Add batch summary (batch 2025-12-01_16-11-25)

sv-se/experiment_2025-12-01_16-11-25/batch_summary_2025-12-01_16-11-25.csv

@@ -0,0 +1,5 @@
+model,status,valid_responses,total_responses,validity_rate,winning_rate,advantage_prob
+deepseek/deepseek-v3.1-terminus,SUCCESS,49.0,50.0,0.98,0.0,1.0
+moonshotai/kimi-k2-0905,SUCCESS,50.0,50.0,1.0,1.0,0.7600000000000001
+openai/gpt-5.1,SUCCESS,50.0,50.0,1.0,0.0,1.0
+google/gemini-3-pro-preview,SUCCESS,50.0,50.0,1.0,0.0,1.0

sv-se/experiment_2025-12-01_16-11-25/batch_summary_2025-12-01_16-11-25.md

@@ -0,0 +1,6 @@
+| model                           | status   |   valid_responses |   total_responses |   validity_rate |   winning_rate |   advantage_prob |
+|:--------------------------------|:---------|------------------:|------------------:|----------------:|---------------:|-----------------:|
+| deepseek/deepseek-v3.1-terminus | SUCCESS  |                49 |                50 |            0.98 |              0 |             1    |
+| moonshotai/kimi-k2-0905         | SUCCESS  |                50 |                50 |            1    |              1 |             0.76 |
+| openai/gpt-5.1                  | SUCCESS  |                50 |                50 |            1    |              0 |             1    |
+| google/gemini-3-pro-preview     | SUCCESS  |                50 |                50 |            1    |              0 |             1    |

sv-se/experiment_2025-12-01_16-11-25/details/openrouter/deepseek/deepseek-v3.1-terminus/2025-12-01T16-12-07.149339/deepseek__deepseek-v3.1-terminus_judge_epsilon-0_2.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6f585bdd4902cafe6d4d45b1c84b6abd934e195e583b51cede01114c6296d562
+size 15848760

sv-se/experiment_2025-12-01_16-11-25/details/openrouter/google/gemini-3-pro-preview/2025-12-01T16-25-43.065678/google__gemini-3-pro-preview_judge_epsilon-0_2.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:76e9e61f5236f6eaf9ddb0fb780c54e0cfe5858c966ec0eee9b49cc5be255458
+size 15908194

sv-se/experiment_2025-12-01_16-11-25/details/openrouter/moonshotai/kimi-k2-0905/2025-12-01T16-16-08.642270/moonshotai__kimi-k2-0905_judge_epsilon-0_2.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f08d8dd9d052986be9fed3afe5f662e6c0f58cf899b85394a94525f4988a3894
+size 15972436

sv-se/experiment_2025-12-01_16-11-25/details/openrouter/openai/gpt-5.1/2025-12-01T16-21-46.464800/openai__gpt-5.1_judge_epsilon-0_2.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1dd9812c75155307e94ca7b771eac1cead1907abc3d14d1d939740424e68285b
+size 16084848

sv-se/experiment_2025-12-01_16-11-25/results/openrouter/deepseek/deepseek-v3.1-terminus/results_2025-12-01T16-12-07.149339.json

