Results for Gemma4 reasoning (#4)

- Results for Gemma4 reasoning (3ddaaaedb3be59fab579196dc569017304deb79d)


Co-authored-by: Bohdan Didenko <transhumanist-already-exists@users.noreply.huggingface.co>

aggregated/results_2026-05-04T11-15-00.000000.json

@@ -0,0 +1,1809 @@
+{
+  "results": {
+    "zno_uk_geography": {
+      "alias": "zno_uk_geography",
+      "exact,strict_letter": 0.9392265193370166,
+      "exact_stderr,strict_letter": 0.007946172946986331,
+      "exact,none": 0.9392265193370166,
+      "exact_stderr,none": 0.007946172946986331
+    },
+    "zno_uk_history": {
+      "alias": "zno_uk_history",
+      "exact,strict_letter": 0.851063829787234,
+      "exact_stderr,strict_letter": 0.008780722477989748,
+      "exact,none": 0.851063829787234,
+      "exact_stderr,none": 0.008780722477989748
+    },
+    "zno_uk_language_and_literature": {
+      "alias": "zno_uk_language_and_literature",
+      "exact,strict_letter": 0.6815020862308763,
+      "exact_stderr,strict_letter": 0.010033732361811326,
+      "exact,none": 0.6815020862308763,
+      "exact_stderr,none": 0.010033732361811326
+    },
+    "zno_uk_math": {
+      "alias": "zno_uk_math",
+      "exact,strict_letter": 0.9782608695652174,
+      "exact_stderr,strict_letter": 0.01245914217229679,
+      "exact,none": 0.9782608695652174,
+      "exact_stderr,none": 0.01245914217229679
+    },
+    "gsm8k_uk": {
+      "alias": "gsm8k_uk",
+      "exact_match,strict-match": 0.9423805913570887,
+      "exact_match_stderr,strict-match": 0.006418593319822864,
+      "exact_match,flexible-extract": 0.9423805913570887,
+      "exact_match_stderr,flexible-extract": 0.006418593319822863
+    },
+    "ifeval_uk": {
+      "alias": "ifeval_uk",
+      "prompt_level_strict_acc,none": 0.8188539741219963,
+      "prompt_level_strict_acc_stderr,none": 0.01657374894371447,
+      "inst_level_strict_acc,none": 0.86810551558753,
+      "inst_level_strict_acc_stderr,none": "N/A",
+      "prompt_level_loose_acc,none": 0.8262476894639557,
+      "prompt_level_loose_acc_stderr,none": 0.01630510287765143,
+      "inst_level_loose_acc,none": 0.8812949640287769,
+      "inst_level_loose_acc_stderr,none": "N/A"
+    },
+    "flores_en-uk": {
+      "alias": "flores_en-uk",
+      "bleu,extract_translation": 32.02354558137718,
+      "bleu_stderr,extract_translation": 0.535057188354709,
+      "chrf,extract_translation": 60.76760815211665,
+      "chrf_stderr,extract_translation": 0.38335034353259617,
+      "bleu,none": 32.02354558137718,
+      "bleu_stderr,none": 0.535057188354709
+    },
+    "long_flores_en-uk": {
+      "alias": "long_flores_en-uk",
+      "bleu,extract_translation": 30.346795006660592,
+      "bleu_stderr,extract_translation": 0.6553890969518961,
+      "chrf,extract_translation": 61.8669726072027,
+      "chrf_stderr,extract_translation": 0.5403178734097245,
+      "bleu,none": 30.346795006660592,
+      "bleu_stderr,none": 0.6553890969518961
+    },
+    "flores_uk-en": {
+      "alias": "flores_uk-en",
+      "bleu,extract_translation": 41.33123040620086,
+      "bleu_stderr,extract_translation": 0.559601838593499,
+      "chrf,extract_translation": 67.03155645382351,
+      "chrf_stderr,extract_translation": 0.37719002595255985,
+      "bleu,none": 41.33123040620086,
+      "bleu_stderr,none": 0.559601838593499
+    },
+    "long_flores_uk-en": {
+      "alias": "long_flores_uk-en",
+      "bleu,extract_translation": 43.61435083784172,
+      "bleu_stderr,extract_translation": 0.6995042566888917,
+      "chrf,extract_translation": 69.83493797333674,
+      "chrf_stderr,extract_translation": 0.4433732141602343,
+      "bleu,none": 43.61435083784172,
+      "bleu_stderr,none": 0.6995042566888917
+    },
+    "wmt_en_uk": {
+      "alias": "wmt_en_uk",
+      "bleu,extract_translation": 30.033035266407538,
+      "bleu_stderr,extract_translation": 0.8616315805539017,
+      "bleu,none": 30.033035266407538,
+      "bleu_stderr,none": 0.8616315805539017
+    },
+    "arc_challenge_uk": {
+      "alias": "arc_challenge_uk",
+      "exact_match,extract_answer": 0.9249146757679181,
+      "exact_match_stderr,extract_answer": 0.007701049554000624,
+      "exact_match,none": 0.9249146757679181,
+      "exact_match_stderr,none": 0.007701049554000624
+    },
+    "triviaqa_uk": {
+      "alias": "triviaqa_uk",
+      "exact_match,remove_whitespace": 0.6310744538564422,
+      "exact_match_stderr,remove_whitespace": 0.003602152838816319
+    },
+    "xlsum_uk": {
+      "alias": "xlsum_uk",
+      "bleu,none": 2.7288584082622362,
+      "bleu_stderr,none": 0.08174601953949884
+    },
+    "arc_easy_uk": {
+      "alias": "arc_easy_uk",
+      "acc,none": 0.9225589225589226,
+      "acc_stderr,none": 0.005484675568425269,
+      "acc_norm,none": 0.7159090909090909,
+      "acc_norm_stderr,none": 0.009253921261885857
+    },
+    "winogrande_uk": {
+      "alias": "winogrande_uk",
+      "acc,none": 0.8445146014206788,
+      "acc_stderr,none": 0.010184308214775674
+    },
+    "belebele_ukr_Cyrl": {
+      "alias": "belebele_ukr_Cyrl",
+      "acc,none": 0.9533333333333334,
+      "acc_stderr,none": 0.0070347057091808655,
+      "acc_norm,none": 0.9533333333333334,
+      "acc_norm_stderr,none": 0.0070347057091808655
+    },
+    "global_mmlu_full_uk": {
+      "alias": "global_mmlu_full_uk",
+      "acc,none": 0.8664007976071785,
+      "acc_stderr,none": 0.002871192657546397
+    },
+    "squad_uk": {
+      "alias": "squad_uk",
+      "exact,none": 60.67327766179541,
+      "exact_stderr,none": "N/A",
+      "f1,none": 71.69369039158182,
+      "f1_stderr,none": "N/A",
+      "HasAns_exact,none": 56.479217603911984,
+      "HasAns_exact_stderr,none": "N/A",
+      "HasAns_f1,none": 78.60472664017195,
+      "HasAns_f1_stderr,none": "N/A",
+      "NoAns_exact,none": 64.83535528596187,
+      "NoAns_exact_stderr,none": "N/A",
+      "NoAns_f1,none": 64.83535528596187,
+      "NoAns_f1_stderr,none": "N/A",
+      "best_exact,none": 61.151704940848994,
+      "best_exact_stderr,none": "N/A",
+      "best_f1,none": 72.04148223398659,
+      "best_f1_stderr,none": "N/A"
+    },
+    "flores_uk": {
+      "alias": "flores_uk",
+      "bleu,none": 36.67738799378902
+    },
+    "long_flores_uk": {
+      "alias": "long_flores_uk",
+      "bleu,none": 36.980572922251156
+    }
+  },
+  "group_subtasks": {
+    "zno_uk_geography": [],
+    "zno_uk_history": [],
+    "zno_uk_language_and_literature": [],
+    "zno_uk_math": [],
+    "gsm8k_uk": [],
+    "ifeval_uk": [],
+    "flores_en-uk": [],
+    "long_flores_en-uk": [],
+    "flores_uk-en": [],
+    "long_flores_uk-en": [],
+    "wmt_en_uk": [],
+    "arc_challenge_uk": [],
+    "triviaqa_uk": [],
+    "xlsum_uk": [],
+    "arc_easy_uk": [],
+    "winogrande_uk": [],
+    "belebele_ukr_Cyrl": [],
+    "global_mmlu_full_uk": [],
+    "squad_uk": []
+  },
+  "configs": {
+    "zno_uk_geography": {
+      "task": "zno_uk_geography",
+      "task_alias": "zno_uk_geography",
+      "doc_to_text": "Питання: {{question.strip()}}\n{% for answer in answers %}{{ 'ABCDE'[loop.index0] }}. {{answer}}\n{% endfor %}Виведи лише одну латинську літеру правильної відповіді: A, B, C, D або E. Без пояснень.\nВідповідь:",
+      "unsafe_code": false,
+      "description": "",
+      "target_delimiter": " ",
+      "fewshot_delimiter": "\n\n",
+      "fewshot_config": {
+        "sampler": "default",
+        "split": null,
+        "process_docs": null,
+        "fewshot_indices": null,
+        "samples": null,
+        "doc_to_text": "Питання: {{question.strip()}}\n{% for answer in answers %}{{ 'ABCDE'[loop.index0] }}. {{answer}}\n{% endfor %}Виведи лише одну латинську літеру правильної відповіді: A, B, C, D або E. Без пояснень.\nВідповідь:",
+        "doc_to_choice": null,
+        "doc_to_target": null,
+        "gen_prefix": null,
+        "fewshot_delimiter": "\n\n",
+        "target_delimiter": " "
+      },
+      "num_fewshot": 0,
+      "output_type": "generate_until",
+      "generation_kwargs": {
+        "until": "<asdasdgfggvvcccx>",
+        "do_sample": true,
+        "temperature": 0.7,
+        "max_gen_toks": 65536,
+        "top_p": 0.95,
+        "skip_special_tokens": false
+      },
+      "repeats": 1,
+      "should_decontaminate": false,
+      "metadata": {
+        "version": "0.0.1",
+        "source_task": "zno_uk_geography",
+        "source_result_file": "eval-results-gemma4-it-zno-reasoning-api-dp4/10431/google__gemma-4-26B-A4B-it/results_strict_letter_2026-04-24T21-33-56.685490.json"
+      },
+      "filter_list": [
+        {
+          "name": "strict_letter",
+          "filter": [
+            {
+              "function": "regex",
