Add files using upload-large-folder tool

.gitattributes

@@ -57,3 +57,7 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 # Video files - compressed
 *.mp4 filter=lfs diff=lfs merge=lfs -text
 *.webm filter=lfs diff=lfs merge=lfs -text
+_workspace_outputs_simpo-skyT1-out_livecodebench_easy_48175f/results.json filter=lfs diff=lfs merge=lfs -text
+_workspace_outputs_simpo-skyT1-out_livecodebench_medium_03c871/results.json filter=lfs diff=lfs merge=lfs -text
+_workspace_outputs_simpo-skyT1-out_livecodebench_hard_a16b19/results.json filter=lfs diff=lfs merge=lfs -text
+_workspace_outputs_simpo-skyT1-out_livecodebench_c1143d/results.json filter=lfs diff=lfs merge=lfs -text

_workspace_outputs_simpo-skyT1-out_aime25_1_b806e9/summary.json

@@ -0,0 +1,55 @@
+{
+    "configuration": {
+        "task": {
+            "name": "aime25_1",
+            "config": {
+                "handler": "aime",
+                "dataset_path": "opencompass/AIME2025",
+                "dataset_subset": "AIME2025-I",
+                "dataset_split": "test",
+                "dataset_kwargs": {},
+                "question_key": "question",
+                "answer_key": "answer",
+                "templating_parameters": {
+                    "template": "{prompt}\nReturn your final response within \\boxed{{}}"
+                },
+                "preprocess_config": {}
+            },
+            "start": 0,
+            "end": -1
+        },
+        "model": {
+            "name": "/workspace/outputs/simpo-skyT1-out",
+            "config": {
+                "model_id": "/workspace/outputs/simpo-skyT1-out",
+                "name": "simpo-skyT1-out",
+                "system_prompt": "Your role as an assistant involves thoroughly exploring questions through a systematic long thinking process before providing the final precise and accurate solutions. This requires engaging in a comprehensive cycle of analysis, summarizing, exploration, reassessment, reflection, backtracing, and iteration to develop well-considered thinking process. Please structure your response into two main sections: Thought and Solution. In the Thought section, detail your reasoning process using the specified format: <|begin_of_thought|> {thought with steps separated with '\n\n'} <|end_of_thought|> Each step should include detailed considerations such as analisying questions, summarizing relevant findings, brainstorming new ideas, verifying the accuracy of the current steps, refining any errors, and revisiting previous steps. In the Solution section, based on various attempts, explorations, and reflections from the Thought section, systematically present the final solution that you deem correct. The solution should remain a logical, accurate, concise expression style and detail necessary step needed to reach the conclusion, formatted as follows: <|begin_of_solution|> {final formatted, precise, and clear solution} <|end_of_solution|> Now, try to solve the following question through the above guidelines:",
+                "user_template": null,
+                "assistant_prefill": null
+            }
+        },
+        "backend": {
+            "name": "vllm",
+            "backend_args": {
+                "tensor_parallel_size": 8
+            }
+        },
+        "sampling_params": {
+            "temperature": 0.6,
+            "top_p": 0.95,
+            "max_tokens": 32768,
+            "n": 8
+        }
+    },
+    "total_completion_tokens": 1479999,
+    "avg_completion_tokens": 12333.325,
+    "total_prompt_tokens": 5547,
+    "avg_prompt_tokens": 46.225,
+    "accuracy": 0.4083,
+    "pass_at_k": {
+        "k=8": 60.0,
+        "k=4": 55.048,
+        "k=2": 48.571,
+        "k=1": 40.833
+    }
+}

