blog articles

explores openai's integration of advanced image generation into chatgpt via the gpt-4o model, producing detailed and realistic images by providing descriptive prompts.

messy handwriting ocr comparison between aya-vision-8b and qwen2vl-ocr-2b presents an experimental evaluation of two vision-language models: aya-vision-8b and qwen2vl-ocr-2b.

comprehensive guide on customizing Google's SigLIP 2 vision-language model for specific image classification tasks. It covers the entire workflow.

comprehensive guide on enhancing the smollm language model using grpo, a reinforcement learning technique that evaluates model outputs in groups to optimize for reasoning, and formatting.

detailed comparison between openai's o3-mini and deepseek's r1 models, focusing on their architectural differences, performance metrics, and practical applications.

comprehensive guide to fine-tuning SmolLM2 models on domain-specific synthetic datasets. It outlines the entire process.