You can deploy this model and start earning money today!

duhaihan/
qwen2.5-2b-lora-merged-2193

Name: qwen2.5-2b-lora-merged-2193
Availability: InStock

Views

No views yet

qwen3.5-2b-lora-merged-2193

This repository contains a merged Hugging Face model exported from LLaMA-Factory.

Summary

Base architecture: Qwen3.5-2B
Export type: merged full model checkpoint
Exported from: a LoRA fine-tuning checkpoint (checkpoint-2193)
Intended usage: multimodal yes/no classification and related reasoning on the target task used during fine-tuning

Files

Core files in this repository:

config.json
model.safetensors
tokenizer.json
tokenizer_config.json
processor_config.json
generation_config.json
chat_template.jinja

Load with Transformers

python

from transformers import AutoModelForCausalLM, AutoProcessor

repo_id = "YOUR_NAME/qwen3.5-2b-lora-merged-2193"
processor = AutoProcessor.from_pretrained(repo_id, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(repo_id, trust_remote_code=True)

Load with LLaMA-Factory

bash

llamafactory-cli chat \
  --model_name_or_path YOUR_NAME/qwen3.5-2b-lora-merged-2193 \
  --template qwen3_5 \
  --trust_remote_code true

Notes

This is a merged model directory, so inference should use model_name_or_path directly.
No separate adapter_name_or_path is needed after merging.
If you register this model in Merlin Model Store, use the Hugging Face repo id after upload.

duhaihan/qwen2.5-2b-lora-merged-2193