+              "regex_pattern": "^\\s*([ABCDE])\\s*$",
+              "group_select": 0,
+              "fallback": "[invalid]"
+            },
+            {
+              "function": "take_first"
+            }
+          ]
+        }
+      ],
+      "alias": "zno_uk_geography"
+    },
+    "zno_uk_history": {
+      "task": "zno_uk_history",
+      "task_alias": "zno_uk_history",
+      "doc_to_text": "Питання: {{question.strip()}}\n{% for answer in answers %}{{ 'ABCDE'[loop.index0] }}. {{answer}}\n{% endfor %}Виведи лише одну латинську літеру правильної відповіді: A, B, C, D або E. Без пояснень.\nВідповідь:",
+      "unsafe_code": false,
+      "description": "",
+      "target_delimiter": " ",
+      "fewshot_delimiter": "\n\n",
+      "fewshot_config": {
+        "sampler": "default",
+        "split": null,
+        "process_docs": null,
+        "fewshot_indices": null,
+        "samples": null,
+        "doc_to_text": "Питання: {{question.strip()}}\n{% for answer in answers %}{{ 'ABCDE'[loop.index0] }}. {{answer}}\n{% endfor %}Виведи лише одну латинську літеру правильної відповіді: A, B, C, D або E. Без пояснень.\nВідповідь:",
+        "doc_to_choice": null,
+        "doc_to_target": null,
+        "gen_prefix": null,
+        "fewshot_delimiter": "\n\n",
+        "target_delimiter": " "
+      },
+      "num_fewshot": 0,
+      "output_type": "generate_until",
+      "generation_kwargs": {
+        "until": "<asdasdgfggvvcccx>",
+        "do_sample": true,
+        "temperature": 0.7,
+        "max_gen_toks": 65536,
+        "top_p": 0.95,
+        "skip_special_tokens": false
+      },
+      "repeats": 1,
+      "should_decontaminate": false,
+      "metadata": {
+        "version": "0.0.1",
+        "source_task": "zno_uk_history",
+        "source_result_file": "eval-results-gemma4-it-zno-reasoning-api-dp4/10431/google__gemma-4-26B-A4B-it/results_strict_letter_2026-04-24T21-33-56.685490.json"
+      },
+      "filter_list": [
+        {
+          "name": "strict_letter",
+          "filter": [
+            {
+              "function": "regex",
+              "regex_pattern": "^\\s*([ABCDE])\\s*$",
+              "group_select": 0,
+              "fallback": "[invalid]"
+            },
+            {
+              "function": "take_first"
+            }
+          ]
+        }
+      ],
+      "alias": "zno_uk_history"
+    },
+    "zno_uk_language_and_literature": {
+      "task": "zno_uk_language_and_literature",
+      "task_alias": "zno_uk_language_and_literature",
+      "doc_to_text": "Питання: {{question.strip()}}\n{% for answer in answers %}{{ 'ABCDE'[loop.index0] }}. {{answer}}\n{% endfor %}Виведи лише одну латинську літеру правильної відповіді: A, B, C, D або E. Без пояснень.\nВідповідь:",
+      "unsafe_code": false,
+      "description": "",
+      "target_delimiter": " ",
+      "fewshot_delimiter": "\n\n",
+      "fewshot_config": {
+        "sampler": "default",
+        "split": null,
+        "process_docs": null,
+        "fewshot_indices": null,
+        "samples": null,
+        "doc_to_text": "Питання: {{question.strip()}}\n{% for answer in answers %}{{ 'ABCDE'[loop.index0] }}. {{answer}}\n{% endfor %}Виведи лише одну латинську літеру правильної відповіді: A, B, C, D або E. Без пояснень.\nВідповідь:",
+        "doc_to_choice": null,
+        "doc_to_target": null,
+        "gen_prefix": null,
+        "fewshot_delimiter": "\n\n",
+        "target_delimiter": " "
+      },
+      "num_fewshot": 0,
+      "output_type": "generate_until",
+      "generation_kwargs": {
+        "until": "<asdasdgfggvvcccx>",
+        "do_sample": true,
+        "temperature": 0.7,
+        "max_gen_toks": 65536,
+        "top_p": 0.95,
+        "skip_special_tokens": false
+      },
+      "repeats": 1,
+      "should_decontaminate": false,
+      "metadata": {
+        "version": "0.0.1",
+        "source_task": "zno_uk_language_and_literature",
+        "source_result_file": "eval-results-gemma4-it-zno-reasoning-api-dp4/10431/google__gemma-4-26B-A4B-it/results_strict_letter_2026-04-24T21-33-56.685490.json"
+      },
+      "filter_list": [
+        {
+          "name": "strict_letter",
+          "filter": [
+            {
+              "function": "regex",
+              "regex_pattern": "^\\s*([ABCDE])\\s*$",
+              "group_select": 0,
+              "fallback": "[invalid]"
+            },
+            {
+              "function": "take_first"
+            }
+          ]
+        }
+      ],
+      "alias": "zno_uk_language_and_literature"
+    },
+    "zno_uk_math": {
+      "task": "zno_uk_math",
+      "task_alias": "zno_uk_math",
+      "doc_to_text": "Питання: {{question.strip()}}\n{% for answer in answers %}{{ 'ABCDE'[loop.index0] }}. {{answer}}\n{% endfor %}Виведи лише одну латинську літеру правильної відповіді: A, B, C, D або E. Без пояснень.\nВідповідь:",
+      "unsafe_code": false,
+      "description": "",
+      "target_delimiter": " ",
+      "fewshot_delimiter": "\n\n",
+      "fewshot_config": {
+        "sampler": "default",
+        "split": null,
+        "process_docs": null,
+        "fewshot_indices": null,
+        "samples": null,
+        "doc_to_text": "Питання: {{question.strip()}}\n{% for answer in answers %}{{ 'ABCDE'[loop.index0] }}. {{answer}}\n{% endfor %}Виведи лише одну латинську літеру правильної відповіді: A, B, C, D або E. Без пояснень.\nВідповідь:",
+        "doc_to_choice": null,
+        "doc_to_target": null,
+        "gen_prefix": null,
+        "fewshot_delimiter": "\n\n",
+        "target_delimiter": " "
+      },
+      "num_fewshot": 0,
+      "output_type": "generate_until",
+      "generation_kwargs": {
+        "until": "<asdasdgfggvvcccx>",
+        "do_sample": true,
+        "temperature": 0.7,
+        "max_gen_toks": 65536,
+        "top_p": 0.95,
+        "skip_special_tokens": false
+      },
+      "repeats": 1,
+      "should_decontaminate": false,
+      "metadata": {
+        "version": "0.0.1",
+        "source_task": "zno_uk_math",
+        "source_result_file": "eval-results-gemma4-it-zno-reasoning-api-dp4/10431/google__gemma-4-26B-A4B-it/results_strict_letter_2026-04-24T21-33-56.685490.json"
+      },
+      "filter_list": [
+        {
+          "name": "strict_letter",
+          "filter": [
+            {
+              "function": "regex",
+              "regex_pattern": "^\\s*([ABCDE])\\s*$",
+              "group_select": 0,
+              "fallback": "[invalid]"
+            },
+            {
+              "function": "take_first"
+            }
+          ]
+        }
+      ],
+      "alias": "zno_uk_math"
+    },
+    "gsm8k_uk": {
+      "task": "gsm8k_uk",
+      "task_alias": "gsm8k_uk",
+      "tag": [
+        "math_word_problems"
+      ],
+      "dataset_path": "INSAIT-Institute/gsm8k_ukr",
+      "training_split": "test",
+      "test_split": "test",
+      "fewshot_split": "test",
+      "doc_to_text": "Питання: {{question}}\nРозв'яжи задачу покроково. Останній рядок виведи рівно у форматі: The answer is N. Не пропускай фінальну крапку.\nВідповідь:",
+      "doc_to_target": "{{answer}}",
+      "unsafe_code": false,
+      "description": "",
+      "target_delimiter": " ",
+      "fewshot_delimiter": "\n\n",
+      "fewshot_config": {
+        "sampler": "default",
+        "split": "test",
+        "process_docs": null,
+        "fewshot_indices": null,
+        "samples": null,
+        "doc_to_text": "Питання: {{question}}\nРозв'яжи задачу покроково. Останній рядок виведи рівно у форматі: The answer is N. Не пропускай фінальну крапку.\nВідповідь:",
+        "doc_to_choice": null,
+        "doc_to_target": "{{answer}}",
+        "gen_prefix": null,
+        "fewshot_delimiter": "\n\n",
+        "target_delimiter": " "
+      },
+      "num_fewshot": 0,
+      "metric_list": [
+        {
+          "metric": "exact_match",
+          "aggregation": "mean",
+          "higher_is_better": true,
+          "ignore_case": true,
+          "ignore_punctuation": false,
+          "regexes_to_ignore": [
+            ",",
+            "\\$",
+            "(?s).*#### ",
+            "\\.$"
+          ]
+        }
+      ],
+      "output_type": "generate_until",
+      "generation_kwargs": {
+        "until": "<asdasdgfggvvcccx>",
+        "do_sample": true,
+        "temperature": 0.7,
+        "max_gen_toks": 65536,
+        "top_p": 0.95,
+        "skip_special_tokens": false
+      },
+      "repeats": 1,
+      "filter_list": [
+        {
+          "name": "strict-match",
+          "filter": [
+            {
+              "function": "regex",
+              "regex_pattern": "The answer is (\\-?[0-9\\.\\,]+)."