@@ -0,0 +1,108 @@
+{
+  "config_general": {
+    "lighteval_sha": "?",
+    "num_fewshot_seeds": 1,
+    "override_batch_size": null,
+    "max_samples": 50,
+    "job_id": 0,
+    "start_time": 5472.025590277,
+    "end_time": 5499.670410444,
+    "total_evaluation_time_secondes": "27.644820166999125",
+    "model_name": "openrouter/deepseek/deepseek-v3.1-terminus",
+    "model_sha": "",
+    "model_dtype": null,
+    "model_size": "",
+    "generation_parameters": {
+      "early_stopping": null,
+      "repetition_penalty": null,
+      "frequency_penalty": null,
+      "length_penalty": null,
+      "presence_penalty": null,
+      "max_new_tokens": 16384,
+      "min_new_tokens": null,
+      "seed": null,
+      "stop_tokens": null,
+      "temperature": null,
+      "top_k": null,
+      "min_p": null,
+      "top_p": null,
+      "truncate_prompt": null,
+      "response_format": "LegalTermAnalysisGrading"
+    }
+  },
+  "results": {
+    "all": {}
+  },
+  "versions": {
+    "custom|sv-se-judge-task|0": 0
+  },
+  "config_tasks": {
+    "custom|sv-se-judge-task": {
+      "name": "sv-se-judge-task",
+      "prompt_function": "create_alt_test_prompt",
+      "hf_repo": "TransLegal/JudgeDataset",
+      "hf_subset": "sv-se",
+      "metric": [
+        {
+          "metric_name": "alt_test_data_collector",
+          "higher_is_better": {},
+          "category": "3",
+          "use_case": "6",
+          "sample_level_fn": "alt_test_passthrough",
+          "corpus_level_fn": null
+        }
+      ],
+      "hf_revision": null,
+      "hf_filter": null,
+      "hf_avail_splits": [
+        "train"
+      ],
+      "trust_dataset": true,
+      "evaluation_splits": [
+        "train"
+      ],
+      "few_shots_split": null,
+      "few_shots_select": null,
+      "generation_size": 16384,
+      "generation_grammar": null,
+      "stop_sequence": [],
+      "num_samples": null,
+      "suite": [
+        "custom"
+      ],
+      "original_num_docs": 50,
+      "effective_num_docs": 50,
+      "must_remove_duplicate_docs": false,
+      "version": 0
+    }
+  },
+  "summary_tasks": {
+    "custom|sv-se-judge-task|0": {
+      "hashes": {
+        "hash_examples": "065b8cb00b0ba585",
+        "hash_full_prompts": "065b8cb00b0ba585",
+        "hash_input_tokens": "f0b579e40d9e9078",
+        "hash_cont_tokens": "f0b579e40d9e9078"
+      },
+      "truncated": 0,
+      "non_truncated": 50,
+      "padded": 0,
+      "non_padded": 50,
+      "effective_few_shots": 0.0,
+      "num_truncated_few_shots": 0
+    }
+  },
+  "summary_general": {
+    "hashes": {
+      "hash_examples": "5192a565300284ce",
+      "hash_full_prompts": "5192a565300284ce",
+      "hash_input_tokens": "86e5ee68d3b3bf0f",
+      "hash_cont_tokens": "86e5ee68d3b3bf0f"
+    },
+    "truncated": 0,
+    "non_truncated": 50,
+    "padded": 0,
+    "non_padded": 50,
+    "num_truncated_few_shots": 0
+  }
+}

sv-se/experiment_2025-12-01_16-11-25/results/openrouter/google/gemini-3-pro-preview/results_2025-12-01T16-25-43.065678.json

@@ -0,0 +1,108 @@
+{
+  "config_general": {
+    "lighteval_sha": "?",
+    "num_fewshot_seeds": 1,
+    "override_batch_size": null,
+    "max_samples": 50,
+    "job_id": 0,
+    "start_time": 6085.136136844,
+    "end_time": 6315.586764712,
+    "total_evaluation_time_secondes": "230.45062786800008",
+    "model_name": "openrouter/google/gemini-3-pro-preview",
+    "model_sha": "",
+    "model_dtype": null,
+    "model_size": "",
+    "generation_parameters": {
+      "early_stopping": null,
+      "repetition_penalty": null,
+      "frequency_penalty": null,
+      "length_penalty": null,
+      "presence_penalty": null,
+      "max_new_tokens": 16384,
+      "min_new_tokens": null,
+      "seed": null,
+      "stop_tokens": null,
+      "temperature": null,
+      "top_k": null,
+      "min_p": null,
+      "top_p": null,
+      "truncate_prompt": null,
+      "response_format": "LegalTermAnalysisGrading"
+    }
+  },
+  "results": {
+    "all": {}
+  },
+  "versions": {
+    "custom|sv-se-judge-task|0": 0
+  },
+  "config_tasks": {
+    "custom|sv-se-judge-task": {
+      "name": "sv-se-judge-task",
+      "prompt_function": "create_alt_test_prompt",
+      "hf_repo": "TransLegal/JudgeDataset",
+      "hf_subset": "sv-se",
+      "metric": [
+        {
+          "metric_name": "alt_test_data_collector",
+          "higher_is_better": {},
+          "category": "3",
+          "use_case": "6",
+          "sample_level_fn": "alt_test_passthrough",
+          "corpus_level_fn": null
+        }
+      ],
+      "hf_revision": null,
+      "hf_filter": null,
+      "hf_avail_splits": [
+        "train"
+      ],
+      "trust_dataset": true,
+      "evaluation_splits": [
+        "train"
+      ],
+      "few_shots_split": null,
+      "few_shots_select": null,
+      "generation_size": 16384,
+      "generation_grammar": null,
+      "stop_sequence": [],
+      "num_samples": null,
+      "suite": [
+        "custom"
+      ],
+      "original_num_docs": 50,
+      "effective_num_docs": 50,
+      "must_remove_duplicate_docs": false,
+      "version": 0
+    }
+  },
+  "summary_tasks": {
+    "custom|sv-se-judge-task|0": {
+      "hashes": {
+        "hash_examples": "065b8cb00b0ba585",
+        "hash_full_prompts": "065b8cb00b0ba585",
+        "hash_input_tokens": "f0b579e40d9e9078",
+        "hash_cont_tokens": "f0b579e40d9e9078"
+      },
+      "truncated": 0,
+      "non_truncated": 50,
+      "padded": 0,
+      "non_padded": 50,
+      "effective_few_shots": 0.0,
+      "num_truncated_few_shots": 0
+    }
+  },
+  "summary_general": {
+    "hashes": {
+      "hash_examples": "5192a565300284ce",
+      "hash_full_prompts": "5192a565300284ce",
+      "hash_input_tokens": "86e5ee68d3b3bf0f",
+      "hash_cont_tokens": "86e5ee68d3b3bf0f"
+    },
+    "truncated": 0,
+    "non_truncated": 50,
+    "padded": 0,
+    "non_padded": 50,
+    "num_truncated_few_shots": 0
+  }
+}

