Add LC_code and make everything run again.

CF_Code.py

@@ -1,8 +1,10 @@
+import importlib
 import os
 
-from flows.flow_verse import load_class
-repository_id = os.environ.get("OpenAIChatAtomicFlow")
-OpenAIChatAtomicFlow = load_class(repository_id, "OpenAIChatAtomicFlow")
+from local.OpenAIChatAtomicFlow import OpenAIChatAtomicFlow
+
+# flow_module = importlib.import_module(os.getenv("OpenAIChatAtomicFlow_module"))
+# OpenAIChatAtomicFlow = getattr(flow_module, "OpenAIChatAtomicFlow")
 
 
 class CF_Code(OpenAIChatAtomicFlow):

CF_CodeCollab.yaml

@@ -22,9 +22,7 @@ output_keys:
   - "code"
 
 subflows_config:
-  - _target_: flows.flow_verse.instantiate_flow
-    repository_id: ${oc.env:CC_FLOWS}
-    class_name: CF_Code
+  - _target_: ${oc.env:CC_flows_module}.CF_Code..instantiate_from_default_config
     overrides:
       name: "CodeGenerator"
       model_name: "gpt-4"
@@ -64,9 +62,7 @@ subflows_config:
       output_keys:
         - "code"
         - "end_of_interaction"
-  - _target_: flows.flow_verse.instantiate_flow
-    repository_id: ${oc.env:CC_FLOWS}
-    class_name: CF_CodeCritic
+  - _target_: ${oc.env:CC_flows_module}.CF_CodeCritic.instantiate_from_default_config
     overrides:
       name: CodeCritic
       output_data_transformations:

CF_CodeCritic.py

@@ -1,9 +1,8 @@
 import os
+import importlib
 
-from flows.flow_verse import load_class
-repository_id = os.environ.get("OpenAIChatAtomicFlow")
-OpenAIChatAtomicFlow = load_class(repository_id, "OpenAIChatAtomicFlow")
-
+flow_module = importlib.import_module(os.getenv("OpenAIChatAtomicFlow_module"))
+OpenAIChatAtomicFlow = getattr(flow_module, "OpenAIChatAtomicFlow")
 
 
 class CF_CodeCritic(OpenAIChatAtomicFlow):

CF_CodeDebug.yaml

@@ -23,9 +23,7 @@ output_keys:
   - "code"
 
 subflows_config:
-  - _target_: flows.flow_verse.instantiate_flow
-    repository_id: ${oc.env:CC_FLOWS}
-    class_name: CF_Code
+  - _target_: ${oc.env:CC_flows_module}.CF_Code.instantiate_from_default_config
     overrides:
       name: "CodeGenerator"
       model_name: "gpt-4"
@@ -61,8 +59,6 @@ subflows_config:
       output_keys:
         - "code"
         - "end_of_interaction"
-  - _target_: flows.flow_verse.instantiate_flow
-    repository_id: ${oc.env:CC_FLOWS}
-    class_name: CF_CodeTesting
+  - _target_: ${oc.env:CC_flows_module}.CF_CodeTesting.instantiate_from_default_config
     overrides:
       name: "CodeTestingCritic"

CF_CodeReflect.yaml

@@ -22,9 +22,7 @@ output_keys:
   - "code"
 
 subflows_config:
-  - _target_: flows.flow_verse.instantiate_flow
-    repository_id: ${oc.env:CC_FLOWS}
-    class_name: CF_Code
+  - _target_: ${oc.env:CC_flows_module}.CF_Code.instantiate_from_default_config
     overrides:
       name: "CodeGenerator"
       model_name: "gpt-4"
@@ -48,6 +46,4 @@ subflows_config:
         - "code"
         - "end_of_interaction"
 
-  - _target_: flows.flow_verse.instantiate_flow
-    repository_id: ${oc.env:CC_FLOWS}
-    class_name: FixedReply_CodeReflect
+  - _target_: ${oc.env:CC_flows_module}.FixedReply_CodeReflect.instantiate_from_default_config

CF_CodeTesting.py

@@ -1,10 +1,10 @@
 from typing import Any, Dict
 
-from flows import utils
+from flows import logging
 from .src.evaluation import testing_utils_codeforces
 from .CodeTesting import CodeTesting
 