+            },
+            {
+              "function": "take_first"
+            }
+          ]
+        },
+        {
+          "name": "flexible-extract",
+          "filter": [
+            {
+              "function": "regex",
+              "group_select": -1,
+              "regex_pattern": "(-?[$0-9.,]{2,})|(-?[0-9]+)"
+            },
+            {
+              "function": "take_first"
+            }
+          ]
+        }
+      ],
+      "should_decontaminate": false,
+      "metadata": {
+        "version": 1.0,
+        "pretrained": "google/gemma-4-26B-A4B-it",
+        "data_parallel_size": 1,
+        "tensor_parallel_size": 1,
+        "gpu_memory_utilization": 0.9,
+        "max_model_len": 131072,
+        "max_num_seqs": 128,
+        "max_num_batched_tokens": 131072,
+        "add_bos_token": true,
+        "reasoning_parser": "gemma4",
+        "enable_thinking": true,
+        "trust_remote_code": true,
+        "source_task": "gsm8k_uk",
+        "source_result_file": "eval-results-gemma4-it-gsm8k-uk-reasoning-dp1/10478/google__gemma-4-26B-A4B-it/results_2026-04-25T11-56-46.578460.json"
+      },
+      "alias": "gsm8k_uk"
+    },
+    "ifeval_uk": {
+      "task": "ifeval_uk",
+      "task_alias": "ifeval_uk",
+      "dataset_path": "INSAIT-Institute/ifeval_ukr",
+      "test_split": "train",
+      "doc_to_text": "Виконай інструкцію нижче буквально. Не додавай вступу чи пояснень поза вимогами.\n\n{{prompt}}",
+      "doc_to_target": 0,
+      "unsafe_code": false,
+      "process_results": "def process_results(doc, results):\n    inp = InputExample(\n        key=doc[\"key\"],\n        instruction_id_list=doc[\"instruction_id_list\"],\n        prompt=doc[\"prompt\"],\n        kwargs=doc[\"kwargs\"],\n    )\n    response = results[0]\n\n    out_strict = test_instruction_following_strict(inp, response)\n    out_loose = test_instruction_following_loose(inp, response)\n\n    return {\n        \"prompt_level_strict_acc\": out_strict.follow_all_instructions,\n        \"inst_level_strict_acc\": out_strict.follow_instruction_list,\n        \"prompt_level_loose_acc\": out_loose.follow_all_instructions,\n        \"inst_level_loose_acc\": out_loose.follow_instruction_list,\n    }\n",
+      "description": "",
+      "target_delimiter": " ",
+      "fewshot_delimiter": "\n\n",
+      "fewshot_config": {
+        "sampler": "default",
+        "split": null,
+        "process_docs": null,
+        "fewshot_indices": null,
+        "samples": null,
+        "doc_to_text": "Виконай інструкцію нижче буквально. Не додавай вступу чи пояснень поза вимогами.\n\n{{prompt}}",
+        "doc_to_choice": null,
+        "doc_to_target": 0,
+        "gen_prefix": null,
+        "fewshot_delimiter": "\n\n",
+        "target_delimiter": " "
+      },
+      "num_fewshot": 0,
+      "metric_list": [
+        {
+          "metric": "prompt_level_strict_acc",
+          "aggregation": "mean",
+          "higher_is_better": true
+        },
+        {
+          "metric": "inst_level_strict_acc",
+          "aggregation": "def agg_inst_level_acc(items):\n    flat_items = [item for sublist in items for item in sublist]\n    inst_level_acc = sum(flat_items) / len(flat_items)\n    return inst_level_acc\n",
+          "higher_is_better": true
+        },
+        {
+          "metric": "prompt_level_loose_acc",
+          "aggregation": "mean",
+          "higher_is_better": true
+        },
+        {
+          "metric": "inst_level_loose_acc",
+          "aggregation": "def agg_inst_level_acc(items):\n    flat_items = [item for sublist in items for item in sublist]\n    inst_level_acc = sum(flat_items) / len(flat_items)\n    return inst_level_acc\n",
+          "higher_is_better": true
+        }
+      ],
+      "output_type": "generate_until",
+      "generation_kwargs": {
+        "until": "<asdasdgfggvvcccx>",
+        "do_sample": true,
+        "temperature": 0.7,
+        "max_gen_toks": 65536,
+        "top_p": 0.95,
+        "skip_special_tokens": false
+      },
+      "repeats": 1,
+      "should_decontaminate": false,
+      "metadata": {
+        "version": 4.0,
+        "pretrained": "google/gemma-4-26B-A4B-it",
+        "data_parallel_size": 4,
+        "tensor_parallel_size": 1,
+        "gpu_memory_utilization": 0.9,
+        "max_model_len": 131072,
+        "max_num_seqs": 128,
+        "max_num_batched_tokens": 131072,
+        "add_bos_token": true,
+        "reasoning_parser": "gemma4",
+        "enable_thinking": true,
+        "trust_remote_code": true,
+        "source_task": "ifeval_uk",
+        "source_result_file": "eval-results-gemma4-it-ifeval-reasoning-api-dp4/10440/google__gemma-4-26B-A4B-it/results_2026-04-24T22-04-45.176450.json"
+      },
+      "alias": "ifeval_uk"
+    },
+    "flores_en-uk": {
+      "task": "flores_en-uk",
+      "task_alias": "flores_en-uk",
+      "dataset_path": "facebook/flores",
+      "dataset_name": "all",
+      "dataset_kwargs": {
+        "trust_remote_code": true
+      },
+      "training_split": "dev",
+      "validation_split": "dev",
+      "test_split": "devtest",
+      "fewshot_split": "dev",
+      "doc_to_text": "Translate this sentence into Ukrainian:\n{{sentence_eng_Latn}}\n\nResponse format must be <translation>переклад українською</translation>.",
+      "doc_to_target": "{{sentence_ukr_Cyrl}}",
+      "unsafe_code": false,
+      "description": "",
+      "target_delimiter": "",
+      "fewshot_delimiter": "\n\n",
+      "fewshot_config": {
+        "sampler": "default",
+        "split": "dev",
+        "process_docs": null,
+        "fewshot_indices": null,
+        "samples": null,
+        "doc_to_text": "Translate this sentence into Ukrainian:\n{{sentence_eng_Latn}}\n\nResponse format must be <translation>переклад українською</translation>.",
+        "doc_to_choice": null,
+        "doc_to_target": "{{sentence_ukr_Cyrl}}",
+        "gen_prefix": null,
+        "fewshot_delimiter": "\n\n",
+        "target_delimiter": ""
+      },
+      "num_fewshot": 0,
+      "metric_list": [
+        {
+          "metric": "bleu",
+          "aggregation": "bleu",
+          "higher_is_better": true
+        },
+        {
+          "metric": "chrf",
+          "aggregation": "chrf",
+          "higher_is_better": true
+        }
+      ],
+      "output_type": "generate_until",
+      "generation_kwargs": {
+        "until": "<asdasdgfggvvcccx>",
+        "temperature": 0.7,
+        "top_p": 0.95,
+        "do_sample": true,
+        "max_gen_toks": 65536,
+        "skip_special_tokens": false
+      },
+      "repeats": 1,
+      "filter_list": [
+        {
+          "name": "extract_translation",
+          "filter": [
+            {
+              "function": "regex",
+              "regex_pattern": "(?s)<translation>\\s*(.*?)\\s*</translation>",
+              "group_select": 0,
+              "fallback": ""
+            },
+            {
+              "function": "take_first"
+            }
+          ]
+        }
+      ],
+      "should_decontaminate": false,
+      "metadata": {
+        "version": 1.0,
+        "pretrained": "google/gemma-4-26B-A4B-it",
+        "data_parallel_size": 2,
+        "tensor_parallel_size": 1,
+        "gpu_memory_utilization": 0.9,
+        "max_model_len": 131072,
+        "max_num_seqs": 128,
+        "max_num_batched_tokens": 131072,
+        "add_bos_token": true,
+        "reasoning_parser": "gemma4",
+        "enable_thinking": true,
+        "trust_remote_code": true,
+        "source_task": "flores_en-uk",
+        "source_result_file": "eval-results-gemma4-it-flores-tagged-reasoning-dp2/10474/google__gemma-4-26B-A4B-it/results_2026-04-25T11-45-02.472693.json"
+      },
+      "alias": "flores_en-uk"
+    },
+    "long_flores_en-uk": {
+      "task": "long_flores_en-uk",
+      "task_alias": "long_flores_en-uk",
+      "dataset_path": "robinhad/long_flores",
+      "dataset_name": "all",
+      "dataset_kwargs": {
+        "trust_remote_code": true
+      },
+      "training_split": "dev",
+      "validation_split": "dev",
+      "test_split": "devtest",
+      "fewshot_split": "dev",
+      "doc_to_text": "Translate this text into Ukrainian:\n{{sentence_eng_Latn}}\n\nResponse format must be <translation>переклад українською</translation>.",
+      "doc_to_target": "{{sentence_ukr_Cyrl}}",
+      "unsafe_code": false,
+      "description": "",
+      "target_delimiter": "",
+      "fewshot_delimiter": "\n\n",
+      "fewshot_config": {
+        "sampler": "default",
+        "split": "dev",
+        "process_docs": null,
+        "fewshot_indices": null,
+        "samples": null,