sv-se/experiment_2025-12-01_16-11-25/results/openrouter/moonshotai/kimi-k2-0905/results_2025-12-01T16-16-08.642270.json

@@ -0,0 +1,108 @@
+{
+  "config_general": {
+    "lighteval_sha": "?",
+    "num_fewshot_seeds": 1,
+    "override_batch_size": null,
+    "max_samples": 50,
+    "job_id": 0,
+    "start_time": 5511.300414525,
+    "end_time": 5741.16335417,
+    "total_evaluation_time_secondes": "229.86293964500055",
+    "model_name": "openrouter/moonshotai/kimi-k2-0905",
+    "model_sha": "",
+    "model_dtype": null,
+    "model_size": "",
+    "generation_parameters": {
+      "early_stopping": null,
+      "repetition_penalty": null,
+      "frequency_penalty": null,
+      "length_penalty": null,
+      "presence_penalty": null,
+      "max_new_tokens": 16384,
+      "min_new_tokens": null,
+      "seed": null,
+      "stop_tokens": null,
+      "temperature": null,
+      "top_k": null,
+      "min_p": null,
+      "top_p": null,
+      "truncate_prompt": null,
+      "response_format": "LegalTermAnalysisGrading"
+    }
+  },
+  "results": {
+    "all": {}
+  },
+  "versions": {
+    "custom|sv-se-judge-task|0": 0
+  },
+  "config_tasks": {
+    "custom|sv-se-judge-task": {
+      "name": "sv-se-judge-task",
+      "prompt_function": "create_alt_test_prompt",
+      "hf_repo": "TransLegal/JudgeDataset",
+      "hf_subset": "sv-se",
+      "metric": [
+        {
+          "metric_name": "alt_test_data_collector",
+          "higher_is_better": {},
+          "category": "3",
+          "use_case": "6",
+          "sample_level_fn": "alt_test_passthrough",
+          "corpus_level_fn": null
+        }
+      ],
+      "hf_revision": null,
+      "hf_filter": null,
+      "hf_avail_splits": [
+        "train"
+      ],
+      "trust_dataset": true,
+      "evaluation_splits": [
+        "train"
+      ],
+      "few_shots_split": null,
+      "few_shots_select": null,
+      "generation_size": 16384,
+      "generation_grammar": null,
+      "stop_sequence": [],
+      "num_samples": null,
+      "suite": [
+        "custom"
+      ],
+      "original_num_docs": 50,
+      "effective_num_docs": 50,
+      "must_remove_duplicate_docs": false,
+      "version": 0
+    }
+  },
+  "summary_tasks": {
+    "custom|sv-se-judge-task|0": {
+      "hashes": {
+        "hash_examples": "065b8cb00b0ba585",
+        "hash_full_prompts": "065b8cb00b0ba585",
+        "hash_input_tokens": "f0b579e40d9e9078",
+        "hash_cont_tokens": "f0b579e40d9e9078"
+      },
+      "truncated": 0,
+      "non_truncated": 50,
+      "padded": 0,
+      "non_padded": 50,
+      "effective_few_shots": 0.0,
+      "num_truncated_few_shots": 0
+    }
+  },
+  "summary_general": {
+    "hashes": {
+      "hash_examples": "5192a565300284ce",
+      "hash_full_prompts": "5192a565300284ce",
+      "hash_input_tokens": "86e5ee68d3b3bf0f",
+      "hash_cont_tokens": "86e5ee68d3b3bf0f"
+    },
+    "truncated": 0,
+    "non_truncated": 50,
+    "padded": 0,
+    "non_padded": 50,
+    "num_truncated_few_shots": 0
+  }
+}