-log = utils.get_pylogger(__name__)
+log = logging.get_logger(__name__)
 
 # ToDo: Add a flags to control whether hidden, public or both tests should be used for evaluation
 

CF_CodeTesting.yaml

@@ -9,10 +9,10 @@ output_keys:
   - "all_tests_passed"
   - "testing_results_summary"
 output_data_transformations:
-  - _target_: CC_flows.src.data_transformations.CorrectnessFlag  # ToDo: This import style would not work if the flow is synced in the current implementation (the outer directory is a hash and not the name; figure out how to do the import robustly; e.g., using relative imports)
+  - _target_: ${oc.env:CC_flows_module}.src.data_transformations.CorrectnessFlag  # ToDo: This import style would not work if the flow is synced in the current implementation (the outer directory is a hash and not the name; figure out how to do the import robustly; e.g., using relative imports)
     input_key: "raw_response.public_tests_results"
     output_key: "all_tests_passed"
-  - _target_: CC_flows.src.data_transformations.TestingResultsSummaryGeneration
+  - _target_: ${oc.env:CC_flows_module}.src.data_transformations.TestingResultsSummaryGeneration
     output_key: "testing_results_summary"
 
     single_test_error_message: True

CodeTesting.py

@@ -1,10 +1,10 @@
 from copy import deepcopy
 from typing import Optional, Any, List, Dict
 
-from flows import utils
+from flows import logging
 from flows.base_flows.abstract import AtomicFlow
 
-log = utils.get_pylogger(__name__)
+log = logging.get_logger(__name__)
 
 
 class CodeTesting(AtomicFlow):

LC_Code.py

@@ -0,0 +1,8 @@
+import os
+
+from local.OpenAIChatAtomicFlow import OpenAIChatAtomicFlow
+
+
+class LC_Code(OpenAIChatAtomicFlow):
+    def __init__(self, **kwargs):
+        super().__init__(**kwargs)

LC_Code.yaml

@@ -0,0 +1,84 @@
+name: "Code_Flow"
+verbose: True
+description: "ToDO: add description"
+
+model_name: "gpt-4"
+generation_parameters:
+  n: 1
+  max_tokens: 3000
+  temperature: 0.3
+
+  model_kwargs:
+    top_p: 0.2
+    frequency_penalty: 0
+    presence_penalty: 0
+
+system_message_prompt_template:
+  _target_: langchain.PromptTemplate
+  template: |2-
+    Your goal is to provide executable Python code that solves a coding interview problem. The code should correctly handle all corner cases in order to pass the hidden test cases, which are used to evaluate the correctness of the solution.
+
+    The user will specify the problem by providing you with:
+      - the problem statement
+      - example test cases
+      - the constraints of the problem
+
+    The user will provide you with a task and an output format that you will strictly follow.
+  input_variables: []
+  template_format: jinja2
+
+human_message_prompt_template:
+  _target_: langchain.PromptTemplate
+  template: "{{query}}"
+  input_variables:
+    - "query"
+  template_format: jinja2
+
+query_message_prompt_template:
+  _target_: langchain.PromptTemplate
+  template: |2-
+    # Problem statement
+    {{problem_description}}
+
+    {{io_description}}
+
+    # Constraints
+    {{constraints}}
+
+
+    Return Python code that solves the problem. The code should extend the following stub:
+    ```python
+    {{python_stub}}
+    ```
+    without changing the method signatures.
+    Reply in the following format:
+    ```python
+    {{code_placeholder}}
+    ```
+  input_variables:
+    - "problem_description"
+    - "io_description"
+    - "constraints"
+    - "python_stub"
+  partial_variables:
+    code_placeholder: "{{python_code}}"
+  template_format: jinja2
+
+input_data_transformations: []
+input_keys:
+    - "problem_description"
+    - "io_description"
+    - "constraints"
+    - "python_stub"
+
+output_data_transformations:
+  - _target_: flows.data_transformations.RegexFirstOccurrenceExtractor
+    regex: '(?<=```python)([\s\S]*?)(?=```)'
+    regex_fallback: '(?<=```)([\s\S]*?)(?=```)'
+    input_key: "raw_response"
+    output_key: "code"
+    strip: True
+    assert_unique: True
+    verbose: True
+output_keys:
+  - "code"

