Qwen3-VL-2B-Instruct
mhlc-training-qwen3vl-qwen3_vl_2b_instruct_hard_mixed_sources_120k
Multi Head Latent Control Training Data - Qwen3-VL 2B Instruct hard Mixed Sources 120k
Dataset Description
This repository contains verified training data for the Multi Head Latent Control paper release. It is part of the Multi Head Latent Control training data Hugging Face collection.
Paper
https://arxiv.org/abs/2607.14277
Code
https://github.com/Amirhosein-gh98/Multi-Head-Latent-Control
Dataset Summary
Field… See the full description on the dataset page: https://huggingface.co/datasets/AmirhoseinGH/mhlc-training-qwen3vl-qwen3_vl_2b_instruct_hard_mixed_sources_120k.Qwen3-VL-2B-Instruct-Post-Metadata-dataset-npy
Qwen-VL UGC Embeddings & Post Metadata
A lightweight dataset of 9,597 precomputed visual vectors (.npy) and structured post metadata for fast SFT/LoRA fine-tuning of compact Qwen-VL models (e.g., 2B) without loading the full vision encoder.
Overview
Samples: 9,597 pairs (~30 MB total)
Visual Features: Extracted .npy patch embeddings from the Qwen-VL Vision Tower
Labels: Distilled titles, detailed descriptions, and tags generated by a larger Qwen-VL teacher… See the full description on the dataset page: https://huggingface.co/datasets/rollenso/Qwen3-VL-2B-Instruct-Post-Metadata-dataset-npy.qwen3-vl_2b-instruct_continuous_resultsqwen3-vl_2b-instruct_hierarchical_standardizedqwen3-vl_2b-instruct_discreteqwen3-vl_2b-instruct_continuous
