Focused on the Text analysis with SLMs in the financial domain.
: Bridging the gap of general LLMs with customized SLMs
Selected Projects
React-vite dashboard for analyzing Financial News with entity graphs, topic modeling, and sentiment analysis across sectors and companies.
LoRA fine-tuning and local inference pipeline for Llama 3.2 3B on Apple Silicon via `mlx-lm`, targeting financial sentiment classification.
LoRA fine-tuning and local inference pipeline for Llama 3.2 3B on Apple Silicon via `mlx-lm`, achieving coreference resolution on the financial texts.
Articles
Requantization of a fine-tuned and fused model with [mlx-lm]
Re-quantizing a fine-tuned model after dequantizing for performance enhancement.
Model fusing (combining fine-tuned adapters to the base model) with [mlx-lm] - unintended behavior
Current limitations of model fusing (mlx_lm.fuse)
Fine-tuning open-source LLMs for Multi-entity Sentiment Analysis task using [mlx-lm]
Enhancing multi-entity Sentiment Analysis task with fine-tuning LLMs
Fine-tuning open-source LLMs for Coreference Resolution task using [mlx-lm]
Modernizing Coreference Resolution task with fine-tuning LLMs
Using Ollama API endpoints
A simple instruction to Ollama endpoints without depending on the Python SDK
Connect
Feel free to contact me via