# Natural Language Generation

An instruction-tuned version based on Qwen2-VL-7B-Instruct, focusing on generating more detailed image descriptions, optimized for text-to-image dataset creation.

InternVideo2-Chat-8B is a video understanding model that combines a large language model (LLM) with video BLIP, built through a progressive learning scheme, capable of video semantic understanding and human-computer interaction.

UniChart is a pre-trained model for chart understanding and reasoning, capable of processing text, data, and visual elements in charts and generating natural language outputs.

LaMini-T5-61M is an instruction-following model based on the T5-small architecture, fine-tuned on the LaMini-instruction dataset with a parameter scale of 61M.

LaMini-Flan-T5-248M is a 248M parameter model fine-tuned on the LaMini-instruction dataset based on google/flan-t5-base, belonging to the LaMini-LM series, specifically designed for responding to natural language instructions.

BLOOMZ-P3 is a multilingual instruction fine-tuned model based on the BLOOM pre-trained model, capable of understanding and executing human instructions in dozens of languages.

MVP is a multitask supervised pretraining model based on the Transformer architecture, specifically designed for natural language generation tasks.

This is a paraphrase model based on the T5-base architecture, fine-tuned on the PAWS, MSRP, and Opinosis datasets, primarily used for text paraphrasing and rewriting tasks.

Unsupervised style transfer model based on GPT2 architecture, achieving text style conversion through paraphrase generation