Update README.md

README.md

@@ -18,7 +18,6 @@ Addresses a critical and often cumbersome aspect of business operations—ensuri
   - Contract Compliance & Review
 
 
-
 ## Dataset Components
 | Dataset Component      | Description                                                  | Data Type    | Example                                                      |
 | ---------------------- | ------------------------------------------------------------ | ------------ | ------------------------------------------------------------ |
@@ -30,12 +29,42 @@ Addresses a critical and often cumbersome aspect of business operations—ensuri
 | Contract Compliance    | Key points from contracts and SLAs, including compliance with specified security requirements. | Unstructured | "SLA includes 99.9% uptime guarantee, penalties for data breaches" |
 
 
+### What to collect:
+
+#### Security Attestation
+- Extract key points from audit reports
+- compliance assessments
+- certification documents.
+
+
+#### BCP Effectiveness
+- Review BCP test reports
+- annual review documents
+- BCP metrics summaries.
+
+#### DRP Readiness
+- Gather information from DRP test results
+- DR drill reports
+- readiness assessments.
+
+
+#### Cybersecurity Maturity
+- Collect maturity assessment reports
+- cybersecurity program evaluations
+- structured maturity scores.
+
+
+#### Financial Health
+- Obtain financial statements
+- ratio analysis reports
+- financial performance summaries.
+
+
+#### Contract Compliance
+- Review contract documents
+- SLA agreements
+- compliance reports.
+
 
 
-## LLM Models to Fine-tune
 
-5. - [**Mixtral**: A high-quality Mixture of Experts (MoE) model, known for its strong performance on various benchmarks. It's praised for its cost/performance trade-offs, making it a potentially excellent choice for a wide range of NLP tasks including pattern recognition and compliance indicators in documents.](https://huggingface.co/mistralai/Mixtral-8x7B-Instruct-v0.1)
-6. - [**DeepSeek Coder**: Trained on two trillion code and natural language tokens, this model is designed for coding tasks but its extensive training on natural language tokens might also offer valuable insights into analyzing and understanding complex documents​.](https://huggingface.co/mistralai/Mixtral-8x7B-Instruct-v0.1)
-7. - [**Vicuna**: A general-use chat model that supports context sizes from 2K to 16K, which could be useful for extracting and interpreting information from lengthy documents​.](https://huggingface.co/liuhaotian/llava-v1.6-vicuna-13b)
-8. - [**Wizard Vicuna Uncensored**: With parameters ranging from 7B, 13B, to 30B, this model is based on Llama 2 uncensored and could provide robust capabilities for understanding nuanced text and generating insights​.](https://huggingface.co/TheBloke/Wizard-Vicuna-30B-Uncensored-GPTQ)
-9. - [**Dolphin Mistral**: An uncensored model excelling at coding tasks, which might be relevant for automated analysis of technical documents or contracts that include specific compliance codes or regulations​.](https://huggingface.co/TheBloke/laser-dolphin-mixtral-2x7b-dpo-GPTQ)