_workspace_outputs_simpo-skyT1-out_aime25_2_2aba16/summary.json

@@ -0,0 +1,55 @@
+{
+    "configuration": {
+        "task": {
+            "name": "aime25_2",
+            "config": {
+                "handler": "aime",
+                "dataset_path": "opencompass/AIME2025",
+                "dataset_subset": "AIME2025-II",
+                "dataset_split": "test",
+                "dataset_kwargs": {},
+                "question_key": "question",
+                "answer_key": "answer",
+                "templating_parameters": {
+                    "template": "{prompt}\nReturn your final response within \\boxed{{}}"
+                },
+                "preprocess_config": {}
+            },
+            "start": 0,
+            "end": -1
+        },
+        "model": {
+            "name": "/workspace/outputs/simpo-skyT1-out",
+            "config": {
+                "model_id": "/workspace/outputs/simpo-skyT1-out",
+                "name": "simpo-skyT1-out",
+                "system_prompt": "Your role as an assistant involves thoroughly exploring questions through a systematic long thinking process before providing the final precise and accurate solutions. This requires engaging in a comprehensive cycle of analysis, summarizing, exploration, reassessment, reflection, backtracing, and iteration to develop well-considered thinking process. Please structure your response into two main sections: Thought and Solution. In the Thought section, detail your reasoning process using the specified format: <|begin_of_thought|> {thought with steps separated with '\n\n'} <|end_of_thought|> Each step should include detailed considerations such as analisying questions, summarizing relevant findings, brainstorming new ideas, verifying the accuracy of the current steps, refining any errors, and revisiting previous steps. In the Solution section, based on various attempts, explorations, and reflections from the Thought section, systematically present the final solution that you deem correct. The solution should remain a logical, accurate, concise expression style and detail necessary step needed to reach the conclusion, formatted as follows: <|begin_of_solution|> {final formatted, precise, and clear solution} <|end_of_solution|> Now, try to solve the following question through the above guidelines:",
+                "user_template": null,
+                "assistant_prefill": null
+            }
+        },
+        "backend": {
+            "name": "vllm",
+            "backend_args": {
+                "tensor_parallel_size": 8
+            }
+        },
+        "sampling_params": {
+            "temperature": 0.6,
+            "top_p": 0.95,
+            "max_tokens": 32768,
+            "n": 8
+        }
+    },
+    "total_completion_tokens": 1777971,
+    "avg_completion_tokens": 14816.425,
+    "total_prompt_tokens": 5834,
+    "avg_prompt_tokens": 48.617,
+    "accuracy": 0.2333,
+    "pass_at_k": {
+        "k=8": 46.667,
+        "k=4": 42.19,
+        "k=2": 33.81,
+        "k=1": 23.333
+    }
+}

_workspace_outputs_simpo-skyT1-out_amc23_8ce04f/summary.json

@@ -0,0 +1,57 @@
+{
+    "configuration": {
+        "task": {
+            "name": "amc23",
+            "config": {
+                "handler": "amc23",
+                "dataset_path": "AI-MO/aimo-validation-amc",
+                "dataset_subset": null,
+                "dataset_split": "train",
+                "dataset_kwargs": {
+                    "trust_remote_code": true
+                },
+                "question_key": "problem",
+                "answer_key": "answer",
+                "templating_parameters": {
+                    "template": "Return your final response within \\boxed{{}}. {problem}"
+                },
+                "preprocess_config": {}
+            },
+            "start": 0,
+            "end": -1
+        },
+        "model": {
+            "name": "/workspace/outputs/simpo-skyT1-out",
+            "config": {
+                "model_id": "/workspace/outputs/simpo-skyT1-out",
+                "name": "simpo-skyT1-out",
+                "system_prompt": "Your role as an assistant involves thoroughly exploring questions through a systematic long thinking process before providing the final precise and accurate solutions. This requires engaging in a comprehensive cycle of analysis, summarizing, exploration, reassessment, reflection, backtracing, and iteration to develop well-considered thinking process. Please structure your response into two main sections: Thought and Solution. In the Thought section, detail your reasoning process using the specified format: <|begin_of_thought|> {thought with steps separated with '\n\n'} <|end_of_thought|> Each step should include detailed considerations such as analisying questions, summarizing relevant findings, brainstorming new ideas, verifying the accuracy of the current steps, refining any errors, and revisiting previous steps. In the Solution section, based on various attempts, explorations, and reflections from the Thought section, systematically present the final solution that you deem correct. The solution should remain a logical, accurate, concise expression style and detail necessary step needed to reach the conclusion, formatted as follows: <|begin_of_solution|> {final formatted, precise, and clear solution} <|end_of_solution|> Now, try to solve the following question through the above guidelines:",
+                "user_template": null,
+                "assistant_prefill": null
+            }
+        },
+        "backend": {
+            "name": "vllm",
+            "backend_args": {
+                "tensor_parallel_size": 8
+            }
+        },
+        "sampling_params": {
+            "temperature": 0.6,
+            "top_p": 0.95,
+            "max_tokens": 32768,
+            "n": 8
+        }
+    },
+    "total_completion_tokens": 1818082,
+    "avg_completion_tokens": 5681.506,
+    "total_prompt_tokens": 13822,
+    "avg_prompt_tokens": 43.194,
+    "accuracy": 0.8094,
+    "pass_at_k": {
+        "k=8": 95.0,
+        "k=4": 93.893,
+        "k=2": 89.554,
+        "k=1": 80.938
+    }
+}