sv-se/experiment_2025-12-01_16-11-25/results/openrouter/openai/gpt-5.1/results_2025-12-01T16-21-46.464800.json

@@ -0,0 +1,108 @@
+{
+  "config_general": {
+    "lighteval_sha": "?",
+    "num_fewshot_seeds": 1,
+    "override_batch_size": null,
+    "max_samples": 50,
+    "job_id": 0,
+    "start_time": 5747.899420918,
+    "end_time": 6078.985879173,
+    "total_evaluation_time_secondes": "331.08645825500025",
+    "model_name": "openrouter/openai/gpt-5.1",
+    "model_sha": "",
+    "model_dtype": null,
+    "model_size": "",
+    "generation_parameters": {
+      "early_stopping": null,
+      "repetition_penalty": null,
+      "frequency_penalty": null,
+      "length_penalty": null,
+      "presence_penalty": null,
+      "max_new_tokens": 16384,
+      "min_new_tokens": null,
+      "seed": null,
+      "stop_tokens": null,
+      "temperature": null,
+      "top_k": null,
+      "min_p": null,
+      "top_p": null,
+      "truncate_prompt": null,
+      "response_format": "LegalTermAnalysisGrading"
+    }
+  },
+  "results": {
+    "all": {}
+  },
+  "versions": {
+    "custom|sv-se-judge-task|0": 0
+  },
+  "config_tasks": {
+    "custom|sv-se-judge-task": {
+      "name": "sv-se-judge-task",
+      "prompt_function": "create_alt_test_prompt",
+      "hf_repo": "TransLegal/JudgeDataset",
+      "hf_subset": "sv-se",
+      "metric": [
+        {
+          "metric_name": "alt_test_data_collector",
+          "higher_is_better": {},
+          "category": "3",
+          "use_case": "6",
+          "sample_level_fn": "alt_test_passthrough",
+          "corpus_level_fn": null
+        }
+      ],
+      "hf_revision": null,
+      "hf_filter": null,
+      "hf_avail_splits": [
+        "train"
+      ],
+      "trust_dataset": true,
+      "evaluation_splits": [
+        "train"
+      ],
+      "few_shots_split": null,
+      "few_shots_select": null,
+      "generation_size": 16384,
+      "generation_grammar": null,
+      "stop_sequence": [],
+      "num_samples": null,
+      "suite": [
+        "custom"
+      ],
+      "original_num_docs": 50,
+      "effective_num_docs": 50,
+      "must_remove_duplicate_docs": false,
+      "version": 0
+    }
+  },
+  "summary_tasks": {
+    "custom|sv-se-judge-task|0": {
+      "hashes": {
+        "hash_examples": "065b8cb00b0ba585",
+        "hash_full_prompts": "065b8cb00b0ba585",
+        "hash_input_tokens": "f0b579e40d9e9078",
+        "hash_cont_tokens": "f0b579e40d9e9078"
+      },
+      "truncated": 0,
+      "non_truncated": 50,
+      "padded": 0,
+      "non_padded": 50,
+      "effective_few_shots": 0.0,
+      "num_truncated_few_shots": 0
+    }
+  },
+  "summary_general": {
+    "hashes": {
+      "hash_examples": "5192a565300284ce",
+      "hash_full_prompts": "5192a565300284ce",
+      "hash_input_tokens": "86e5ee68d3b3bf0f",
+      "hash_cont_tokens": "86e5ee68d3b3bf0f"
+    },
+    "truncated": 0,
+    "non_truncated": 50,
+    "padded": 0,
+    "non_padded": 50,
+    "num_truncated_few_shots": 0
+  }
+}