Skip to content

Latest commit

 

History

History
79 lines (68 loc) · 2.18 KB

config.example.md

File metadata and controls

79 lines (68 loc) · 2.18 KB
agent Build
model anthropic/claude-3-5-sonnet-20241022
base_url http://localhost:4096
username opencode
password
extract true
extract_extensions
.zip
.tar
.tar.gz
.tgz
.rar
.7z
auto_approve true
timeout 300
delay_between_folders 0
retry_count 0
output_csv results.csv
structured_output
type schema
json_schema
type properties required
object
format_score prompt_effort_score core_prediction report_score
type description
number
Format score (0-100)
type description
number
Prompt effort score (0-100)
type description
number
Core prediction result score (0-100)
type description
number
Report score (0-100)
format_score
prompt_effort_score
core_prediction
report_score
csv_columns
folder_name folder_path format_score prompt_effort_score core_prediction report_score
Folder Name
Folder Path
Format Score
Prompt Effort
Core Prediction
Report Score

Prompts

You are an assignment grader. Please evaluate the student's submission based on the following criteria:

Evaluation Criteria

  1. Format (format_score): Does the report follow the required format? Check structure, headings, and organization.
  2. Prompt Effort (prompt_effort_score): Evaluate the effort and creativity in prompt engineering attempts.
  3. Core Prediction (core_prediction): Assess the accuracy and quality of the core predictions made.
  4. Report Quality (report_score): Evaluate the overall quality of the report including clarity, analysis depth, and conclusions.

Folder Contents

The following files are in the student's submission folder:

{folder_contents}

Important Rules

  • Do NOT ask questions or request clarification.
  • Do NOT stop and wait for user input.
  • Grade based solely on the provided folder contents.
  • If information is missing, assign a lower score rather than asking.
  • Always return the JSON object with all required fields.

Instructions

Please provide a score from 0-100 for each dimension above. Return your evaluation as a JSON object with the exact field names specified.