Pinned Loading

1. llm-reliability-and-consistency-evaluation llm-reliability-and-consistency-evaluation Public

   Evaluating LLMs' factual accuracy, consistency, and robustness to prompt variations using diverse response and question formats.

   HTML 1

2. TruthEval TruthEval Public

   A curated collection of challenging statements on sensitive topics for LLM benchmarking. Designed to distinguish LLMs' abilities from their stochastic nature.

   2

3. GPT3-Reliability-Check GPT3-Reliability-Check Public

   Systematic analysis of the responses of GPT-3 to different categories of statements and the potential vulnerabilities to simple prompting changes. We analyze what confuses GPT-3: how the model resp…

   Jupyter Notebook 5 2

4. KG-LM-Integration KG-LM-Integration Public

   A research project to combine information from Knowledge Graphs (KG) into Large Languge Models (LLM) to improve LLM factual accuracy while retaining fluency. The intention is to de-bias and prevent…

   Jupyter Notebook 3

5. Authorship-Attribution-using-Transfer-Learning Authorship-Attribution-using-Transfer-Learning Public

   Transfer learning for authorship attribution with unsupervised training of a language model that teaches a model the working and structure of Bangla language, followed by authorship attribution spe…

   Jupyter Notebook 1

6. Wikidata-WDQS-Analysis Wikidata-WDQS-Analysis Public

   Analysis on Wikidata and Wikidata Query Service to help figure out ways to scale the service. Repository contains analysis code, written articles on the findings and visualizations.

   Jupyter Notebook 1