__init__.py

@@ -1,5 +1,16 @@
+import os
+from flows import flow_verse
+
+dependencies = [
+    {"url": os.getenv("OpenAIChatAtomicFlow_repo"), "revision": "main"},
+    # {"url": os.getenv("CC_flows_repo"), "revision": "main"}
+]
+flow_verse.sync_dependencies(dependencies)
+
 # cf-code
 from .CF_Code import CF_Code
+# lc-code
+from .LC_Code import LC_Code
 
 # cf-code_reflect
 from .FixedReply_CodeReflect import FixedReply_CodeReflect
@@ -10,30 +21,30 @@ from .CF_CodeCritic import CF_CodeCritic
 from .CF_CodeCollab import CF_CodeCollab
 
 # cf-plan-code (and cf-plan_oracle-code)
-from .CF_Plan import CF_Plan
-from .CF_CodeWithPlan import CF_CodeWithPlan
-from .CF_Plan_Code import CF_Plan_Code
-
-# cf-plan_reflect-code
-from .FixedReply_PlanReflect import FixedReply_PlanReflect
-from .CF_PlanReflect import CF_PlanReflect
-from .CF_PlanReflect_Code import CF_PlanReflect_Code
-
-# cf-plan_collab-code
-from .CF_PlanCritic import CF_PlanCritic
-from .CF_PlanCollab import CF_PlanCollab
-from .CF_PlanCollab_Code import CF_PlanCollab_Code
-
+# from .CF_Plan import CF_Plan
+# from .CF_CodeWithPlan import CF_CodeWithPlan
+# from .CF_Plan_Code import CF_Plan_Code
+#
+# # cf-plan_reflect-code
+# from .FixedReply_PlanReflect import FixedReply_PlanReflect
+# from .CF_PlanReflect import CF_PlanReflect
+# from .CF_PlanReflect_Code import CF_PlanReflect_Code
+#
+# # cf-plan_collab-code
+# from .CF_PlanCritic import CF_PlanCritic
+# from .CF_PlanCollab import CF_PlanCollab
+# from .CF_PlanCollab_Code import CF_PlanCollab_Code
+#
 # cf-code_debug
 from .CF_CodeTesting import CF_CodeTesting
 from .CF_CodeDebug import CF_CodeDebug
 
-# cf-code_debug_collab
-from .CF_CodeCriticWrongAttempt import CF_CodeCriticWrongAttempt
-from .CF_CodeDebugCritic import CF_CodeDebugCritic
-from .CF_CodeDebugCollab import CF_CodeDebugCollab
-
-# cf-plan_oracle-code_debug_collab
-from .CF_CodeCriticWrongAttemptWithPlan import CF_CodeCriticWrongAttemptWithPlan
-from .CF_CodeDebugCriticWithPlan import CF_CodeDebugCriticWithPlan
-from .CF_CodeDebugCollabWithPlan import CF_CodeDebugCollabWithPlan
+# # cf-code_debug_collab
+# from .CF_CodeCriticWrongAttempt import CF_CodeCriticWrongAttempt
+# from .CF_CodeDebugCritic import CF_CodeDebugCritic
+# from .CF_CodeDebugCollab import CF_CodeDebugCollab
+#
+# # cf-plan_oracle-code_debug_collab
+# from .CF_CodeCriticWrongAttemptWithPlan import CF_CodeCriticWrongAttemptWithPlan
+# from .CF_CodeDebugCriticWithPlan import CF_CodeDebugCriticWithPlan
+# from .CF_CodeDebugCollabWithPlan import CF_CodeDebugCollabWithPlan

src/evaluation/testing_utils_codeforces.py

@@ -19,9 +19,9 @@ import threading
 
 from src.datasets.schema import assert_test_format_codeforces
 
-import logging
+from flows import logging
 
-log = logging.getLogger()
+log = logging.get_logger(__name__)
 lock = threading.Lock()