wilsonramos/qwen3-4b-2507-agentic-questionnaire-V2-merged-hf

TEXT GENERATIONConcurrent Unit Cost:1Model Size:4BQuant:BF16Context Size:32kTool Calling:SupportedPublished:Jul 3, 2026Architecture:Transformer Featherless Exclusive Cold

This is a 4 billion parameter Qwen3-based instruction-tuned causal language model, developed by wilsonramos, with a merged LoRA adapter after Supervised Fine-Tuning (SFT). It is designed for agentic questionnaire tasks and supports a context length of 32768 tokens. The model is provided in Hugging Face/safetensors format, with GGUF weights available for local use with llama.cpp/Ollama.

Loading preview...

Model Overview

This model, wilsonramos/qwen3-4b-2507-agentic-questionnaire-V2-merged-hf, is an instruction-tuned variant of the Qwen/Qwen3-4B-Instruct-2507 base model. It incorporates a merged LoRA adapter following Supervised Fine-Tuning (SFT), enhancing its capabilities for specific applications.

Key Capabilities

  • Base Architecture: Built upon the Qwen3-4B-Instruct-2507 model.
  • Fine-Tuning: Features a merged LoRA adapter after SFT, indicating specialized training.
  • Context Length: Supports a substantial context window of 32768 tokens.
  • Format: Available in Hugging Face/safetensors format for direct integration.

Good For

  • Agentic Questionnaire Tasks: The model's fine-tuning suggests optimization for agent-based question-answering or interactive questionnaire scenarios.
  • Local Deployment: GGUF weights are available for efficient local inference using tools like llama.cpp and Ollama, making it suitable for edge or privacy-sensitive applications.