+        "doc_to_text": "Translate this text into Ukrainian:\n{{sentence_eng_Latn}}\n\nResponse format must be <translation>переклад українською</translation>.",
+        "doc_to_choice": null,
+        "doc_to_target": "{{sentence_ukr_Cyrl}}",
+        "gen_prefix": null,
+        "fewshot_delimiter": "\n\n",
+        "target_delimiter": ""
+      },
+      "num_fewshot": 0,
+      "metric_list": [
+        {
+          "metric": "bleu",
+          "aggregation": "bleu",
+          "higher_is_better": true
+        },
+        {
+          "metric": "chrf",
+          "aggregation": "chrf",
+          "higher_is_better": true
+        }
+      ],
+      "output_type": "generate_until",
+      "generation_kwargs": {
+        "until": "<asdasdgfggvvcccx>",
+        "temperature": 0.7,
+        "top_p": 0.95,
+        "do_sample": true,
+        "max_gen_toks": 65536,
+        "skip_special_tokens": false
+      },
+      "repeats": 1,
+      "filter_list": [
+        {
+          "name": "extract_translation",
+          "filter": [
+            {
+              "function": "regex",
+              "regex_pattern": "(?s)<translation>\\s*(.*?)\\s*</translation>",
+              "group_select": 0,
+              "fallback": ""
+            },
+            {
+              "function": "take_first"
+            }
+          ]
+        }
+      ],
+      "should_decontaminate": false,
+      "metadata": {
+        "version": 1.0,
+        "pretrained": "google/gemma-4-26B-A4B-it",
+        "data_parallel_size": 2,
+        "tensor_parallel_size": 1,
+        "gpu_memory_utilization": 0.9,
+        "max_model_len": 131072,
+        "max_num_seqs": 128,
+        "max_num_batched_tokens": 131072,
+        "add_bos_token": true,
+        "reasoning_parser": "gemma4",
+        "enable_thinking": true,
+        "trust_remote_code": true,
+        "source_task": "long_flores_en-uk",
+        "source_result_file": "eval-results-gemma4-it-long-flores-tagged-reasoning-dp2/10475/google__gemma-4-26B-A4B-it/results_2026-04-25T11-43-14.875363.json"
+      },
+      "alias": "long_flores_en-uk"
+    },
+    "flores_uk-en": {
+      "task": "flores_uk-en",
+      "task_alias": "flores_uk-en",
+      "dataset_path": "facebook/flores",
+      "dataset_name": "all",
+      "dataset_kwargs": {
+        "trust_remote_code": true
+      },
+      "training_split": "dev",
+      "validation_split": "dev",
+      "test_split": "devtest",
+      "fewshot_split": "dev",
+      "doc_to_text": "Translate this sentence into English:\n{{sentence_ukr_Cyrl}}\n\nResponse format must be <translation>English translation</translation>.",
+      "doc_to_target": "{{sentence_eng_Latn}}",
+      "unsafe_code": false,
+      "description": "",
+      "target_delimiter": "",
+      "fewshot_delimiter": "\n\n",
+      "fewshot_config": {
+        "sampler": "default",
+        "split": "dev",
+        "process_docs": null,
+        "fewshot_indices": null,
+        "samples": null,
+        "doc_to_text": "Translate this sentence into English:\n{{sentence_ukr_Cyrl}}\n\nResponse format must be <translation>English translation</translation>.",
+        "doc_to_choice": null,
+        "doc_to_target": "{{sentence_eng_Latn}}",
+        "gen_prefix": null,
+        "fewshot_delimiter": "\n\n",
+        "target_delimiter": ""
+      },
+      "num_fewshot": 0,
+      "metric_list": [
+        {
+          "metric": "bleu",
+          "aggregation": "bleu",
+          "higher_is_better": true
+        },
+        {
+          "metric": "chrf",
+          "aggregation": "chrf",
+          "higher_is_better": true
+        }
+      ],
+      "output_type": "generate_until",
+      "generation_kwargs": {
+        "max_gen_toks": 65536,
+        "until": "<asdasdgfggvvcccx>",
+        "temperature": 0.7,
+        "top_p": 0.95,
+        "skip_special_tokens": false
+      },
+      "repeats": 1,
+      "filter_list": [
+        {
+          "name": "extract_translation",
+          "filter": [
+            {
+              "function": "regex",
+              "regex_pattern": "(?s)<translation>\\s*(.*?)\\s*</translation>",
+              "group_select": 0,
+              "fallback": ""
+            },
+            {
+              "function": "take_first"
+            }
+          ]
+        }
+      ],
+      "should_decontaminate": false,
+      "metadata": {
+        "version": 1.0,
+        "pretrained": "google/gemma-4-26B-A4B-it",
+        "data_parallel_size": 2,
+        "tensor_parallel_size": 1,
+        "gpu_memory_utilization": 0.9,
+        "max_model_len": 131072,
+        "max_num_seqs": 64,
+        "max_num_batched_tokens": 131072,
+        "add_bos_token": true,
+        "reasoning_parser": "gemma4",
+        "enable_thinking": true,
+        "trust_remote_code": true,
+        "source_task": "flores_uk-en",
+        "source_result_file": "eval-results-gemma4-it-flores-uk-en-tagged-reasoning-dp2/google__gemma-4-26B-A4B-it/results_2026-05-03T00-06-56.117521.json"
+      },
+      "alias": "flores_uk-en"
+    },
+    "long_flores_uk-en": {
+      "task": "long_flores_uk-en",
+      "task_alias": "long_flores_uk-en",
+      "dataset_path": "robinhad/long_flores",
+      "dataset_name": "all",
+      "dataset_kwargs": {
+        "trust_remote_code": true
+      },
+      "training_split": "dev",
+      "validation_split": "dev",
+      "test_split": "devtest",
+      "fewshot_split": "dev",
+      "doc_to_text": "Translate this text into English:\n{{sentence_ukr_Cyrl}}\n\nResponse format must be <translation>English translation</translation>.",
+      "doc_to_target": "{{sentence_eng_Latn}}",
+      "unsafe_code": false,
+      "description": "",
+      "target_delimiter": "",
+      "fewshot_delimiter": "\n\n",
+      "fewshot_config": {
+        "sampler": "default",
+        "split": "dev",
+        "process_docs": null,
+        "fewshot_indices": null,
+        "samples": null,
+        "doc_to_text": "Translate this text into English:\n{{sentence_ukr_Cyrl}}\n\nResponse format must be <translation>English translation</translation>.",
+        "doc_to_choice": null,
+        "doc_to_target": "{{sentence_eng_Latn}}",
+        "gen_prefix": null,
+        "fewshot_delimiter": "\n\n",
+        "target_delimiter": ""
+      },
+      "num_fewshot": 0,
+      "metric_list": [
+        {
+          "metric": "bleu",
+          "aggregation": "bleu",
+          "higher_is_better": true
+        },
+        {
+          "metric": "chrf",
+          "aggregation": "chrf",
+          "higher_is_better": true
+        }
+      ],
+      "output_type": "generate_until",
+      "generation_kwargs": {
+        "max_gen_toks": 65536,
+        "until": "<asdasdgfggvvcccx>",
+        "temperature": 0.7,
+        "top_p": 0.95,
+        "skip_special_tokens": false
+      },
+      "repeats": 1,
+      "filter_list": [
+        {
+          "name": "extract_translation",
+          "filter": [
+            {
+              "function": "regex",
+              "regex_pattern": "(?s)<translation>\\s*(.*?)\\s*</translation>",
+              "group_select": 0,
+              "fallback": ""
+            },
+            {
+              "function": "take_first"
+            }
+          ]
+        }
+      ],
+      "should_decontaminate": false,
+      "metadata": {
+        "version": 1.0,
+        "pretrained": "google/gemma-4-26B-A4B-it",
+        "data_parallel_size": 2,
+        "tensor_parallel_size": 1,
+        "gpu_memory_utilization": 0.9,
+        "max_model_len": 131072,
+        "max_num_seqs": 64,
+        "max_num_batched_tokens": 131072,
+        "add_bos_token": true,
+        "reasoning_parser": "gemma4",
+        "enable_thinking": true,
+        "trust_remote_code": true,
+        "source_task": "long_flores_uk-en",
+        "source_result_file": "eval-results-gemma4-it-flores-uk-en-tagged-reasoning-dp2/google__gemma-4-26B-A4B-it/results_2026-05-03T00-06-56.117521.json"
+      },
+      "alias": "long_flores_uk-en"
+    },
+    "wmt_en_uk": {
+      "task": "wmt_en_uk",
+      "task_alias": "wmt_en_uk",
+      "dataset_path": "google/wmt24pp",
+      "dataset_name": "en-uk_UA",
+      "dataset_kwargs": {
+        "trust_remote_code": true
+      },
+      "training_split": "train",
+      "validation_split": "train",
+      "test_split": "train",
+      "fewshot_split": "train",
+      "doc_to_text": "Translate this text into Ukrainian:\n{{source}}\n\nResponse format must be <translation>переклад українською</translation>.",
+      "doc_to_target": "{{target}}",
+      "unsafe_code": false,
+      "description": "",
+      "target_delimiter": " ",