_workspace_outputs_simpo-skyT1-out_livecodebench_c1143d/results.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ebf547d1189234a642e2690843b9822acbb1b2415f026670747168f29b4cee6f
+size 8496745232

_workspace_outputs_simpo-skyT1-out_livecodebench_c1143d/summary.json

@@ -0,0 +1,59 @@
+{
+    "configuration": {
+        "task": {
+            "name": "livecodebench",
+            "config": {
+                "handler": "livecodebench",
+                "dataset_path": "livecodebench/code_generation_lite",
+                "dataset_subset": null,
+                "dataset_split": "test",
+                "dataset_kwargs": {
+                    "version_tag": "release_v2",
+                    "trust_remote_code": true
+                },
+                "question_key": "task_id",
+                "answer_key": null,
+                "templating_parameters": {
+                    "stdin_template": "Generate an executable Python function generated from the given prompt. The function should take stdin as input and print the output. Simply call the function after the definition. {prompt}",
+                    "non_stdin_template": "Generate an executable Python function generated from the given prompt. Return the function body without invoking it at the final solution. {prompt}"
+                },
+                "preprocess_config": {}
+            },
+            "start": 0,
+            "end": -1
+        },
+        "model": {
+            "name": "/workspace/outputs/simpo-skyT1-out",
+            "config": {
+                "model_id": "/workspace/outputs/simpo-skyT1-out",
+                "name": "simpo-skyT1-out",
+                "system_prompt": "Your role as an assistant involves thoroughly exploring questions through a systematic long thinking process before providing the final precise and accurate solutions. This requires engaging in a comprehensive cycle of analysis, summarizing, exploration, reassessment, reflection, backtracing, and iteration to develop well-considered thinking process. Please structure your response into two main sections: Thought and Solution. In the Thought section, detail your reasoning process using the specified format: <|begin_of_thought|> {thought with steps separated with '\n\n'} <|end_of_thought|> Each step should include detailed considerations such as analisying questions, summarizing relevant findings, brainstorming new ideas, verifying the accuracy of the current steps, refining any errors, and revisiting previous steps. In the Solution section, based on various attempts, explorations, and reflections from the Thought section, systematically present the final solution that you deem correct. The solution should remain a logical, accurate, concise expression style and detail necessary step needed to reach the conclusion, formatted as follows: <|begin_of_solution|> {final formatted, precise, and clear solution} <|end_of_solution|> Now, try to solve the following question through the above guidelines:",
+                "user_template": null,
+                "assistant_prefill": null
+            }
+        },
+        "backend": {
+            "name": "vllm",
+            "backend_args": {
+                "tensor_parallel_size": 8
+            }
+        },
+        "sampling_params": {
+            "temperature": 0.6,
+            "top_p": 0.95,
+            "max_tokens": 32768,
+            "n": 8
+        }
+    },
+    "total_completion_tokens": 38923519,
+    "avg_completion_tokens": 9521.409,
+    "total_prompt_tokens": 382012,
+    "avg_prompt_tokens": 93.447,
+    "accuracy": 0.5453,
+    "pass_at_k": {
+        "k=8": 80.822,
+        "k=4": 74.428,
+        "k=2": 65.963,
+        "k=1": 54.525
+    }
+}

_workspace_outputs_simpo-skyT1-out_livecodebench_easy_48175f/results.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:92bfda27178ff1d5728f008995f27bb8863c4efda80354807b146d396caf8e87
+size 222832915

