to-do.md

1. groq + openai
2. hf dataset + local dir + csv upload
3. generate graphs
4. generate report - tab 1/ tab 2
5. Patent Acceptance Prediction
6. Patentibility check based on Abstract + Claims
7. Abstract + Claim Summarization
8. Abstract-to-Claim and Claim-to-Abstract Generation

Questions:

• What is the trend of patent filings over time?
• What is the distribution of patents categorized as 'ACCEPTED', 'REJECTED', or 'PENDING'?
• What is the average time difference between the filing date and the patent issue date?
• What are the most common CPC labels in the dataset?
• Which examiners have reviewed the most patents?
• How does the acceptance or rejection rate vary across years?
• What is the average length of abstracts for accepted versus rejected patents?
• Which IPC labels are most associated with accepted patents?
• Are there any seasonal trends in the filing of patents?
• What is the distribution of CPC or IPC labels for patents still marked as pending?

Observation:

1. Fails when the size of the dataset is large.
2. Sometimes fails even when the dataset size is small; unreliable!