+      "fewshot_delimiter": "\n\n",
+      "fewshot_config": {
+        "sampler": "default",
+        "split": "train",
+        "process_docs": null,
+        "fewshot_indices": null,
+        "samples": null,
+        "doc_to_text": "Translate this text into Ukrainian:\n{{source}}\n\nResponse format must be <translation>переклад українською</translation>.",
+        "doc_to_choice": null,
+        "doc_to_target": "{{target}}",
+        "gen_prefix": null,
+        "fewshot_delimiter": "\n\n",
+        "target_delimiter": " "
+      },
+      "num_fewshot": 0,
+      "metric_list": [
+        {
+          "metric": "bleu",
+          "aggregation": "bleu",
+          "higher_is_better": true
+        }
+      ],
+      "output_type": "generate_until",
+      "generation_kwargs": {
+        "until": "<asdasdgfggvvcccx>",
+        "temperature": 0.7,
+        "top_p": 0.95,
+        "do_sample": true,
+        "max_gen_toks": 65536,
+        "skip_special_tokens": false
+      },
+      "repeats": 1,
+      "filter_list": [
+        {
+          "name": "extract_translation",
+          "filter": [
+            {
+              "function": "regex",
+              "regex_pattern": "(?s)<translation>\\s*(.*?)\\s*</translation>",
+              "group_select": 0,
+              "fallback": ""
+            },
+            {
+              "function": "take_first"
+            }
+          ]
+        }
+      ],
+      "should_decontaminate": false,
+      "metadata": {
+        "version": 1.0,
+        "pretrained": "google/gemma-4-26B-A4B-it",
+        "data_parallel_size": 2,
+        "tensor_parallel_size": 1,
+        "gpu_memory_utilization": 0.9,
+        "max_model_len": 131072,
+        "max_num_seqs": 128,
+        "max_num_batched_tokens": 131072,
+        "add_bos_token": true,
+        "reasoning_parser": "gemma4",
+        "enable_thinking": true,
+        "trust_remote_code": true,
+        "source_task": "wmt_en_uk",
+        "source_result_file": "eval-results-gemma4-it-wmt-en-uk-tagged-reasoning-dp2/10498/google__gemma-4-26B-A4B-it/results_2026-04-25T15-17-26.066466.json"
+      },
+      "alias": "wmt_en_uk"
+    },
+    "arc_challenge_uk": {
+      "task": "arc_challenge_uk",
+      "task_alias": "arc_challenge_uk",
+      "tag": [
+        "llama"
+      ],
+      "dataset_path": "INSAIT-Institute/arc-challenge_ukr",
+      "training_split": "train",
+      "validation_split": "validation",
+      "test_split": "test",
+      "fewshot_split": "train",
+      "doc_to_text": "Маючи наступне запитання та чотири варіанти відповіді (A, B, C and D), оберіть найкращу відповідь.\\nПитання: {{question.strip()}}\\nA. {{choices.text[0]}}\\nB. {{choices.text[1]}}\\nC. {{choices.text[2]}}{% if choices.text|length > 3 %}\\nD. {{choices.text[3]}}{% endif %}\\nВиведи лише фінальну відповідь у форматі \"Найкраща відповідь — X\", де X — це одна з A, B, C або D.",
+      "doc_to_target": "{{ 'ABCD'[answerKey|int - 1] if answerKey|string in '1234' else answerKey }}",
+      "unsafe_code": false,
+      "description": "",
+      "target_delimiter": " ",
+      "fewshot_delimiter": "\n\n",
+      "fewshot_config": {
+        "sampler": "default",
+        "split": "train",
+        "process_docs": null,
+        "fewshot_indices": null,
+        "samples": null,
+        "doc_to_text": "Маючи наступне запитання та чотири варіанти відповіді (A, B, C and D), оберіть найкращу відповідь.\\nПитання: {{question.strip()}}\\nA. {{choices.text[0]}}\\nB. {{choices.text[1]}}\\nC. {{choices.text[2]}}{% if choices.text|length > 3 %}\\nD. {{choices.text[3]}}{% endif %}\\nВиведи лише фінальну відповідь у форматі \"Найкраща відповідь — X\", де X — це одна з A, B, C або D.",
+        "doc_to_choice": null,
+        "doc_to_target": "{{ 'ABCD'[answerKey|int - 1] if answerKey|string in '1234' else answerKey }}",
+        "gen_prefix": "Найкраща відповідь — ",
+        "fewshot_delimiter": "\n\n",
+        "target_delimiter": " "
+      },
+      "num_fewshot": 0,
+      "metric_list": [
+        {
+          "metric": "exact_match",
+          "aggregation": "mean",
+          "higher_is_better": true,
+          "ignore_case": true,
+          "ignore_punctuation": true
+        }
+      ],
+      "output_type": "generate_until",
+      "generation_kwargs": {
+        "max_gen_toks": 16384,
+        "until": "<asdasdgfggvvcccx>",
+        "temperature": 0.7,
+        "top_p": 0.95,
+        "do_sample": true,
+        "skip_special_tokens": false
+      },
+      "repeats": 1,
+      "filter_list": [
+        {
+          "name": "extract_answer",
+          "filter": [
+            {
+              "function": "regex",
+              "group_select": -1,
+              "regex_pattern": "(?is)(?:Найкраща\\s+відповідь\\s*[—:-]\\s*([ABCD]))|(?:^\\s*([ABCD])\\b)",
+              "fallback": "[invalid]"
+            },
+            {
+              "function": "take_first"
+            }
+          ]
+        }
+      ],
+      "should_decontaminate": false,
+      "gen_prefix": "Найкраща відповідь — ",
+      "metadata": {
+        "version": 1.0,
+        "pretrained": "google/gemma-4-26B-A4B-it",
+        "data_parallel_size": 2,
+        "tensor_parallel_size": 1,
+        "gpu_memory_utilization": 0.9,
+        "max_model_len": 131072,
+        "max_num_seqs": 128,
+        "max_num_batched_tokens": 131072,
+        "add_bos_token": true,
+        "reasoning_parser": "gemma4",
+        "enable_thinking": true,
+        "trust_remote_code": true,
+        "source_task": "arc_challenge_uk",
+        "source_result_file": "eval-results-gemma4-it-arc-challenge-uk-budget16k-parserfix-reasoning-dp2/google__gemma-4-26B-A4B-it/results_2026-04-26T01-33-50.856789.json"
+      },
+      "alias": "arc_challenge_uk"
+    },
+    "triviaqa_uk": {
+      "task": "triviaqa_uk",
+      "task_alias": "triviaqa_uk",
+      "dataset_path": "INSAIT-Institute/trivia_qa_ukr",
+      "training_split": "validation",
+      "validation_split": "validation",
+      "test_split": "validation",
+      "doc_to_text": "Питання: {{question.rstrip('?')}}?\nДай лише точне ім'я або назву, без пояснень. Не перекладай власні назви, якщо це не потрібно.\nВідповідь:",
+      "doc_to_target": "{{answer.aliases}}",
+      "unsafe_code": false,
+      "description": "",
+      "target_delimiter": " ",
+      "fewshot_delimiter": "\n\n",
+      "fewshot_config": {
+        "sampler": "default",
+        "split": null,
+        "process_docs": null,
+        "fewshot_indices": null,
+        "samples": null,
+        "doc_to_text": "Питання: {{question.rstrip('?')}}?\nДай лише точне ім'я або назву, без пояснень. Не перекладай власні назви, якщо це не потрібно.\nВідповідь:",
+        "doc_to_choice": null,
+        "doc_to_target": "{{answer.aliases}}",
+        "gen_prefix": null,
+        "fewshot_delimiter": "\n\n",
+        "target_delimiter": " "
+      },
+      "num_fewshot": 0,
+      "metric_list": [
+        {
+          "metric": "exact_match",
+          "aggregation": "mean",
+          "higher_is_better": true,
+          "ignore_case": true,
+          "ignore_punctuation": true
+        }
+      ],
+      "output_type": "generate_until",
+      "generation_kwargs": {
+        "max_gen_toks": 8192,
+        "until": "<asdasdgfggvvcccx>",
+        "do_sample": true,
+        "temperature": 0.7,
+        "top_p": 0.95,
+        "skip_special_tokens": false
+      },
+      "repeats": 1,
+      "filter_list": [
+        {
+          "name": "remove_whitespace",
+          "filter": [
+            {
+              "function": "remove_whitespace"
+            },
+            {
+              "function": "take_first"
+            }
+          ]
+        }
+      ],
+      "should_decontaminate": true,
+      "doc_to_decontamination_query": "question",
+      "metadata": {
+        "version": 3.0,
+        "pretrained": "google/gemma-4-26B-A4B-it",
+        "data_parallel_size": 2,
+        "tensor_parallel_size": 1,
+        "gpu_memory_utilization": 0.9,
+        "max_model_len": 131072,
+        "max_num_seqs": 128,
+        "max_num_batched_tokens": 131072,
+        "add_bos_token": true,
+        "reasoning_parser": "gemma4",
+        "enable_thinking": true,
+        "trust_remote_code": true,
+        "source_task": "triviaqa_uk",
+        "source_result_file": "eval-results-gemma4-it-triviaqa-uk-budget8k-reasoning-dp2/google__gemma-4-26B-A4B-it/results_2026-04-26T01-10-13.387281.json"
+      },
+      "alias": "triviaqa_uk"
+    },
+    "xlsum_uk": {
+      "task": "xlsum_uk",
+      "task_alias": "xlsum_uk",
+      "dataset_path": "csebuetnlp/xlsum",
