# Low-Resource Optimization

Sailor2 is a multilingual large language model focused on the Southeast Asian region, supporting 15 languages including English, Chinese, and various Southeast Asian languages.

Seed-Coder-8B-Reasoning is a large language model with 8B parameters developed by ByteDance-Seed, focusing on code generation and reasoning tasks.

This is a sentence-transformers model fine-tuned from intfloat/multilingual-e5-small, specifically optimized for Korean retrieval tasks, mapping text to a 384-dimensional vector space.

A quantized version of the Qwen2.5-VL-72B-Instruct multimodal large language model, supporting image-text-to-text tasks, suitable for various quantization levels from high precision to low memory requirements.

A multilingual text-to-speech model fine-tuned based on Orpheus-3b, optimized for African low-resource languages, supporting voice cloning and emotion synthesis

EraX-Translator-V1.0 is a multilingual translation model that supports translation tasks in various languages, including Vietnamese, English, Russian, French, Chinese, and more. It is particularly suitable for low-resource languages and the translation of ancient Chinese and Buddhist scriptures.

Russian speech recognition model based on TIES fusion method, integrating two Whisper-large-v3 Russian variants, optimized for telephone call scenarios

A Turkish text-to-speech model based on Microsoft's SpeechT5 architecture, fine-tuned and optimized with specific datasets

IndicF5 is a near-human multilingual text-to-speech (TTS) model trained on 1,417 hours of high-quality speech data, supporting 11 Indian languages.

LogicLlama is a fine-tuned language model developed by Goppa AI, based on the 1-billion-parameter version of LLaMA, enhanced with knowledge injection and logical reasoning capabilities, suitable for edge device applications.

The GGUF quantized version of Wan Video is a text-to-video generation model suitable for older or low-end machines, supporting efficient inference via GGUF files.

A high-quality speech language model trained on a single GPU within 24 hours, fine-tuned based on Qwen2.5-0.5B, using Hubert tokens as vocabulary

This model is a fine-tuned speech recognition model based on openai/whisper-small for Sinhala language datasets, supporting Sinhala speech-to-text tasks.

ViTucano is a natively Portuguese pre-trained visual assistant that integrates visual understanding and language capabilities, suitable for multimodal tasks.

GGUF format model file of OpenBuddy/openbuddy-deepseek-10b-v17.1-4k, supporting multiple languages, suitable for text generation tasks.

A TTS model supporting Urdu text-to-speech and voice cloning

This is a GGUF format model converted from Chuxin-Embedding, primarily used for Chinese text retrieval tasks.

A compact multilingual translation model based on the T5 architecture, supporting mutual translation between Malay, Chinese, Tamil, and other languages, with special optimization for translating between Malaysian localized texts and standard texts.

This is a multi-target neural machine translation model specifically designed for translating from multiple Romance languages into German, English, French, Portuguese, and Spanish.

A small language model for low-resource African languages, supporting five African languages and English/French bilingual

A vision-language model capable of conversing about input images, supporting Japanese interaction

This is a Transformer-based machine translation model from Arabic to English, trained exclusively on HPLT data.

A 7-billion-parameter Bengali large language model optimized based on the LLaMA-2 architecture, supporting instruction-following tasks

A speech-to-text model fine-tuned based on openai/whisper-small, supporting audio translation and text generation tasks.

TowerInstruct-13B is a 13-billion-parameter language model obtained by fine-tuning TowerBase on the TowerBlocks supervised fine-tuning dataset, focusing on translation-related tasks

Qwen-1.8B model fine-tuned on English-Vietnamese bilingual data using ChatML prompt template format

Mixtral-8x7B is a pre-trained generative sparse mixture of experts large language model that outperforms Llama 2 70B on most benchmarks.

This model is a fine-tuned version of openai/whisper-small for speech recognition, supporting multiple languages and suitable for automatic speech recognition tasks.

A speech processing model based on the SpeechT5 architecture, fine-tuned for Taiwanese Mandarin

This is a machine translation model supporting bidirectional translation between 22 Indian languages and English, optimized using distillation techniques with a parameter size of 200M.

Heron GIT Japanese StableLM Base 7B is a vision-language model capable of conversing about input images.

Hindi automatic speech recognition (ASR) model based on Wav2Vec2 architecture, developed by AI4Bharat

A bidirectional translation model for English and Tamil based on the M2M100 architecture, released under the OpenRAIL license

This is a 1.3 billion parameter Yakut language model, fine-tuned based on mGPT-XL (1.3B), supporting Yakut text generation tasks.

A Dutch speech processing model fine-tuned on the common_voice_13_0 dataset based on the SpeechT5 architecture

A video caption generation model based on Timesformer and BERT architectures, capable of generating descriptive captions for video content.

An image captioning model based on Swin Transformer and AraGPT2 architecture, capable of generating textual descriptions for input images.

OpenCALM is a series of decoder-only language models pre-trained on Japanese datasets, developed by CyberAgent.

This repository contains a LLaMA-7b model internally fine-tuned on the Stanford Alpaca dataset, primarily for machine translation research.

A Korean cross-encoder model fine-tuned from albert-small-kor-v1 for sentence similarity calculation