Data: 10k Synthetic WandBot QA with Evaluations

Created on June 2|Last edited on June 2
Comment
Questions were created based off the docs using this notebook:
﻿https://github.com/wandb/wandbot/blob/main/examples/synthetic_wandbot_question_generation.ipynb﻿
Answers & Grades were created using this notebook:
﻿https://github.com/wandb/wandbot/blob/main/examples/synthetic_wandbot_answer_and_evaluation.ipynb﻿
﻿
﻿W&B Run link﻿
﻿
Questions & Answers with Feedback﻿
runs.summary["logs/qa_with_eval_final"]
 - 5 of 9463
request_timestamp
request_id
user
question
question_id
wandbot_answer
retrived_source_documents
answer_elapsed_time_s
synth_grader_feedback
grader_justification
prompt_tokens
completion_tokens
total_tokens
answer_cost_usd
successful_requests
answer_system_prompt_version
eval_elapsed_time_s
eval_cost_usd
eval_total_tokens
eval_tokens_for_source_docs
eval_prompt_tokens
eval_completion_tokens
is_source_documents_truncated
wandb_run_id
wandbot_model
wandbot_temperature
hyde_llm_temperature
eval_model
1
2
3
4
5
Run: synth_answer_generation_final1
﻿
﻿
3 WandBot System Prompt Template VariantsUsed for answering the question
﻿
Run set1
﻿
﻿
LLM Grader Evaluation Prompt Template and HyDE Prompt Template﻿
Run set1
﻿
﻿
Add a comment