# Multi-dataset Fine-tuning

DeQA-Score-Mix3 is a no-reference image quality assessment model fine-tuned based on the MAGAer13/mplug-owl2-llama2-7b foundation model, demonstrating excellent performance across multiple datasets.

A Malayalam automatic speech recognition model fine-tuned based on facebook/w2v-bert-2.0, trained on multiple Malayalam datasets and using a trigram language model trained with the KENLM library.

DistilViT is an image caption generation model based on Vision Transformer (ViT) and distilled GPT-2, capable of converting images into textual descriptions.

Matter 7B is a fine-tuned model based on Mistral 7B, specifically designed for text generation tasks, supporting conversational interaction and function calling.

A Brazilian Portuguese abstract summarization generation model fine-tuned based on PTT5, supporting concise summary generation from texts such as news articles.

This model is an Indonesian text summarization model based on the PEGASUS architecture, fine-tuned on the Indosum, Liputan6, and XLSum datasets, suitable for news text summarization tasks.

This is a version fine-tuned on multiple summarization datasets based on the google/flan-t5-xxl model, aiming to create a multi-functional summarizer suitable for both academic and general scenarios.

A Kannada automatic speech recognition model fine-tuned based on openai/whisper-tiny, trained on multiple public Kannada ASR corpora

Malayalam automatic speech recognition model fine-tuned based on OpenAI Whisper-medium, trained on datasets including Common Voice 11.0

A Brazilian Portuguese text abstract summarization model fine-tuned based on PTT5

A model fine-tuned based on PTT5 for generating abstractive summaries of Brazilian Portuguese texts.

This is a Portuguese legal domain-specific model based on the BERTimbau large version, optimized for semantic textual similarity tasks.

This model is a fine-tuned version of facebook/wav2vec2-xls-r-1b on public Japanese speech datasets, supporting automatic speech recognition tasks in Japanese.

A Russian natural language inference model fine-tuned from DeepPavlov/rubert-base-cased, capable of predicting logical relationships (entailment/contradiction/neutral) between two texts

An automatic speech recognition model fine-tuned on multiple Italian speech datasets based on facebook/wav2vec2-xls-r-300m

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.

BERT model fine-tuned for Thai Q&A tasks based on WangchanBERTa, trained using Wikipedia and other datasets

A sentiment analysis model fine-tuned based on the CAMeLBERT Classical Arabic model, supporting Arabic text sentiment classification

A sentiment analysis model fine-tuned based on the CAMeLBERT dialect Arabic model, supporting Arabic text sentiment classification

Albert model fine-tuned for Persian named entity recognition task, supporting 10 entity types

A sentiment analysis model fine-tuned based on the CAMeLBERT Modern Standard Arabic model, supporting Arabic text sentiment classification