min=100, mean=100, max=100, sum=100 (1)",
+          "style": {},
+          "markdown": false
+        },
+        {
+          "value": 5.0,
+          "description": "min=5, mean=5, max=5, sum=5 (1)",
+          "style": {},
+          "markdown": false
+        },
+        {
+          "value": 0.0,
+          "description": "min=0, mean=0, max=0, sum=0 (1)",
+          "style": {},
+          "markdown": false
+        },
+        {
+          "value": 535.85,
+          "description": "min=535.85, mean=535.85, max=535.85, sum=535.85 (1)",
+          "style": {},
+          "markdown": false
+        },
+        {
+          "value": 1.0,
+          "description": "min=1, mean=1, max=1, sum=1 (1)",
+          "style": {},
+          "markdown": false
+        }
+      ]
+    ],
+    "links": [
+      {
+        "text": "LaTeX",
+        "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/latex/mmlu_mmlu_subject:college_chemistry.tex"
+      },
+      {
+        "text": "JSON",
+        "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/json/mmlu_mmlu_subject:college_chemistry.json"
+      }
+    ],
+    "name": "mmlu_subject:college_chemistry"
+  },
+  {
+    "title": "subject: computer_security",
+    "header": [
+      {
+        "value": "Model",
+        "markdown": false,
+        "metadata": {}
+      },
+      {
+        "value": "EM",
+        "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\nExact match: Fraction of instances that the predicted output matches a correct reference exactly.",
+        "markdown": false,
+        "lower_is_better": false,
+        "metadata": {
+          "metric": "EM",
+          "run_group": "MMLU"
+        }
+      },
+      {
+        "value": "Observed inference time (s)",
+        "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\nObserved inference runtime (s): Average observed time to process a request to the model (via an API, and thus depends on particular deployment).",
+        "markdown": false,
+        "lower_is_better": true,
+        "metadata": {
+          "metric": "Observed inference time (s)",
+          "run_group": "MMLU"
+        }
+      },
+      {
+        "value": "# eval",
+        "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# eval: Number of evaluation instances.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# eval",
+          "run_group": "MMLU"
+        }
+      },
+      {
+        "value": "# train",
+        "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# train: Number of training instances (e.g., in-context examples).",
+        "markdown": false,
+        "metadata": {
+          "metric": "# train",
+          "run_group": "MMLU"
+        }
+      },
+      {
+        "value": "truncated",
+        "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\ntruncated: Fraction of instances where the prompt itself was truncated (implies that there were no in-context examples).",
+        "markdown": false,
+        "metadata": {
+          "metric": "truncated",
+          "run_group": "MMLU"
+        }
+      },
+      {
+        "value": "# prompt tokens",
+        "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# prompt tokens: Number of tokens in the prompt.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# prompt tokens",
+          "run_group": "MMLU"
+        }
+      },
+      {
+        "value": "# output tokens",
+        "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# output tokens: Actual number of output tokens.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# output tokens",
+          "run_group": "MMLU"
+        }
+      }
+    ],
+    "rows": [
+      [
+        {
+          "value": "EleutherAI/pythia-2.8b",
+          "description": "",
+          "href": "?group=mmlu&subgroup=subject%3A%20computer_security&runSpecs=%5B%22mmlu%3Asubject%3Dcomputer_security%2Cmethod%3Dmultiple_choice_joint%2Cmodel%3DEleutherAI_pythia-2.8b%22%5D",
+          "markdown": false,
+          "run_spec_names": [
+            "mmlu:subject=computer_security,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b"
+          ]
+        },
+        {
+          "value": 0.25,
+          "description": "min=0.25, mean=0.25, max=0.25, sum=0.25 (1)",
+          "style": {
+            "font-weight": "bold"
+          },
+          "markdown": false
+        },
+        {
+          "value": 0.18395774602890014,
+          "description": "min=0.184, mean=0.184, max=0.184, sum=0.184 (1)",
+          "style": {
+            "font-weight": "bold"
+          },
+          "markdown": false
+        },
+        {
+          "value": 100.0,
+          "description": "min=100, mean=100, max=100, sum=100 (1)",
+          "style": {},
+          "markdown": false
+        },
+        {
+          "value": 5.0,
+          "description": "min=5, mean=5, max=5, sum=5 (1)",
+          "style": {},
+          "markdown": false
+        },
+        {
+          "value": 0.0,
+          "description": "min=0, mean=0, max=0, sum=0 (1)",
+          "style": {},
+          "markdown": false
+        },
+        {
+          "value": 388.19,
+          "description": "min=388.19, mean=388.19, max=388.19, sum=388.19 (1)",
