Upload configuration file

config/d1_science_load_in_qwq_together.yaml

@@ -0,0 +1,72 @@
+operators:
+- id: load_in_math
+  config:
+    type: load_preexisting
+    framework_name: load_in_science_0d
+- id: fuzzy_dedup
+  config:
+    type: cpu_function
+    num_cpus: 80
+    function: data_strategies.DeepSeekR1.utils.fuzzy_dedup
+    function_config:
+      column: instruction_seed
+      similarity_threshold: 100.0
+  input_ids:
+    - load_in_math
+- id: sample_dataset_final
+  config:
+    type: function
+    function: data_strategies.commons.uniform_sample_fixed
+    function_config:
+      num_samples: 3_950
+  input_ids:
+    - fuzzy_dedup
+- id: add_row_ids
+  config:
+    type: function
+    function: data_strategies.commons.add_row_ids
+    function_config:
+      column_name: ms_id
+  input_ids:
+    - sample_dataset_final
+- id: duplicate
+  config:
+    type: function
+    function: data_strategies.TestTime.generator.duplicate_rows
+    function_config:
+      n_copies: 16
+  input_ids:
+    - add_row_ids
+- id: annotate
+  config:
+    type: completions
+    map: together_qwen
+    map_config:
+      input_problem_column: instruction_seed
+    model: Qwen/QwQ-32B
+    batch: false
+    backend: openai
+    require_all_responses: false
+    temperature: 0.999
+    backend_params:
+      base_url: "https://api.together.ai"
+      api_key: ab025288d5a46a9a3acf23b7af22b5d824f9623cb3d62e1e69debd48291be8cc
+      max_requests_per_minute: 250
+      max_tokens_per_minute: 500000
+      invalid_finish_reasons:
+      - content_filter
+    generation_params:
+      max_tokens: 32000
+  input_ids:
+  - duplicate
+
+- id: convert_to_sharegpt
+  config:
+    type: function
+    function: data_strategies.commons.convert_instruction_response_to_sharegpt
+    function_config:
+      input_instruction_column: instruction_seed
+      input_response_column: qwq_response
+      output_sharegpt_column: conversations
+  input_ids:
+  - annotate