# Lightweight Fine-tuning

Smoothie Qwen is a lightweight fine-tuning tool that significantly improves the balance of multilingual generation by smoothing the token probability distribution of Qwen and similar models.

Text generation model fine-tuned based on google/gemma-2b-it, supporting tool call reasoning process

ThinkEdit is a lightweight weight editing method that identifies and edits a small number of attention heads to mitigate the issue of large language models generating overly short reasoning chains in inference tasks, thereby improving reasoning accuracy.

This is a DreamBooth LoRA weight trained based on Alpha-VLLM/Lumina-Image-2.0, specifically designed for generating yarn art-style images.

A LoRA fine-tuned version based on the Lightricks/LTX-Video model, specializing in text-to-video generation tasks

Qwen 2.5 0.5B is a compact yet powerful language model, fine-tuned based on Qwen/Qwen2.5-0.5B-Instruct, supporting multiple languages with performance close to the Llama 3.2 1B model.

phi3 is an open-source model based on the Apache-2.0 license, supporting English language, primarily used for summarization tasks.

Mamba is an efficient sequence model compatible with transformers, with 790 million parameters, suitable for causal language modeling tasks.

Mamba is a transformer-compatible sequence modeling model with efficient inference capabilities.

Mamba is an efficient language model based on the State Space Model (SSM) architecture, with 1.4B parameters, supporting text generation tasks

A 2.8 billion parameter language model based on the Mamba architecture, compatible with HuggingFace Transformers library

A Tarot card interpretation model fine-tuned based on TinyLlama-1.1B, capable of making predictions and interpretations based on Tarot cards.

BLIP2 is a vision-language model based on OPT-2.7B, focusing on visual question answering tasks. It can understand image content and answer related questions.

A large language model fine-tuned based on MythoMax-L2-13b, optimized through LoRA technology merging, suitable for creative text generation and dialogue tasks

Image caption generation model fine-tuned based on Salesforce/blip-image-captioning-base

This model is a fine-tuned sentence transformer based on a news dataset, specifically designed for semantic search and sentence similarity calculation.

A model fine-tuned based on distilroberta-base, with specific uses and training data not clearly stated

A Turkish text encoder fine-tuned from dbmdz/distilbert-base-turkish-cased, designed to work with CLIP's ViT-B/32 image encoder

A named entity recognition model based on ELECTRA-small architecture, fine-tuned on the wikiann dataset