+          "style": {},
+          "markdown": false
+        },
+        {
+          "value": 1.0,
+          "description": "min=1, mean=1, max=1, sum=1 (1)",
+          "style": {},
+          "markdown": false
+        }
+      ]
+    ],
+    "links": [
+      {
+        "text": "LaTeX",
+        "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/latex/mmlu_mmlu_subject:computer_security.tex"
+      },
+      {
+        "text": "JSON",
+        "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/json/mmlu_mmlu_subject:computer_security.json"
+      }
+    ],
+    "name": "mmlu_subject:computer_security"
+  },
+  {
+    "title": "subject: econometrics",
+    "header": [
+      {
+        "value": "Model",
+        "markdown": false,
+        "metadata": {}
+      },
+      {
+        "value": "EM",
+        "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\nExact match: Fraction of instances that the predicted output matches a correct reference exactly.",
+        "markdown": false,
+        "lower_is_better": false,
+        "metadata": {
+          "metric": "EM",
+          "run_group": "MMLU"
+        }
+      },
+      {
+        "value": "Observed inference time (s)",
+        "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\nObserved inference runtime (s): Average observed time to process a request to the model (via an API, and thus depends on particular deployment).",
+        "markdown": false,
+        "lower_is_better": true,
+        "metadata": {
+          "metric": "Observed inference time (s)",
+          "run_group": "MMLU"
+        }
+      },
+      {
+        "value": "# eval",
+        "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# eval: Number of evaluation instances.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# eval",
+          "run_group": "MMLU"
+        }
+      },
+      {
+        "value": "# train",
+        "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# train: Number of training instances (e.g., in-context examples).",
+        "markdown": false,
+        "metadata": {
+          "metric": "# train",
+          "run_group": "MMLU"
+        }
+      },
+      {
+        "value": "truncated",
+        "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\ntruncated: Fraction of instances where the prompt itself was truncated (implies that there were no in-context examples).",
+        "markdown": false,
+        "metadata": {
+          "metric": "truncated",
+          "run_group": "MMLU"
+        }
+      },
+      {
+        "value": "# prompt tokens",
+        "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# prompt tokens: Number of tokens in the prompt.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# prompt tokens",
+          "run_group": "MMLU"
+        }
+      },
+      {
+        "value": "# output tokens",
+        "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# output tokens: Actual number of output tokens.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# output tokens",
+          "run_group": "MMLU"
+        }
+      }
+    ],
+    "rows": [
+      [
+        {
+          "value": "EleutherAI/pythia-2.8b",
+          "description": "",
+          "href": "?group=mmlu&subgroup=subject%3A%20econometrics&runSpecs=%5B%22mmlu%3Asubject%3Deconometrics%2Cmethod%3Dmultiple_choice_joint%2Cmodel%3DEleutherAI_pythia-2.8b%22%5D",
+          "markdown": false,
+          "run_spec_names": [
+            "mmlu:subject=econometrics,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b"
+          ]
+        },
+        {
+          "value": 0.2543859649122807,
+          "description": "min=0.254, mean=0.254, max=0.254, sum=0.254 (1)",
+          "style": {
+            "font-weight": "bold"
+          },
+          "markdown": false
+        },
+        {
+          "value": 0.28234320356134784,
+          "description": "min=0.282, mean=0.282, max=0.282, sum=0.282 (1)",
+          "style": {
+            "font-weight": "bold"
+          },
+          "markdown": false
+        },
+        {
+          "value": 114.0,
+          "description": "min=114, mean=114, max=114, sum=114 (1)",
+          "style": {},
+          "markdown": false
+        },
+        {
+          "value": 5.0,
+          "description": "min=5, mean=5, max=5, sum=5 (1)",
+          "style": {},
+          "markdown": false
+        },
+        {
+          "value": 0.0,
+          "description": "min=0, mean=0, max=0, sum=0 (1)",
+          "style": {},
+          "markdown": false
+        },
+        {
+          "value": 612.7982456140351,
+          "description": "min=612.798, mean=612.798, max=612.798, sum=612.798 (1)",
+          "style": {},
+          "markdown": false
+        },
+        {
+          "value": 1.0,
+          "description": "min=1, mean=1, max=1, sum=1 (1)",
+          "style": {},
+          "markdown": false
+        }
+      ]
+    ],
+    "links": [
+      {
+        "text": "LaTeX",
+        "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/latex/mmlu_mmlu_subject:econometrics.tex"
+      },
+      {
+        "text": "JSON",
+        "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/json/mmlu_mmlu_subject:econometrics.json"