_workspace_outputs_simpo-skyT1-out_livecodebench_easy_48175f/summary.json

@@ -0,0 +1,61 @@
+{
+    "configuration": {
+        "task": {
+            "name": "livecodebench_easy",
+            "config": {
+                "handler": "livecodebench",
+                "dataset_path": "livecodebench/code_generation_lite",
+                "dataset_subset": null,
+                "dataset_split": "test",
+                "dataset_kwargs": {
+                    "version_tag": "release_v2",
+                    "trust_remote_code": true
+                },
+                "question_key": "task_id",
+                "answer_key": null,
+                "templating_parameters": {
+                    "stdin_template": "Generate an executable Python function generated from the given prompt. The function should take stdin as input and print the output. Simply call the function after the definition. {prompt}",
+                    "non_stdin_template": "Generate an executable Python function generated from the given prompt. Return the function body without invoking it at the final solution. {prompt}"
+                },
+                "preprocess_config": {
+                    "difficulty": "easy"
+                }
+            },
+            "start": 0,
+            "end": -1
+        },
+        "model": {
+            "name": "/workspace/outputs/simpo-skyT1-out",
+            "config": {
+                "model_id": "/workspace/outputs/simpo-skyT1-out",
+                "name": "simpo-skyT1-out",
+                "system_prompt": "Your role as an assistant involves thoroughly exploring questions through a systematic long thinking process before providing the final precise and accurate solutions. This requires engaging in a comprehensive cycle of analysis, summarizing, exploration, reassessment, reflection, backtracing, and iteration to develop well-considered thinking process. Please structure your response into two main sections: Thought and Solution. In the Thought section, detail your reasoning process using the specified format: <|begin_of_thought|> {thought with steps separated with '\n\n'} <|end_of_thought|> Each step should include detailed considerations such as analisying questions, summarizing relevant findings, brainstorming new ideas, verifying the accuracy of the current steps, refining any errors, and revisiting previous steps. In the Solution section, based on various attempts, explorations, and reflections from the Thought section, systematically present the final solution that you deem correct. The solution should remain a logical, accurate, concise expression style and detail necessary step needed to reach the conclusion, formatted as follows: <|begin_of_solution|> {final formatted, precise, and clear solution} <|end_of_solution|> Now, try to solve the following question through the above guidelines:",
+                "user_template": null,
+                "assistant_prefill": null
+            }
+        },
+        "backend": {
+            "name": "vllm",
+            "backend_args": {
+                "tensor_parallel_size": 8
+            }
+        },
+        "sampling_params": {
+            "temperature": 0.6,
+            "top_p": 0.95,
+            "max_tokens": 32768,
+            "n": 8
+        }
+    },
+    "total_completion_tokens": 6096452,
+    "avg_completion_tokens": 4187.124,
+    "total_prompt_tokens": 115539,
+    "avg_prompt_tokens": 79.354,
+    "accuracy": 0.8537,
+    "pass_at_k": {
+        "k=8": 99.451,
+        "k=4": 97.692,
+        "k=2": 94.113,
+        "k=1": 85.371
+    }
+}

_workspace_outputs_simpo-skyT1-out_livecodebench_hard_a16b19/results.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:42faa57e3a211fcd7b85cf657b288546e4d0352511f48e3cd7d63eed5b92b151
+size 4239481969

_workspace_outputs_simpo-skyT1-out_livecodebench_hard_a16b19/summary.json

@@ -0,0 +1,61 @@
+{
+    "configuration": {
+        "task": {
+            "name": "livecodebench_hard",
+            "config": {
+                "handler": "livecodebench",
+                "dataset_path": "livecodebench/code_generation_lite",
+                "dataset_subset": null,
+                "dataset_split": "test",
+                "dataset_kwargs": {
+                    "version_tag": "release_v2",
+                    "trust_remote_code": true
+                },
+                "question_key": "task_id",
+                "answer_key": null,
+                "templating_parameters": {
+                    "stdin_template": "Generate an executable Python function generated from the given prompt. The function should take stdin as input and print the output. Simply call the function after the definition. {prompt}",
+                    "non_stdin_template": "Generate an executable Python function generated from the given prompt. Return the function body without invoking it at the final solution. {prompt}"
+                },
+                "preprocess_config": {
+                    "difficulty": "hard"
+                }
+            },
+            "start": 0,
+            "end": -1
+        },
+        "model": {
+            "name": "/workspace/outputs/simpo-skyT1-out",
+            "config": {
+                "model_id": "/workspace/outputs/simpo-skyT1-out",
+                "name": "simpo-skyT1-out",
+                "system_prompt": "Your role as an assistant involves thoroughly exploring questions through a systematic long thinking process before providing the final precise and accurate solutions. This requires engaging in a comprehensive cycle of analysis, summarizing, exploration, reassessment, reflection, backtracing, and iteration to develop well-considered thinking process. Please structure your response into two main sections: Thought and Solution. In the Thought section, detail your reasoning process using the specified format: <|begin_of_thought|> {thought with steps separated with '\n\n'} <|end_of_thought|> Each step should include detailed considerations such as analisying questions, summarizing relevant findings, brainstorming new ideas, verifying the accuracy of the current steps, refining any errors, and revisiting previous steps. In the Solution section, based on various attempts, explorations, and reflections from the Thought section, systematically present the final solution that you deem correct. The solution should remain a logical, accurate, concise expression style and detail necessary step needed to reach the conclusion, formatted as follows: <|begin_of_solution|> {final formatted, precise, and clear solution} <|end_of_solution|> Now, try to solve the following question through the above guidelines:",
+                "user_template": null,
+                "assistant_prefill": null
+            }
+        },
+        "backend": {
+            "name": "vllm",
+            "backend_args": {
+                "tensor_parallel_size": 8
+            }
+        },
+        "sampling_params": {
+            "temperature": 0.6,
+            "top_p": 0.95,
+            "max_tokens": 32768,
+            "n": 8
+        }
+    },
+    "total_completion_tokens": 16255767,
+    "avg_completion_tokens": 16520.088,
+    "total_prompt_tokens": 112101,
+    "avg_prompt_tokens": 113.924,
+    "accuracy": 0.1301,
+    "pass_at_k": {
+        "k=8": 39.024,
+        "k=4": 29.338,
+        "k=2": 20.557,
+        "k=1": 13.008
+    }
+}

