Garance Colomer

This experience was intense and challenging, but with my team, we challenged ourselves by trying to add our version and vision of the mechanistic interpretability of large language models.

We proved the neural pathway coverage of the model by using its features.

My role in the team was AI Scientist, where I contributed my experience in model thinking.

See our research paper in my portfolio: [email protected]

Learned: TensorFlow, Pytorch, TensorBoard, Django

Fine-tuning and Model Evaluation Performance and computational cost optimization (PEFT) Multi-Modal, OCR, Classifier, LLM Django as backend Visualization and analysis of ongoing training with TensorBoard

Learned: HF, FastAPI, Ragas

Deployment of LLMs with Hugging Face RAG systems, integrating embedding, retrieval, and NLP Prompts engineering Python and Docker for application programming and containerization Model evaluation with Ragas for benchmarking