# Multi-task Fine-tuning

A semantic text similarity and embedding model specifically designed for Persian, capable of mapping sentences into a dense vector space where semantically similar texts are positioned close to each other.

OLMo 2 1B SFT is a supervised fine-tuned version of the OLMo-2-0425-1B model, trained on the Tulu 3 dataset, designed to achieve state-of-the-art performance across multiple tasks.

TRL is a reinforcement learning library based on the Transformer architecture for training and fine-tuning language models.

PaliGemma 2 is a vision-language model based on Gemma 2, supporting image and text inputs with text generation output, suitable for various vision-language tasks.

CamemBERTav2 is a French language model pretrained on 275 billion French text tokens, utilizing the DebertaV2 architecture, and excels in multiple French NLP tasks.

An instruction-following model fine-tuned based on T5-base, specifically designed to understand and execute complex instructions

PaliGemma is a lightweight vision-language model developed by Google, built upon the SigLIP vision model and Gemma language model, supporting multilingual text generation and visual understanding tasks.

TookaBERT is a family of encoder models trained on Persian, including base and large versions, suitable for various natural language processing tasks.

OpenELM is a set of open-source efficient language models that employ a hierarchical scaling strategy to optimize parameter allocation and improve model accuracy. It includes four parameter scales: 270M, 450M, 1.1B, and 3B, offering both pre-trained and instruction-tuned versions.

A multi-domain language model fine-tuned on the Hyperion-v2.0 dataset, excelling in scientific reasoning and complex task processing.

A large German language model developed based on Llama2 70B, specializing in German business scenarios

A 1.8B parameter base language model trained by H2O.ai, based on an improved Llama 2 architecture with 16K context length support

This is a fine-tuned large language model for mental health prediction using online text data.

A 13-billion-parameter pre-trained model based on the LLaMa architecture, capable of translation, programming, text classification, information extraction, summarization, copywriting, commonsense Q&A, and mathematical calculations

Sentence embedding model based on MPNet architecture, mapping text to a 768-dimensional vector space, suitable for semantic search and text similarity tasks

GLM-2B is a general-purpose language model pre-trained with autoregressive blank filling objectives, supporting various natural language understanding and generation tasks.

This model is a question-answering model based on the DeBERTa architecture, fine-tuned on the SQuAD1, AQA, and NewsQA datasets.

A Chinese sentiment analysis model based on the MegatronBert architecture, fine-tuned on multiple sentiment analysis tasks

A fine-tuned version based on the Chinese RoBERTa-wwm-ext-large model on multiple natural language inference task datasets

A QA and NER model fine-tuned on the conll2003 dataset based on distilbert-base-uncased-squad2

TAPAS is a table question answering model based on the BERT architecture, pre-trained in a self-supervised manner on Wikipedia table data, supporting natural language question answering on table content

RobBERT is the state-of-the-art Dutch BERT model, pretrained on a large scale and adaptable to various text tasks through fine-tuning.

Data2Vec NLP Base is a natural language processing model converted from the fairseq framework, suitable for tasks such as text classification.

SEW-D-mid is a speech pre-training model developed by ASAPP Research, focusing on automatic speech recognition tasks, achieving a good balance between performance and efficiency.

TAPAS is a Transformer-based table question answering model, pre-trained in a self-supervised manner on English Wikipedia table data and fine-tuned with supervision on the WikiSQL dataset.

T5 1.1 is Google's improved text-to-text transfer model, utilizing GEGLU activation function and optimized architecture, focusing on unsupervised pre-training

A lightweight DistilBERT model pre-trained on a large-scale Italian corpus, suitable for various natural language processing tasks.

TAPAS is a BERT-based Transformer model specifically designed for table question answering tasks. It is pre-trained in a self-supervised manner on English Wikipedia table data and supports weakly supervised table parsing.

This is a sentence transformer model based on allenai/biomed_roberta_base, specifically fine-tuned for sentence similarity tasks, capable of mapping text to a 768-dimensional vector space.

BART is a Transformer model combining a bidirectional encoder and an autoregressive decoder, suitable for text generation and understanding tasks.

MiniLM is a compact and efficient pre-trained language model, compressed through deep self-attention distillation technology, suitable for language understanding and generation tasks.

ELECTRA is an efficient self-supervised language representation learning method that replaces traditional generative pretraining with discriminative pretraining, significantly improving computational efficiency.

TAPAS is a BERT-like Transformer model designed for table-based question answering tasks. It is pre-trained in a self-supervised manner on English Wikipedia table corpora and fine-tuned on the WikiSQL dataset.

This model is a small version of TAPAS, specifically fine-tuned on the WikiTable Questions dataset for table-based question answering tasks.

This model is a medium-sized table question answering model based on TAPAS architecture, fine-tuned on WikiTable Questions dataset, suitable for table data QA tasks.

A lightweight BERT model for self-supervised learning of Persian language representations

TAPAS is a tiny Transformer model optimized for table question answering tasks, achieving table comprehension capabilities through intermediate pretraining and chained multi-dataset fine-tuning

A Japanese question generation and answer extraction model fine-tuned based on MT5-small, supporting generating questions or extracting answers from given text.

A RoBERTa BASE model pretrained on Thai Wikipedia, suitable for Thai text processing tasks

TAPAS is a Transformer-based table question answering model, pre-trained on Wikipedia table data through self-supervised learning and fine-tuned on datasets like WTQ.