# Long text processing

MiniCPM4 is an efficient large - language model designed specifically for edge devices. Through systematic innovation, it achieves extreme efficiency improvements in four key dimensions: model architecture, training data, training algorithm, and inference system.

MiniCPM4 is an efficient large language model designed specifically for edge devices. Through systematic innovation, it achieves extreme efficiency improvements in four dimensions: model architecture, training data, training algorithm, and inference system. It can achieve over 5 times faster generation speed on edge chips.

Qwen3-4B is the latest version in the Qwen series of large language models with 4B parameters, supporting switching between reasoning and non-reasoning modes, excelling at inference, instruction following, and multilingual processing.

Qwen3-8B-AWQ is the latest generation of large language model with 8.2B parameters in the Tongyi Qianwen series, which uses AWQ 4-bit quantization technology to optimize inference efficiency. It supports the switching between thinking and non-thinking modes and has excellent reasoning, instruction-following, and intelligent agent capabilities.

Qwen3-4B is the latest large language model in the Qwen series, featuring the ability to switch thinking modes, powerful reasoning capabilities, excellent human preference alignment, outstanding agent capabilities, and multilingual support.

GGUF format model converted from VIDraft/Gemma-3-R1984-27B, supporting multilingual text generation

This is a cross-encoder model fine-tuned from ModernBERT-large, used to calculate the scores of text pairs, suitable for text re-ranking and semantic search tasks.

Croatian text generation model based on Mistral architecture, trained with Unsloth acceleration

Enhanced version of QwQ-32B, suitable for all usage scenarios, with outstanding reasoning and output capabilities.

This is a GGUF-format converted 8B-parameter language model specialized for fantasy fiction, supporting 128K context length.

Frigg-v2-8b-ACADEMIC-128K-Q4_K_M-GGUF is an 8B-parameter large language model in GGUF format, suitable for various text generation tasks.

A multimodal large model fine-tuned on the LongWriter-V-22K dataset based on Qwen2.5-VL-72B-Instruct

A 70B-parameter large language model based on the Llama3 architecture, specially optimized

Italian ModernBERT is a specialized version of ModernBERT for Italian language, pre-trained specifically on Italian text.

ModernBERT is a modern bidirectional encoder-only Transformer model (BERT-style), pre-trained on 2 trillion tokens of English and code data, with a native context length of up to 8,192 tokens.

KURE-v1 is an embedding model specifically optimized for Korean text retrieval, fine-tuned based on BAAI/bge-m3, and excels in Korean retrieval tasks.

A sentence embedding model fine-tuned on the Korean triplet dataset based on the Alibaba-NLP/gte-multilingual-base model for semantic similarity calculation

This is a sentence-transformers model fine-tuned on a Korean triplet dataset based on Alibaba NLP/gte-multilingual-base, designed for semantic textual similarity tasks.

A 3-billion parameter instruction-tuned language model developed by IBM, based on Granite-3.0 architecture, supporting multilingual tasks and commercial applications

This model is a large language model fine-tuned based on Qwen-2 72B Instruct, aiming to replicate the prose quality of the Claude 3 series of models and is the seventh version in the series of models.

A Korean sentence embedding model fine - tuned on Alibaba - NLP/gte - multilingual - base, supporting tasks such as semantic text similarity calculation and semantic search.

Jais series 1.3 billion parameter Arabic-English bilingual large language model, optimized for exceptional Arabic capabilities while maintaining strong English proficiency

Transformer encoder pretrained based on Megatron-LM, specifically designed for Japanese scenarios

Llama-2-7b-Ukrainian Version is a bilingual pre-trained model supporting Ukrainian and English, based on continued pre-training of Llama-2-7b using 5 billion tokens of data from CulturaX.

Turkish-Llama-8b-v0.1 is a fully fine-tuned Turkish text generation model based on LLaMA-3 8B, trained on a 30GB Turkish dataset.

Yi-1.5 is an upgraded version of the Yi model, demonstrating superior performance in programming, mathematics, reasoning, and instruction-following capabilities.

Yi-1.5 is an upgraded version of the Yi model, excelling in programming, mathematics, reasoning, and instruction-following capabilities while maintaining outstanding language understanding, commonsense reasoning, and reading comprehension.

This is an Estonian text summarization model based on the T5 architecture, specifically designed for summarizing stenographic records of the Estonian Parliament discussions.

360 Zhinao is an open-source large language model series developed by Qihoo 360, including base models and dialogue models with various context lengths, supporting both Chinese and English.

PrunaAI's compressed MPT-7B story-writing model, enabling efficient inference through llm-int8 technology

A model specifically designed for efficient zero-shot classification, supporting multilingual text classification tasks without requiring training data

A multilingual zero-shot text classification model trained based on BAAI/bge-m3-retromae, specifically designed for commercial-friendly scenarios

This is a base BERT model for computing compact embedding vectors of Russian sentences, developed based on cointegrated/rubert-tiny2, with the number of layers increased from 3 to 7.

Qra is a series of Polish-optimized large language models jointly developed by the Polish National Information Processing Institute and Gdańsk University of Technology, initialized based on TinyLlama-1.1B and trained on 90 billion Polish tokens

A Russian encoder model based on the RoPEBert architecture, trained using cloning methods, supports 512-token context, and surpasses the original ruBert-base model in quality

polka-1.1b is a bilingual (Polish and English) text generation model enhanced by continuing pre-training on 5.7 billion Polish tokens based on the TinyLlama-1.1B model.

An experimental fine-tuned model based on yi-34b-200k, suitable for creative writing, role-playing, and other tasks, without DPO stage applied.

Law LLM 13B is a specific domain foundation model developed based on LLaMA-1-13B, focusing on tasks in the legal domain.

A bilingual and cross-language reranking model optimized for RAG, supporting Chinese, English, Japanese, and Korean, providing explainable absolute scores

TituLM-1B-BN-V1 is a large language model specifically trained for generating and understanding Bengali text, extensively trained on a dataset containing 4.51 billion Bengali tokens.