+      "dataset_name": "ukrainian",
+      "training_split": "train",
+      "validation_split": "validation",
+      "test_split": "test",
+      "fewshot_split": "train",
+      "process_docs": "def process_xlsum(dataset):\n    \"\"\"\n    Process XLSum dataset by cleaning up text and summary fields.\n    Removes any double spaces from both fields using regex substitution.\n    \"\"\"\n    def _process_doc(doc):\n        # Remove double spaces\n        doc[\"text\"] = re.sub(r\" +\", \" \", doc[\"text\"])\n        doc[\"summary\"] = re.sub(r\" +\", \" \", doc[\"summary\"])\n        return doc\n\n    return dataset.map(_process_doc)\n",
+      "doc_to_text": "Текст: {{text}}\n\nНапиши один короткий підсумок українською. Лише головний факт. Без префіксів, заголовка чи пояснень.",
+      "doc_to_target": "{{summary}}",
+      "unsafe_code": false,
+      "description": "",
+      "target_delimiter": " ",
+      "fewshot_delimiter": "\n\n",
+      "fewshot_config": {
+        "sampler": "default",
+        "split": "train",
+        "process_docs": "<function process_xlsum at 0x7ff649bbcea0>",
+        "fewshot_indices": null,
+        "samples": null,
+        "doc_to_text": "Текст: {{text}}\n\nНапиши один коротки�� підсумок українською. Лише головний факт. Без префіксів, заголовка чи пояснень.",
+        "doc_to_choice": null,
+        "doc_to_target": "{{summary}}",
+        "gen_prefix": null,
+        "fewshot_delimiter": "\n\n",
+        "target_delimiter": " "
+      },
+      "num_fewshot": 0,
+      "metric_list": [
+        {
+          "metric": "bleu",
+          "aggregation": "bleu",
+          "higher_is_better": true
+        }
+      ],
+      "output_type": "generate_until",
+      "generation_kwargs": {
+        "max_gen_toks": 65536,
+        "until": "<asdasdgfggvvcccx>",
+        "do_sample": true,
+        "temperature": 0.7,
+        "top_p": 0.95,
+        "skip_special_tokens": false
+      },
+      "repeats": 1,
+      "should_decontaminate": false,
+      "metadata": {
+        "version": 1.0,
+        "pretrained": "google/gemma-4-26B-A4B-it",
+        "data_parallel_size": 1,
+        "tensor_parallel_size": 1,
+        "gpu_memory_utilization": 0.9,
+        "max_model_len": 131072,
+        "max_num_seqs": 128,
+        "max_num_batched_tokens": 131072,
+        "add_bos_token": true,
+        "reasoning_parser": "gemma4",
+        "enable_thinking": true,
+        "trust_remote_code": true,
+        "source_task": "xlsum_uk",
+        "source_result_file": "eval-results-gemma4-it-xlsum-uk-repro-best-reasoning-dp1/google__gemma-4-26B-A4B-it/results_2026-04-30T17-26-30.384224.json"
+      },
+      "alias": "xlsum_uk"
+    },
+    "arc_easy_uk": {
+      "task": "arc_easy_uk",
+      "dataset_path": "parquet",
+      "dataset_kwargs": {
+        "data_files": {
+          "test": "cot_traces/arc_easy_uk/arc_easy_uk_cot.parquet"
+        }
+      },
+      "test_split": "test",
+      "doc_to_text": "def arc_easy_doc_to_text_cot(doc) -> str:\n    labels = doc[\"choices\"][\"label\"]\n    choices = doc[\"choices\"][\"text\"]\n    options = \"\\n\".join(f\"{label}. {choice}\" for label, choice in zip(labels, choices))\n    prompt = f\"Питання: {doc['question']}\\n{options}\"\n    return _append_trace(prompt, doc)\n",
+      "doc_to_target": "def arc_easy_doc_to_target_index(doc) -> int:\n    return doc[\"choices\"][\"label\"].index(doc[\"answerKey\"])\n",
+      "unsafe_code": false,
+      "doc_to_choice": "def arc_easy_doc_to_choice_text(doc) -> list[str]:\n    return doc[\"choices\"][\"text\"]\n",
+      "description": "",
+      "target_delimiter": " ",
+      "fewshot_delimiter": "\n\n",
+      "fewshot_config": {
+        "sampler": "default",
+        "split": null,
+        "process_docs": null,
+        "fewshot_indices": null,
+        "samples": null,
+        "doc_to_text": "<function arc_easy_doc_to_text_cot at 0x7fe357932e80>",
+        "doc_to_choice": "<function arc_easy_doc_to_choice_text at 0x7fe312775300>",
+        "doc_to_target": "<function arc_easy_doc_to_target_index at 0x7fe312774680>",
+        "gen_prefix": null,
+        "fewshot_delimiter": "\n\n",
+        "target_delimiter": " "
+      },
+      "num_fewshot": 0,
+      "metric_list": [
+        {
+          "metric": "acc",
+          "aggregation": "mean",
+          "higher_is_better": true
+        },
+        {
+          "metric": "acc_norm",
+          "aggregation": "mean",
+          "higher_is_better": true
+        }
+      ],
+      "output_type": "multiple_choice",
+      "repeats": 1,
+      "should_decontaminate": false,
+      "metadata": {
+        "version": 0.1,
+        "reasoning_mode": "cot_conditioned_likelihood",
+        "stage2_trace_max_chars": 30000,
+        "pretrained": "google/gemma-4-26B-A4B-it",
+        "data_parallel_size": 1,
+        "tensor_parallel_size": 1,
+        "gpu_memory_utilization": 0.55,
+        "max_model_len": 32768,
+        "max_num_seqs": 32,
+        "max_num_batched_tokens": 32768,
+        "enforce_eager": true,
+        "add_bos_token": true,
+        "enable_thinking": true,
+        "trust_remote_code": true,
+        "source_task": "arc_easy_uk_cot_likelihood",
+        "source_result_file": "eval-results-gemma4-it-arc_easy_uk_cot_likelihood-stage2-dp1-32768-tracecap30000-tracefix/google__gemma-4-26B-A4B-it/results_2026-05-04T01-23-28.838093.json"
+      },
+      "alias": "arc_easy_uk"
+    },
+    "winogrande_uk": {
+      "task": "winogrande_uk",
+      "dataset_path": "parquet",
+      "dataset_kwargs": {
+        "data_files": {
+          "validation": "cot_traces/winogrande_uk/winogrande_uk_cot.parquet"
+        }
+      },
+      "validation_split": "validation",
+      "test_split": "validation",
+      "doc_to_text": "def winogrande_doc_to_text_cot(doc) -> str:\n    prompt = (\n        f\"Речення з пропуском: {doc['sentence']}\\n\"\n        f\"1. {doc['option1']}\\n\"\n        f\"2. {doc['option2']}\"\n    )\n    return _append_trace(prompt, doc)\n",
+      "doc_to_target": "def winogrande_doc_to_target_index(doc) -> int:\n    return int(doc[\"answer\"]) - 1\n",
+      "unsafe_code": false,
+      "doc_to_choice": "def winogrande_doc_to_choice_options(doc) -> list[str]:\n    return [doc[\"option1\"], doc[\"option2\"]]\n",
+      "description": "",
+      "target_delimiter": " ",
+      "fewshot_delimiter": "\n\n",
+      "fewshot_config": {
+        "sampler": "default",
+        "split": null,
+        "process_docs": null,
+        "fewshot_indices": null,
+        "samples": null,
+        "doc_to_text": "<function winogrande_doc_to_text_cot at 0x7fe86beac680>",
+        "doc_to_choice": "<function winogrande_doc_to_choice_options at 0x7fe86be96fc0>",
+        "doc_to_target": "<function winogrande_doc_to_target_index at 0x7fe86bead4e0>",
+        "gen_prefix": null,
+        "fewshot_delimiter": "\n\n",
+        "target_delimiter": " "
+      },
+      "num_fewshot": 0,
+      "metric_list": [
+        {
+          "metric": "acc",
+          "aggregation": "mean",
+          "higher_is_better": true
+        }
+      ],
+      "output_type": "multiple_choice",
+      "repeats": 1,
+      "should_decontaminate": false,
+      "metadata": {
+        "version": 0.1,
+        "reasoning_mode": "cot_conditioned_likelihood",
+        "stage2_trace_max_chars": 30000,
+        "pretrained": "google/gemma-4-26B-A4B-it",
+        "data_parallel_size": 1,
+        "tensor_parallel_size": 1,
+        "gpu_memory_utilization": 0.55,
+        "max_model_len": 32768,
+        "max_num_seqs": 32,
+        "max_num_batched_tokens": 32768,
+        "enforce_eager": true,
+        "add_bos_token": true,
+        "enable_thinking": true,
+        "trust_remote_code": true,
+        "source_task": "winogrande_uk_cot_likelihood",
+        "source_result_file": "eval-results-gemma4-it-winogrande_uk_cot_likelihood-stage2-dp1-32768-tracecap30000-tracefix/google__gemma-4-26B-A4B-it/results_2026-05-04T01-30-14.239658.json"
+      },
+      "alias": "winogrande_uk"
+    },
+    "belebele_ukr_Cyrl": {
+      "task": "belebele_ukr_Cyrl",
+      "dataset_path": "parquet",
+      "dataset_kwargs": {
+        "data_files": {
+          "test": "cot_traces/belebele_ukr_Cyrl/belebele_ukr_Cyrl_cot.parquet"
+        }
+      },
+      "test_split": "test",
+      "doc_to_text": "def belebele_doc_to_text_cot(doc) -> str:\n    prompt = (\n        f\"P: {doc['flores_passage']}\\n\"\n        f\"Q: {doc['question'].strip()}\\n\"\n        f\"1. {doc['mc_answer1']}\\n\"\n        f\"2. {doc['mc_answer2']}\\n\"\n        f\"3. {doc['mc_answer3']}\\n\"\n        f\"4. {doc['mc_answer4']}\"\n    )\n    return _append_trace(prompt, doc)\n",
