Upload configuration file

config/fig1_all_bespoke_stratos.yaml

@@ -0,0 +1,125 @@
+operators:
+- id: load_hf
+  config:
+    type: hf_source
+    dataset: bespokelabs/Bespoke-Stratos-17k
+    split: train
+    trust_remote_code: True
+- id: unravel_conversations
+  config:
+    type: function
+    function: data_strategies.commons.convert_sharegpt_to_instruction_response
+    function_config:
+      input_sharegpt_column: conversations
+      output_instruction_column: original_problem
+      output_response_column: original_reasoning_trace
+      role_tag: from
+      user_tag: user
+      content_tag: value
+      assistant_tag: assistant
+      drop_non_output_columns: True
+  input_ids:
+  - load_hf
+- id: strip_problem
+  config:
+    type: function
+    function: data_strategies.commons.strip_column
+    function_config:
+      input_column: original_problem
+      strip_str: "Return your final response within \\boxed{}. "
+      output_column: problem
+  input_ids:
+  - unravel_conversations
+- id: decompose_reasoning_trace
+  config:
+    type: function
+    function: data_strategies.commons.convert_final_to_reasoning_trace
+    function_config:
+      input_column: original_reasoning_trace
+      output_reasoning_column: deepseek_reasoning
+      output_solution_column: deepseek_solution
+  input_ids:
+  - strip_problem
+- id: reassemble_reasoning_trace
+  config:
+    type: function
+    function: data_strategies.commons.convert_reasoning_trace_to_final
+    function_config:
+      reasoning_column: deepseek_reasoning
+      solution_column: deepseek_solution
+      output_column: final_reasoning_trace
+  input_ids:
+  - decompose_reasoning_trace
+- id: decontaminate
+  config:
+    type: cpu_function
+    sharded: false
+    num_cpus: 16
+    function:  data_strategies._A1_Source.utils.decontaminate_fuzzy_w_ngram
+    function_config:
+      column: problem
+      eval_datasets:
+        - HuggingFaceH4/MATH-500
+        - Maxwell-Jia/AIME_2024
+        - AI-MO/aimo-validation-amc
+        - livecodebench/code_generation_lite
+        - mlfoundations-dev/AIME2025_combined
+        - cais/hle
+        - open-r1/codeforces
+        - Idavidrein/gpqa
+        - daman1209arora/jeebench
+        - mlfoundations-dev/mmlu_pro_eval_full
+        - Qwen/CodeElo
+        - open-r1/ioi
+      eval_columns:
+        - problem
+        - Problem
+        - problem
+        - question_content
+        - question
+        - question
+        - description
+        - Question
+        - question
+        - prompt
+        - description
+        - statement
+      eval_splits:
+        - test
+        - train
+        - train
+        - test
+        - train
+        - test
+        - test
+        - train
+        - test
+        - test
+        - test
+        - test
+      eval_subsets:
+        Idavidrein/gpqa: gpqa_diamond
+      similarity_threshold: 75.0
+      ngram_size: 13
+  input_ids:
+  - decompose_reasoning_trace
+- id: convert_reasoning_trace_to_final
+  config:
+    type: function
+    function: data_strategies.commons.convert_reasoning_trace_to_final
+    function_config:
+      reasoning_column: deepseek_reasoning
+      solution_column: deepseek_solution
+      output_column: final_reasoning_trace
+  input_ids:
+  - decontaminate
+- id: convert_to_sharegpt
+  config:
+    type: function
+    function: data_strategies.commons.convert_instruction_response_to_sharegpt
+    function_config:
+      input_instruction_column: problem
+      input_response_column: final_reasoning_trace
+      output_sharegpt_column: conversations
+  input_ids:
+    - convert_reasoning_trace_to_final