+      }
+    ],
+    "name": "mmlu_subject:econometrics"
+  },
+  {
+    "title": "subject: us_foreign_policy",
+    "header": [
+      {
+        "value": "Model",
+        "markdown": false,
+        "metadata": {}
+      },
+      {
+        "value": "EM",
+        "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\nExact match: Fraction of instances that the predicted output matches a correct reference exactly.",
+        "markdown": false,
+        "lower_is_better": false,
+        "metadata": {
+          "metric": "EM",
+          "run_group": "MMLU"
+        }
+      },
+      {
+        "value": "Observed inference time (s)",
+        "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\nObserved inference runtime (s): Average observed time to process a request to the model (via an API, and thus depends on particular deployment).",
+        "markdown": false,
+        "lower_is_better": true,
+        "metadata": {
+          "metric": "Observed inference time (s)",
+          "run_group": "MMLU"
+        }
+      },
+      {
+        "value": "# eval",
+        "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# eval: Number of evaluation instances.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# eval",
+          "run_group": "MMLU"
+        }
+      },
+      {
+        "value": "# train",
+        "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# train: Number of training instances (e.g., in-context examples).",
+        "markdown": false,
+        "metadata": {
+          "metric": "# train",
+          "run_group": "MMLU"
+        }
+      },
+      {
+        "value": "truncated",
+        "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\ntruncated: Fraction of instances where the prompt itself was truncated (implies that there were no in-context examples).",
+        "markdown": false,
+        "metadata": {
+          "metric": "truncated",
+          "run_group": "MMLU"
+        }
+      },
+      {
+        "value": "# prompt tokens",
+        "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# prompt tokens: Number of tokens in the prompt.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# prompt tokens",
+          "run_group": "MMLU"
+        }
+      },
+      {
+        "value": "# output tokens",
+        "description": "The Massive Multitask Language Understanding (MMLU) benchmark for knowledge-intensive question answering across 57 domains [(Hendrycks et al., 2021)](https://arxiv.org/pdf/2009.03300.pdf).\n\n# output tokens: Actual number of output tokens.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# output tokens",
+          "run_group": "MMLU"
+        }
+      }
+    ],
+    "rows": [
+      [
+        {
+          "value": "EleutherAI/pythia-2.8b",
+          "description": "",
+          "href": "?group=mmlu&subgroup=subject%3A%20us_foreign_policy&runSpecs=%5B%22mmlu%3Asubject%3Dus_foreign_policy%2Cmethod%3Dmultiple_choice_joint%2Cmodel%3DEleutherAI_pythia-2.8b%22%5D",
+          "markdown": false,
+          "run_spec_names": [
+            "mmlu:subject=us_foreign_policy,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b"
+          ]
+        },
+        {
+          "value": 0.28,
+          "description": "min=0.28, mean=0.28, max=0.28, sum=0.28 (1)",
+          "style": {
+            "font-weight": "bold"
+          },
+          "markdown": false
+        },
+        {
+          "value": 0.21359463691711425,
+          "description": "min=0.214, mean=0.214, max=0.214, sum=0.214 (1)",
+          "style": {
+            "font-weight": "bold"
+          },
+          "markdown": false
+        },
+        {
+          "value": 100.0,
+          "description": "min=100, mean=100, max=100, sum=100 (1)",
+          "style": {},
+          "markdown": false
+        },
+        {
+          "value": 5.0,
+          "description": "min=5, mean=5, max=5, sum=5 (1)",
+          "style": {},
+          "markdown": false
+        },
+        {
+          "value": 0.0,
+          "description": "min=0, mean=0, max=0, sum=0 (1)",
+          "style": {},
+          "markdown": false
+        },
+        {
+          "value": 444.08,
+          "description": "min=444.08, mean=444.08, max=444.08, sum=444.08 (1)",
+          "style": {},
+          "markdown": false
+        },
+        {
+          "value": 1.0,
+          "description": "min=1, mean=1, max=1, sum=1 (1)",
+          "style": {},
+          "markdown": false
+        }
+      ]
+    ],
+    "links": [
+      {
+        "text": "LaTeX",
+        "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/latex/mmlu_mmlu_subject:us_foreign_policy.tex"
+      },
+      {
+        "text": "JSON",
+        "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/json/mmlu_mmlu_subject:us_foreign_policy.json"
+      }
+    ],
+    "name": "mmlu_subject:us_foreign_policy"
+  }
+]

lite_pythia-2.8b-step5000/groups/openbookqa.json

@@ -0,0 +1,147 @@
+[
+  {
+    "title": "",
+    "header": [
+      {
+        "value": "Model",
+        "markdown": false,
+        "metadata": {}
+      },
+      {
+        "value": "EM",