+      "doc_to_target": "def belebele_doc_to_target_index(doc) -> int:\n    return int(doc[\"correct_answer_num\"]) - 1\n",
+      "unsafe_code": false,
+      "doc_to_choice": "def belebele_doc_to_choice_numbers(doc) -> list[str]:\n    return [\"1\", \"2\", \"3\", \"4\"]\n",
+      "description": "",
+      "target_delimiter": " ",
+      "fewshot_delimiter": "\n\n",
+      "fewshot_config": {
+        "sampler": "default",
+        "split": null,
+        "process_docs": null,
+        "fewshot_indices": null,
+        "samples": null,
+        "doc_to_text": "<function belebele_doc_to_text_cot at 0x7f9559cd40e0>",
+        "doc_to_choice": "<function belebele_doc_to_choice_numbers at 0x7f958d8932e0>",
+        "doc_to_target": "<function belebele_doc_to_target_index at 0x7f9559cd4f40>",
+        "gen_prefix": null,
+        "fewshot_delimiter": "\n\n",
+        "target_delimiter": " "
+      },
+      "num_fewshot": 0,
+      "metric_list": [
+        {
+          "metric": "acc",
+          "aggregation": "mean",
+          "higher_is_better": true
+        },
+        {
+          "metric": "acc_norm",
+          "aggregation": "mean",
+          "higher_is_better": true
+        }
+      ],
+      "output_type": "multiple_choice",
+      "repeats": 1,
+      "should_decontaminate": false,
+      "metadata": {
+        "version": 0.1,
+        "reasoning_mode": "cot_conditioned_likelihood",
+        "stage2_trace_max_chars": 30000,
+        "pretrained": "google/gemma-4-26B-A4B-it",
+        "data_parallel_size": 1,
+        "tensor_parallel_size": 1,
+        "gpu_memory_utilization": 0.55,
+        "max_model_len": 32768,
+        "max_num_seqs": 32,
+        "max_num_batched_tokens": 32768,
+        "enforce_eager": true,
+        "add_bos_token": true,
+        "enable_thinking": true,
+        "trust_remote_code": true,
+        "source_task": "belebele_ukr_Cyrl_cot_likelihood",
+        "source_result_file": "eval-results-gemma4-it-belebele_ukr_Cyrl_cot_likelihood-stage2-dp1-32768-tracecap30000-tracefix/google__gemma-4-26B-A4B-it/results_2026-05-04T01-36-21.390956.json"
+      },
+      "alias": "belebele_ukr_Cyrl"
+    },
+    "global_mmlu_full_uk": {
+      "task": "global_mmlu_full_uk",
+      "dataset_path": "parquet",
+      "dataset_kwargs": {
+        "data_files": {
+          "test": "cot_traces/global_mmlu_full_uk/global_mmlu_full_uk_cot.parquet"
+        }
+      },
+      "test_split": "test",
+      "doc_to_text": "def global_mmlu_doc_to_text_cot(doc) -> str:\n    prompt = (\n        f\"{doc['question'].strip()}\\n\"\n        f\"A. {doc['option_a']}\\n\"\n        f\"B. {doc['option_b']}\\n\"\n        f\"C. {doc['option_c']}\\n\"\n        f\"D. {doc['option_d']}\"\n    )\n    return _append_trace(prompt, doc)\n",
+      "doc_to_target": "def global_mmlu_doc_to_target_letter(doc) -> str:\n    return doc[\"answer\"]\n",
+      "unsafe_code": false,
+      "doc_to_choice": "def global_mmlu_doc_to_choice_letters(doc) -> list[str]:\n    return [\"A\", \"B\", \"C\", \"D\"]\n",
+      "description": "",
+      "target_delimiter": " ",
+      "fewshot_delimiter": "\n\n",
+      "fewshot_config": {
+        "sampler": "default",
+        "split": null,
+        "process_docs": null,
+        "fewshot_indices": null,
+        "samples": null,
+        "doc_to_text": "<function global_mmlu_doc_to_text_cot at 0x7fe11d00fec0>",
+        "doc_to_choice": "<function global_mmlu_doc_to_choice_letters at 0x7fe11d00f4c0>",
+        "doc_to_target": "<function global_mmlu_doc_to_target_letter at 0x7fe0e9cdcd60>",
+        "gen_prefix": null,
+        "fewshot_delimiter": "\n\n",
+        "target_delimiter": " "
+      },
+      "num_fewshot": 0,
+      "metric_list": [
+        {
+          "metric": "acc",
+          "aggregation": "mean",
+          "higher_is_better": true
+        }
+      ],
+      "output_type": "multiple_choice",
+      "repeats": 1,
+      "should_decontaminate": false,
+      "metadata": {
+        "version": 0.1,
+        "reasoning_mode": "cot_conditioned_likelihood",
+        "stage2_trace_max_chars": 30000,
+        "pretrained": "google/gemma-4-26B-A4B-it",
+        "data_parallel_size": 1,
+        "tensor_parallel_size": 1,
+        "gpu_memory_utilization": 0.55,
+        "max_model_len": 32768,
+        "max_num_seqs": 32,
+        "max_num_batched_tokens": 32768,
+        "enforce_eager": true,
+        "add_bos_token": true,
+        "enable_thinking": true,
+        "trust_remote_code": true,
+        "source_task": "global_mmlu_full_uk_cot_likelihood",
+        "source_result_file": "eval-results-gemma4-it-global_mmlu_full_uk_cot_likelihood-stage2-dp1-32768-tracecap30000-tracefix/google__gemma-4-26B-A4B-it/results_2026-05-04T03-08-10.239062.json"
+      },
+      "alias": "global_mmlu_full_uk"
+    },
+    "squad_uk": {
+      "task": "squad_uk",
+      "dataset_path": "parquet",
+      "dataset_kwargs": {
+        "data_files": {
+          "validation": "cot_traces/squad_uk/squad_uk_cot.parquet"
+        }
+      },
+      "validation_split": "validation",
+      "test_split": "validation",
+      "unsafe_code": false,
+      "description": "",
+      "target_delimiter": " ",
+      "fewshot_delimiter": "\n\n",
+      "fewshot_config": {
+        "sampler": "default",
+        "split": null,
+        "process_docs": null,
+        "fewshot_indices": null,
+        "samples": null,
+        "doc_to_text": null,
+        "doc_to_choice": null,
+        "doc_to_target": null,
+        "gen_prefix": null,
+        "fewshot_delimiter": "\n\n",
+        "target_delimiter": " "
+      },
+      "num_fewshot": 0,
+      "output_type": "loglikelihood",
+      "repeats": 1,
+      "should_decontaminate": false,
+      "metadata": {
+        "version": 0.1,
+        "reasoning_mode": "cot_conditioned_likelihood",
+        "stage2_trace_max_chars": 30000,
+        "source_task": "squad_uk_cot_likelihood",
+        "source_result_file": "eval-results-gemma4-it-squad_uk_cot_likelihood-stage2-dp2-32768-tracecap30000-tracefix/google__gemma-4-26B-A4B-it/results_2026-05-03T22-12-20.783594.json"
+      },
+      "alias": "squad_uk"
+    }
+  },
+  "versions": {
+    "zno_uk_geography": "0.0.1",
+    "zno_uk_history": "0.0.1",
+    "zno_uk_language_and_literature": "0.0.1",
+    "zno_uk_math": "0.0.1",
+    "gsm8k_uk": 1.0,
+    "ifeval_uk": 4.0,
+    "flores_en-uk": 1.0,
+    "long_flores_en-uk": 1.0,
+    "flores_uk-en": 1.0,
+    "long_flores_uk-en": 1.0,
+    "wmt_en_uk": 1.0,
+    "arc_challenge_uk": 1.0,
+    "triviaqa_uk": 3.0,
+    "xlsum_uk": 1.0,
+    "arc_easy_uk": 0.1,
+    "winogrande_uk": 0.1,
+    "belebele_ukr_Cyrl": 0.1,
+    "global_mmlu_full_uk": 0.1,
+    "squad_uk": 0.1
+  },
+  "n-shot": {
+    "zno_uk_geography": 0,
+    "zno_uk_history": 0,
+    "zno_uk_language_and_literature": 0,
+    "zno_uk_math": 0,
+    "gsm8k_uk": 0,
+    "ifeval_uk": 0,
+    "flores_en-uk": 0,
+    "long_flores_en-uk": 0,
+    "flores_uk-en": 0,
+    "long_flores_uk-en": 0,
+    "wmt_en_uk": 0,
+    "arc_challenge_uk": 0,
+    "triviaqa_uk": 0,
+    "xlsum_uk": 0,
+    "arc_easy_uk": 0,
+    "winogrande_uk": 0,
+    "belebele_ukr_Cyrl": 0,
+    "global_mmlu_full_uk": 0,
+    "squad_uk": 0,
+    "flores_uk": 0,
+    "long_flores_uk": 0
+  },
+  "higher_is_better": {
+    "zno_uk_geography": {
+      "exact": true
+    },
+    "zno_uk_history": {
+      "exact": true
+    },
+    "zno_uk_language_and_literature": {
+      "exact": true
+    },
+    "zno_uk_math": {
+      "exact": true
+    },
+    "gsm8k_uk": {
+      "exact_match": true
+    },
+    "ifeval_uk": {
+      "prompt_level_strict_acc": true,
+      "inst_level_strict_acc": true,
+      "prompt_level_loose_acc": true,
+      "inst_level_loose_acc": true
+    },
+    "flores_en-uk": {
+      "bleu": true,
+      "chrf": true
+    },
+    "long_flores_en-uk": {
+      "bleu": true,
+      "chrf": true
+    },
+    "flores_uk-en": {
+      "bleu": true,
+      "chrf": true
+    },
+    "long_flores_uk-en": {
+      "bleu": true,
+      "chrf": true
+    },
+    "wmt_en_uk": {
+      "bleu": true
+    },
+    "arc_challenge_uk": {
+      "exact_match": true
+    },
+    "triviaqa_uk": {
+      "exact_match": true
+    },
+    "xlsum_uk": {
+      "bleu": true
+    },
+    "arc_easy_uk": {
+      "acc": true,
+      "acc_norm": true
+    },
+    "winogrande_uk": {
+      "acc": true
+    },
+    "belebele_ukr_Cyrl": {
+      "acc": true,
+      "acc_norm": true
+    },
+    "global_mmlu_full_uk": {
+      "acc": true
+    },