_workspace_outputs_simpo-skyT1-out_livecodebench_medium_03c871/results.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a024716bd243a4a96ce94896a27c508621a3f6ec918e36e9a2073c40121200eb
+size 4037789565

_workspace_outputs_simpo-skyT1-out_livecodebench_medium_03c871/summary.json

@@ -0,0 +1,61 @@
+{
+    "configuration": {
+        "task": {
+            "name": "livecodebench_medium",
+            "config": {
+                "handler": "livecodebench",
+                "dataset_path": "livecodebench/code_generation_lite",
+                "dataset_subset": null,
+                "dataset_split": "test",
+                "dataset_kwargs": {
+                    "version_tag": "release_v2",
+                    "trust_remote_code": true
+                },
+                "question_key": "task_id",
+                "answer_key": null,
+                "templating_parameters": {
+                    "stdin_template": "Generate an executable Python function generated from the given prompt. The function should take stdin as input and print the output. Simply call the function after the definition. {prompt}",
+                    "non_stdin_template": "Generate an executable Python function generated from the given prompt. Return the function body without invoking it at the final solution. {prompt}"
+                },
+                "preprocess_config": {
+                    "difficulty": "medium"
+                }
+            },
+            "start": 0,
+            "end": -1
+        },
+        "model": {
+            "name": "/workspace/outputs/simpo-skyT1-out",
+            "config": {
+                "model_id": "/workspace/outputs/simpo-skyT1-out",
+                "name": "simpo-skyT1-out",
+                "system_prompt": "Your role as an assistant involves thoroughly exploring questions through a systematic long thinking process before providing the final precise and accurate solutions. This requires engaging in a comprehensive cycle of analysis, summarizing, exploration, reassessment, reflection, backtracing, and iteration to develop well-considered thinking process. Please structure your response into two main sections: Thought and Solution. In the Thought section, detail your reasoning process using the specified format: <|begin_of_thought|> {thought with steps separated with '\n\n'} <|end_of_thought|> Each step should include detailed considerations such as analisying questions, summarizing relevant findings, brainstorming new ideas, verifying the accuracy of the current steps, refining any errors, and revisiting previous steps. In the Solution section, based on various attempts, explorations, and reflections from the Thought section, systematically present the final solution that you deem correct. The solution should remain a logical, accurate, concise expression style and detail necessary step needed to reach the conclusion, formatted as follows: <|begin_of_solution|> {final formatted, precise, and clear solution} <|end_of_solution|> Now, try to solve the following question through the above guidelines:",
+                "user_template": null,
+                "assistant_prefill": null
+            }
+        },
+        "backend": {
+            "name": "vllm",
+            "backend_args": {
+                "tensor_parallel_size": 8
+            }
+        },
+        "sampling_params": {
+            "temperature": 0.6,
+            "top_p": 0.95,
+            "max_tokens": 32768,
+            "n": 8
+        }
+    },
+    "total_completion_tokens": 17383231,
+    "avg_completion_tokens": 10548.077,
+    "total_prompt_tokens": 154372,
+    "avg_prompt_tokens": 93.672,
+    "accuracy": 0.5091,
+    "pass_at_k": {
+        "k=8": 83.01,
+        "k=4": 76.227,
+        "k=2": 65.603,
+        "k=1": 50.91
+    }
+}