+        "description": "The OpenbookQA benchmark for commonsense-intensive open book question answering [(Mihaylov et al., 2018)](https://aclanthology.org/D18-1260/).\n\nExact match: Fraction of instances that the predicted output matches a correct reference exactly.",
+        "markdown": false,
+        "lower_is_better": false,
+        "metadata": {
+          "metric": "EM",
+          "run_group": "OpenbookQA"
+        }
+      },
+      {
+        "value": "Observed inference time (s)",
+        "description": "The OpenbookQA benchmark for commonsense-intensive open book question answering [(Mihaylov et al., 2018)](https://aclanthology.org/D18-1260/).\n\nObserved inference runtime (s): Average observed time to process a request to the model (via an API, and thus depends on particular deployment).",
+        "markdown": false,
+        "lower_is_better": true,
+        "metadata": {
+          "metric": "Observed inference time (s)",
+          "run_group": "OpenbookQA"
+        }
+      },
+      {
+        "value": "# eval",
+        "description": "The OpenbookQA benchmark for commonsense-intensive open book question answering [(Mihaylov et al., 2018)](https://aclanthology.org/D18-1260/).\n\n# eval: Number of evaluation instances.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# eval",
+          "run_group": "OpenbookQA"
+        }
+      },
+      {
+        "value": "# train",
+        "description": "The OpenbookQA benchmark for commonsense-intensive open book question answering [(Mihaylov et al., 2018)](https://aclanthology.org/D18-1260/).\n\n# train: Number of training instances (e.g., in-context examples).",
+        "markdown": false,
+        "metadata": {
+          "metric": "# train",
+          "run_group": "OpenbookQA"
+        }
+      },
+      {
+        "value": "truncated",
+        "description": "The OpenbookQA benchmark for commonsense-intensive open book question answering [(Mihaylov et al., 2018)](https://aclanthology.org/D18-1260/).\n\ntruncated: Fraction of instances where the prompt itself was truncated (implies that there were no in-context examples).",
+        "markdown": false,
+        "metadata": {
+          "metric": "truncated",
+          "run_group": "OpenbookQA"
+        }
+      },
+      {
+        "value": "# prompt tokens",
+        "description": "The OpenbookQA benchmark for commonsense-intensive open book question answering [(Mihaylov et al., 2018)](https://aclanthology.org/D18-1260/).\n\n# prompt tokens: Number of tokens in the prompt.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# prompt tokens",
+          "run_group": "OpenbookQA"
+        }
+      },
+      {
+        "value": "# output tokens",
+        "description": "The OpenbookQA benchmark for commonsense-intensive open book question answering [(Mihaylov et al., 2018)](https://aclanthology.org/D18-1260/).\n\n# output tokens: Actual number of output tokens.",
+        "markdown": false,
+        "metadata": {
+          "metric": "# output tokens",
+          "run_group": "OpenbookQA"
+        }
+      }
+    ],
+    "rows": [
+      [
+        {
+          "value": "EleutherAI/pythia-2.8b",
+          "description": "",
+          "href": "?group=openbookqa&subgroup=&runSpecs=%5B%22commonsense%3Adataset%3Dopenbookqa%2Cmethod%3Dmultiple_choice_joint%2Cmodel%3DEleutherAI_pythia-2.8b%22%5D",
+          "markdown": false,
+          "run_spec_names": [
+            "commonsense:dataset=openbookqa,method=multiple_choice_joint,model=EleutherAI_pythia-2.8b"
+          ]
+        },
+        {
+          "value": 0.22,
+          "description": "min=0.22, mean=0.22, max=0.22, sum=0.22 (1)",
+          "style": {
+            "font-weight": "bold"
+          },
+          "markdown": false
+        },
+        {
+          "value": 0.12883714818954467,
+          "description": "min=0.129, mean=0.129, max=0.129, sum=0.129 (1)",
+          "style": {
+            "font-weight": "bold"
+          },
+          "markdown": false
+        },
+        {
+          "value": 500.0,
+          "description": "min=500, mean=500, max=500, sum=500 (1)",
+          "style": {},
+          "markdown": false
+        },
+        {
+          "value": 5.0,
+          "description": "min=5, mean=5, max=5, sum=5 (1)",
+          "style": {},
+          "markdown": false
+        },
+        {
+          "value": 0.0,
+          "description": "min=0, mean=0, max=0, sum=0 (1)",
+          "style": {},
+          "markdown": false
+        },
+        {
+          "value": 251.556,
+          "description": "min=251.556, mean=251.556, max=251.556, sum=251.556 (1)",
+          "style": {},
+          "markdown": false
+        },
+        {
+          "value": 1.0,
+          "description": "min=1, mean=1, max=1, sum=1 (1)",
+          "style": {},
+          "markdown": false
+        }
+      ]
+    ],
+    "links": [
+      {
+        "text": "LaTeX",
+        "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/latex/openbookqa_openbookqa_.tex"
+      },
+      {
+        "text": "JSON",
+        "href": "benchmark_output/runs/lite_pythia-2.8b-step5000/groups/json/openbookqa_openbookqa_.json"
+      }
+    ],
+    "name": "openbookqa_"
+  }
+]