+    "squad_uk": {
+      "exact": true,
+      "f1": true,
+      "HasAns_exact": true,
+      "HasAns_f1": true,
+      "NoAns_exact": true,
+      "NoAns_f1": true,
+      "best_exact": true,
+      "best_f1": true
+    },
+    "flores_uk": {
+      "bleu": true
+    },
+    "long_flores_uk": {
+      "bleu": true
+    }
+  },
+  "n-samples": {
+    "zno_uk_geography": {
+      "original": 905,
+      "effective": 905
+    },
+    "zno_uk_history": {
+      "original": 1645,
+      "effective": 1645
+    },
+    "zno_uk_language_and_literature": {
+      "original": 2157,
+      "effective": 2157
+    },
+    "zno_uk_math": {
+      "original": 138,
+      "effective": 138
+    },
+    "gsm8k_uk": {
+      "original": 1319,
+      "effective": 1319
+    },
+    "ifeval_uk": {
+      "original": 541,
+      "effective": 541
+    },
+    "flores_en-uk": {
+      "original": 1012,
+      "effective": 1012
+    },
+    "long_flores_en-uk": {
+      "original": 281,
+      "effective": 281
+    },
+    "flores_uk-en": {
+      "original": 1012,
+      "effective": 1012
+    },
+    "long_flores_uk-en": {
+      "original": 281,
+      "effective": 281
+    },
+    "wmt_en_uk": {
+      "original": 998,
+      "effective": 998
+    },
+    "arc_challenge_uk": {
+      "original": 1172,
+      "effective": 1172
+    },
+    "triviaqa_uk": {
+      "original": 17944,
+      "effective": 17944
+    },
+    "xlsum_uk": {
+      "original": 5399,
+      "effective": 5399
+    },
+    "arc_easy_uk": {
+      "original": 2376,
+      "effective": 2376
+    },
+    "winogrande_uk": {
+      "original": 1267,
+      "effective": 1267
+    },
+    "belebele_ukr_Cyrl": {
+      "original": 900,
+      "effective": 900
+    },
+    "global_mmlu_full_uk": {
+      "original": 14042,
+      "effective": 14042
+    },
+    "squad_uk": {
+      "original": 11496,
+      "effective": 11496
+    }
+  },
+  "config": {
+    "model": "vllm",
+    "model_args": {
+      "pretrained": "google/gemma-4-26B-A4B-it",
+      "data_parallel_size": 2,
+      "tensor_parallel_size": 1,
+      "gpu_memory_utilization": 0.55,
+      "max_model_len": 32768,
+      "max_num_seqs": 1,
+      "max_num_batched_tokens": 32768,
+      "enforce_eager": true,
+      "add_bos_token": true,
+      "enable_thinking": true,
+      "trust_remote_code": true
+    },
+    "batch_size": "auto",
+    "batch_sizes": [],
+    "device": "cuda:0",
+    "use_cache": null,
+    "limit": null,
+    "bootstrap_iters": 100000,
+    "gen_kwargs": {},
+    "random_seed": 0,
+    "numpy_seed": 1234,
+    "torch_seed": 1234,
+    "fewshot_seed": 1234
+  },
+  "git_hash": "8c8c110",
+  "date": 1777843903.646594,
+  "pretty_env_info": "PyTorch version: 2.11.0+cu129\nIs debug build: False\nCUDA used to build PyTorch: 12.9\nROCM used to build PyTorch: N/A\n\nOS: NVIDIA DGX Server (x86_64)\nGCC version: (GCC) 11.4.1 20231218 (Red Hat 11.4.1-4)\nClang version: Could not collect\nCMake version: version 3.26.5\nLibc version: glibc-2.34\n\nPython version: 3.12.7 | packaged by Anaconda, Inc. | (main, Oct  4 2024, 13:27:36) [GCC 11.2.0] (64-bit runtime)\nPython platform: Linux-5.14.0-427.42.1.el9_4.x86_64-x86_64-with-glibc2.34\nIs CUDA available: True\nCUDA runtime version: Could not collect\nCUDA_MODULE_LOADING set to: \nGPU models and configuration: \nGPU 0: NVIDIA H200\nGPU 1: NVIDIA H200\n\nNvidia driver version: 565.57.01\ncuDNN version: Could not collect\nIs XPU available: False\nHIP runtime version: N/A\nMIOpen runtime version: N/A\nIs XNNPACK available: True\nCaching allocator config: {'PYTORCH_CUDA_ALLOC_CONF': 'expandable_segments:True'}\n\nCPU:\nArchitecture:                         x86_64\nCPU op-mode(s):                       32-bit, 64-bit\nAddress sizes:                        52 bits physical, 57 bits virtual\nByte Order:                           Little Endian\nCPU(s):                               224\nOn-line CPU(s) list:                  0-223\nVendor ID:                            GenuineIntel\nModel name:                           Intel(R) Xeon(R) Platinum 8480C\nCPU family:                           6\nModel:                                143\nThread(s) per core:                   2\nCore(s) per socket:                   56\nSocket(s):                            2\nStepping:                             8\nCPU(s) scaling MHz:                   99%\nCPU max MHz:                          3800.0000\nCPU min MHz:                          800.0000\nBogoMIPS:                             4000.00\nFlags:                                fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx pdpe1gb rdtscp lm constant_tsc art arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc cpuid aperfmperf tsc_known_freq pni pclmulqdq dtes64 monitor ds_cpl vmx smx est tm2 ssse3 sdbg fma cx16 xtpr pdcm pcid dca sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand lahf_lm abm 3dnowprefetch cpuid_fault epb cat_l3 cat_l2 cdp_l3 intel_ppin cdp_l2 ssbd mba ibrs ibpb stibp ibrs_enhanced tpr_shadow flexpriority ept vpid ept_ad fsgsbase tsc_adjust bmi1 avx2 smep bmi2 erms invpcid cqm rdt_a avx512f avx512dq rdseed adx smap avx512ifma clflushopt clwb intel_pt avx512cd sha_ni avx512bw avx512vl xsaveopt xsavec xgetbv1 xsaves cqm_llc cqm_occup_llc cqm_mbm_total cqm_mbm_local split_lock_detect avx_vnni avx512_bf16 wbnoinvd dtherm ida arat pln pts hwp hwp_act_window hwp_epp hwp_pkg_req vnmi avx512vbmi umip pku ospke waitpkg avx512_vbmi2 gfni vaes vpclmulqdq avx512_vnni avx512_bitalg tme avx512_vpopcntdq la57 rdpid bus_lock_detect cldemote movdiri movdir64b enqcmd fsrm md_clear serialize tsxldtrk pconfig arch_lbr ibt amx_bf16 avx512_fp16 amx_tile amx_int8 flush_l1d arch_capabilities\nVirtualization:                       VT-x\nL1d cache:                            5.3 MiB (112 instances)\nL1i cache:                            3.5 MiB (112 instances)\nL2 cache:                             224 MiB (112 instances)\nL3 cache:                             210 MiB (2 instances)\nNUMA node(s):                         2\nNUMA node0 CPU(s):                    0-55,112-167\nNUMA node1 CPU(s):                    56-111,168-223\nVulnerability Gather data sampling:   Not affected\nVulnerability Itlb multihit:          Not affected\nVulnerability L1tf:                   Not affected\nVulnerability Mds:                    Not affected\nVulnerability Meltdown:               Not affected\nVulnerability Mmio stale data:        Not affected\nVulnerability Reg file data sampling: Not affected\nVulnerability Retbleed:               Not affected\nVulnerability Spec rstack overflow:   Not affected\nVulnerability Spec store bypass:      Mitigation; Speculative Store Bypass disabled via prctl\nVulnerability Spectre v1:             Mitigation; usercopy/swapgs barriers and __user pointer sanitization\nVulnerability Spectre v2:             Mitigation; Enhanced / Automatic IBRS; IBPB conditional; RSB filling; PBRSB-eIBRS SW sequence; BHI BHI_DIS_S\nVulnerability Srbds:                  Not affected\nVulnerability Tsx async abort:        Not affected\n\nVersions of relevant libraries:\n[pip3] numpy==2.2.6\n[pip3] nvidia-cublas-cu12==12.9.1.4\n[pip3] nvidia-cuda-cupti-cu12==12.9.79\n[pip3] nvidia-cuda-nvrtc-cu12==12.9.86\n[pip3] nvidia-cuda-runtime-cu12==12.9.79\n[pip3] nvidia-cudnn-cu12==9.17.1.4\n[pip3] nvidia-cudnn-frontend==1.18.0\n[pip3] nvidia-cufft-cu12==11.4.1.4\n[pip3] nvidia-curand-cu12==10.3.10.19\n[pip3] nvidia-cusolver-cu12==11.7.5.82\n[pip3] nvidia-cusparse-cu12==12.5.10.65\n[pip3] nvidia-cusparselt-cu12==0.7.1\n[pip3] nvidia-nccl-cu12==2.28.9\n[pip3] nvidia-nvjitlink-cu12==12.9.86\n[pip3] nvidia-nvtx-cu12==12.9.79\n[pip3] torch==2.11.0+cu129\n[pip3] torch_c_dlpack_ext==0.1.5\n[pip3] torchaudio==2.11.0+cu129\n[pip3] torchvision==0.26.0+cu129\n[pip3] triton==3.6.0\n[conda] Could not collect",
+  "transformers_version": "5.5.0",
+  "lm_eval_version": "0.4.11",
+  "upper_git_hash": null,
+  "tokenizer_pad_token": [
+    "<pad>",
+    "0"
+  ],
+  "tokenizer_eos_token": [
+    "<eos>",
+    "1"
+  ],
+  "tokenizer_bos_token": [
+    "<bos>",
+    "2"
+  ],
+  "eot_token_id": 1,
+  "max_length": 32768,
+  "task_hashes": {
+    "squad_uk_cot_likelihood": "def3984ac5a810669523869824d3e7af6bc51d8f6fa7f33b2b7f9267cd68563f"
+  },
+  "model_source": "vllm",
+  "model_name": "google/gemma-4-26B-A4B-it (reasoning)",
+  "model_name_sanitized": "google__gemma-4-26B-A4B-it-reasoning",
+  "system_instruction": null,
+  "system_instruction_sha": null,
+  "fewshot_as_multiturn": null,
+  "chat_template": null,
+  "chat_template_sha": null,
+  "total_evaluation_time_seconds": "2475.159956113028",
+  "groups": {}
+}