From 7ee0d9d4121324c7b550d6b95b44f74652f47be8 Mon Sep 17 00:00:00 2001 From: Tyrin Todd Date: Thu, 23 Apr 2026 23:07:10 -0700 Subject: [PATCH] server-inference push seekbench --- README.md | 5 ++++- config.json | 56 ++++++++++++++++++++++++++++++++++++++++++++++++++++ program.json | 48 ++++++++++++++++++++++++++++++++++++++++++++ 3 files changed, 108 insertions(+), 1 deletion(-) create mode 100644 config.json create mode 100644 program.json diff --git a/README.md b/README.md index c8c874f..5ef1aa8 100644 --- a/README.md +++ b/README.md @@ -1,2 +1,5 @@ -# seekbench +--- +{} +--- +# seekbench \ No newline at end of file diff --git a/config.json b/config.json new file mode 100644 index 0000000..93734f5 --- /dev/null +++ b/config.json @@ -0,0 +1,56 @@ +{ + "model": null, + "signature": { + "description": "Given question + traces, classify whether the current reasoning is grounded. As you reason, note any uncertainties, ambiguities, and gaps \u2014 both in how the instructions apply to the task at hand and in whether you have the knowledge needed to solve it. Be honest in your reasoning when you are unsure about your answer.", + "properties": { + "question": { + "__dspy_field_type": "input", + "desc": "Original question.", + "prefix": "Question:", + "title": "Question", + "type": "string" + }, + "previous_traces": { + "__dspy_field_type": "input", + "desc": "All previous traces before the current reasoning trace.", + "prefix": "Previous Traces:", + "title": "Previous Traces", + "type": "string" + }, + "current_trace": { + "__dspy_field_type": "input", + "desc": "Current reasoning trace.", + "prefix": "Current Trace:", + "title": "Current Trace", + "type": "string" + }, + "reasoning": { + "__dspy_field_type": "output", + "desc": "Step-by-step reasoning, including uncertainty, ambiguity, or uncertainty in your thought process when relevant.", + "prefix": "Reasoning:", + "title": "Reasoning", + "type": "string" + }, + "groundness": { + "__dspy_field_type": "output", + "desc": "Groundness label for current reasoning trace.", + "enum": [ + "Grounded", + "Not Grounded" + ], + "prefix": "Groundness:", + "title": "Groundness", + "type": "string" + } + }, + "required": [ + "question", + "previous_traces", + "current_trace", + "reasoning", + "groundness" + ], + "title": "SeekBenchReasoningSig", + "type": "object" + } +} \ No newline at end of file diff --git a/program.json b/program.json new file mode 100644 index 0000000..4907b86 --- /dev/null +++ b/program.json @@ -0,0 +1,48 @@ +{ + "traces": [], + "train": [], + "demos": [], + "signature": { + "instructions": "Given question + traces, classify whether the current reasoning is grounded. As you reason, note any uncertainties, ambiguities, and gaps \u2014 both in how the instructions apply to the task at hand and in whether you have the knowledge needed to solve it. Be honest in your reasoning when you are unsure about your answer.", + "fields": [ + { + "prefix": "Question:", + "description": "Original question." + }, + { + "prefix": "Previous Traces:", + "description": "All previous traces before the current reasoning trace." + }, + { + "prefix": "Current Trace:", + "description": "Current reasoning trace." + }, + { + "prefix": "Reasoning:", + "description": "Step-by-step reasoning, including uncertainty, ambiguity, or uncertainty in your thought process when relevant." + }, + { + "prefix": "Groundness:", + "description": "Groundness label for current reasoning trace." + } + ] + }, + "lm": { + "model": "together_ai/Qwen/Qwen2.5-7B-Instruct-Turbo", + "model_type": "chat", + "cache": true, + "num_retries": 3, + "finetuning_model": null, + "launch_kwargs": {}, + "train_kwargs": {}, + "temperature": null, + "max_tokens": null + }, + "metadata": { + "dependency_versions": { + "python": "3.11", + "dspy": "3.1.3", + "cloudpickle": "3.1" + } + } +} \ No newline at end of file