# Instruction fine-tuning

The quantized version of Tencent Hunyuan A13B Instruction Model, which uses technical means to improve operational efficiency while ensuring performance.

Gemma 3n-E2B-it is a lightweight open-source multimodal model launched by Google, built on the same technology as Gemini and optimized for low-resource devices.

Gemma 3n is a lightweight and state-of-the-art open-source multimodal model family launched by Google. It is built on the same research and technology as the Gemini model and supports text, audio, and visual inputs.

Fanar-1-9B-Instruct is a powerful Arabic-English large language model developed by the Qatar Computing Research Institute (QCRI). It supports Modern Standard Arabic and multiple Arabic dialects, and is aligned with Islamic values and Arab culture.

Seed-Coder-8B-Instruct is a powerful open-source code model with features such as model-centricity, transparency, and high performance, and it performs excellently in various coding tasks.

This model is a text generation model fine-tuned based on Qwen/Qwen2.5-1.5B-Instruct. It is trained using TRL and provides strong support for text generation tasks.

Spatial-LLaVA-7B is a multimodal model fine-tuned based on the LLaVA model, focusing on improving the ability of spatial relationship reasoning and suitable for multimodal research and chatbot development.

This is a quantized version based on the Gryphe/Pantheon-Proto-RP-1.8-30B-A3B model, using llama.cpp for quantization, suitable for role-playing and text generation tasks.

This model is the FP8 quantized version of Bielik-4.5B-v3.0-Instruct, utilizing AutoFP8 technology to quantize weights and activations into FP8 data type, reducing approximately 50% of disk space and GPU memory requirements.

This is an FP8 dynamic quantization version based on the Bielik-1.5B-v3.0-Instruct model, adapted for vLLM or SGLang inference frameworks. It uses AutoFP8 quantization technology to reduce parameter bytes from 16-bit to 8-bit, significantly lowering disk space and GPU VRAM requirements.

This is a 1.5B parameter instruction fine-tuned model for Polish, developed based on the SpeakLeash Bielik series, suitable for text generation tasks.

A lightweight instruction fine-tuned language model optimized for mobile deployment

An instruction fine-tuned model optimized based on the Mistral architecture, suitable for zero-shot classification tasks

Llama-Krikri-8B-Instruct is a Greek instruction-tuned large language model developed by the Athena Research Center, based on Llama-3.1-8B, with enhanced multitasking capabilities in Greek and English.

An instruction fine-tuned model based on the Llama-3-3B architecture, supporting tasks such as dialogue, reasoning, and role-playing, suitable for general artificial intelligence assistance scenarios.

MKLLM-7B is an open-source large language model for the Macedonian language, built through continued pre-training on mixed Macedonian and English texts based on the Mistral-7B-v0.1 model.

The Gemma-2-Llama-Swallow series of models are multilingual large models constructed through continuous pre-training based on Gemma-2, with a particular enhancement in Japanese ability.

Turkish-Gemma-9b-v0.1 is a Turkish text generation model developed based on Gemma-2-9b, optimized through continued pretraining, supervised fine-tuning (SFT), direct preference optimization (DPO), and model merging techniques.

Bielik-4.5B-v3-Instruct is a 4.6 billion parameter Polish generative text model, fine-tuned based on Bielik-4.5B-v3, demonstrating exceptional Polish language comprehension and processing capabilities.

An 8B-parameter language model merged using the SLERP method, combining the strengths of DeepSeek-R1 and Nemotron-8B models

Falcon-E-1B-Instruct is an efficient language model based on a 1.58-bit architecture, optimized for edge devices with low memory footprint and high performance.

Latxa is a 70B-parameter Basque language large language model based on Llama-3.1, fine-tuned with instructions and FP8 quantization, specifically optimized for Basque

This is a multi-model fusion version based on the Gemma-2-9B series models, merged using the DARE TIES method, integrating the strengths of multiple Gemma variants.

A 1.5B parameter instruction fine-tuned model designed for Gensyn RL Swarm, supporting local fine-tuning via peer-to-peer reinforcement learning

A Persian-specific model fine-tuned based on the Gemma 3 architecture, utilizing QLoRA 4-bit quantization technology, suitable for running on ordinary hardware.

Gemma is a lightweight open model series launched by Google, built on the same research technology as Gemini. The llamafile version is packaged as an executable file by Mozilla for easy use on multiple platforms.

Latxa 3.1 70B Instruct is an instruction-tuned version based on Llama-3.1 (Instruct), specifically optimized for Basque language, demonstrating excellent performance on multiple Basque benchmarks.

GaMS-9B-Instruct is a Slovenian large language model developed based on Google Gemma 2 series, supporting Slovenian, English, and some South Slavic languages.

Lughaat-1.0-8B-Instruct is a large Urdu language model based on the Llama 3.1 8B architecture, specifically trained on the largest Urdu dataset and excels in Urdu language tasks.

Quantized version based on mlabonne/gemma-3-12b-it-abliterated model, using llama.cpp for imatrix quantization, suitable for text generation tasks.

Qwen2-VL-72B-Instruct is a multimodal vision-language model that supports interaction between images and text, suitable for complex vision-language tasks.

This is a 12B-parameter language model merged using the mergekit tool, combining the strengths of multiple specialized models with a focus on reasoning and psychology-related tasks.

The Gemma-3-4b Pure Text Model is a pure text version converted from google/gemma-3-4b-it, removing the visual component and focusing on text generation tasks.

The GGUF quantized version of the Gemma 3 1B model, suitable for text generation tasks.

A version fine-tuned on the OpenR1-Math-220k dataset based on the Qwen2.5-7B-Instruct model, focusing on mathematics-related tasks.

Instella is a series of 3-billion-parameter open-source language models developed by AMD, trained on AMD Instinct™ MI300X GPUs, outperforming other fully open-source models of the same scale.

A fine-tuned reasoning model based on LLaMA-3B, enhanced with reasoning capabilities and multilingual processing support

An intermediate model during the Tantum merging process, created by merging multiple 24B-parameter Mistral and Llama3 variants, suitable for role-playing and text generation tasks.

Latxa is a Basque language instruction-tuned large language model based on the Llama-3.1 architecture, specifically optimized for Basque, demonstrating excellent performance across multiple benchmarks.

OLMo 2 32B is the largest 32B-parameter model in the open language model series released by the Allen Institute for AI (AI2). It is open-sourced under the Apache 2.0 license and supports English language processing.