# Low-resource adaptation

A version fine-tuned on scientific datasets based on Qwen/Qwen2.5-7B-Instruct

Valkyrie-49B-v1 is a 49B-parameter large language model based on llama.cpp, offering multiple quantization versions suitable for different hardware configurations.

Smoothie Qwen is a lightweight adjustment tool that can smooth the token probabilities in Qwen and similar models and enhance the multilingual balanced generation ability.

A lightweight adjustment tool that can smooth the token probabilities in Qwen and similar models and enhance the balance ability of multilingual generation.

This model is a fine-tuned version of Llama-3.2-1B-Instruct using LoRA (Low-Rank Adaptation) weights, primarily designed to assist in answering various topic-related questions and providing relevant information.

Llamacpp imatrix quantized version based on PocketDoc/Dans-SakuraKaze-V1.0.0-12b, supporting multiple quantization types, suitable for text generation tasks.

Skyfall-36B-v2 is a 36B-parameter large language model, processed with llama.cpp's imatrix quantization, offering multiple quantized versions to accommodate different hardware requirements.

A text-to-image generation model based on LoRA technology, utilizing a diffuser architecture, capable of generating images with a Maniac Box style.

This is a Thai text-to-speech model based on the VITS architecture, specifically fine-tuned for male narration style.

A BERT model fine-tuned based on google-bert/bert-base-uncased, with unspecified specific use

NorMistral-11b-warm is a Norwegian large language model initialized based on Mistral-Nemo-Base-2407, supporting multiple Scandinavian languages and Sami.

FLUX.1-dev is a LoRA fine-tuned text-to-image diffusion model focused on generating high-quality artistic style images.

This is a LoRA weight model trained based on black-forest-labs/FLUX.1-dev, specifically designed for generating paper-cut style images.

A checkpoint model based on the MMS zero-shot project, capable of transcribing speech in nearly any language with only a small amount of unannotated text in the target language.

RoLlama3 is a series of Romanian language generation text models based on Meta Llama 3, this is the 8B parameter instruction fine-tuned version, specifically optimized for Romanian language tasks.

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.

Llamacpp imatrix quantized version of WizardLM-2-7B-abliterated, offering multiple quantization options for different hardware configurations.

The Llamacpp imatrix quantization version of bagel-8b-v1.0, providing multiple quantization files to meet different hardware and performance requirements.

Korean text generation model based on Llama 2 architecture, specialized for Korean natural language processing tasks

Thai instruction-optimized model fine-tuned from Typhoon-7B using Direct Preference Optimization (DPO) technology

SambaLingo-Turkish-Base is a bilingual (Turkish and English) model based on Llama-2-7b pre-training, adapted for Turkish by training on 42 billion tokens from the Turkish portion of the Cultura-X dataset.

A 13B-parameter Bengali-optimized large language model based on LLaMA-2 architecture, supporting causal language modeling in Bengali and English

This is an open-source model based on the Apache-2.0 license. Specific functionalities depend on the actual model

A Vietnamese-optimized dialogue model based on LLaMA2-7B, enhanced for Vietnamese language capabilities through continued self-supervised learning and supervised fine-tuning

This is a version fine-tuned in Korean based on the mistral-7b model

TinyLLama-v0 is a lightweight language model provided in GGUF format, suitable for text generation tasks.

ReMM-SLERP is a reconstructed version of the original MythoMax-L2-13B, utilizing SLERP technology to integrate multiple updated 13B parameter models, including Chronos-Beluga, Airoboros, and Huginn.

A medical Q&A model fine-tuned on the MedQuAD dataset based on Llama-2-7b-chat

A bilingual (Portuguese/English) text generation model fine-tuned from Falcon-7B using QLoRA method

Stable Diffusion model fine-tuned with Dreambooth technology for generating high-quality personalized images

This is a stable diffusion model trained with Dreambooth technology for text-to-image generation tasks.

This model is a fine-tuned speech intent classification model based on facebook/wav2vec2-conformer-rel-pos-large-960h-ft, achieving an accuracy of 58.33% on the evaluation set.

This model is an English-Japanese translation model fine-tuned based on Helsinki-NLP/opus-tatoeba-en-ja, optimized on an unknown dataset.

This model is a fine-tuned version of facebook/bart-base on the SQuAD dataset, suitable for question answering tasks.

A classification model fine-tuned based on bert-base-uncased for unknown classification tasks

This model is an automatic speech recognition (ASR) model based on the wav2vec2-large-lv60 architecture, fine-tuned on the GARY109/AI_LIGHT_DANCE - ONSET-SINGING dataset, primarily used for singing voice recognition tasks.

A fine-tuned version of the bert-base-uncased model, suitable for specific task processing

UniSpeech is a unified speech representation learning model that combines labeled and unlabeled data for pre-training, specifically fine-tuned for French.

A multilingual Q&A model based on XLM-RoBERTa-large, trained on the XQuAD dataset, supporting extractive Q&A tasks in 11 languages.

Pre-trained language model based on Transformer architecture, using MASK tokens for self-supervised learning