These GGUF releases are optimized deployment variants of Frame2KG models. They are provided for practical inference use and may not exactly match the original weights, checkpoints, or evaluation configuration reported in the Frame2KG paper.

Model Details

Family: Qwen3.5
Size: 4b
Output format: JSON Frame2KG graph output
Base model: Qwen/Qwen3.5-4B
Model type: qwen3_5
Architecture: Qwen3_5ForConditionalGeneration
Fine-tuning method: PEFT LoRA
LoRA rank: 8
LoRA alpha: 16
Trainable added token count: 0

Files

File	Size	Notes
`Frame2KG-Qwen3.5-4b-JSON.f16.gguf`	7.85 GB	model weights
`Frame2KG-Qwen3.5-4b-JSON.IQ4_XS.gguf`	2.36 GB	model weights
`Frame2KG-Qwen3.5-4b-JSON.Q2_K.gguf`	1.78 GB	model weights
`Frame2KG-Qwen3.5-4b-JSON.Q3_K_L.gguf`	2.26 GB	model weights
`Frame2KG-Qwen3.5-4b-JSON.Q3_K_M.gguf`	2.11 GB	model weights
`Frame2KG-Qwen3.5-4b-JSON.Q3_K_S.gguf`	1.93 GB	model weights
`Frame2KG-Qwen3.5-4b-JSON.Q4_K_M.gguf`	2.52 GB	model weights
`Frame2KG-Qwen3.5-4b-JSON.Q4_K_S.gguf`	2.39 GB	model weights
`Frame2KG-Qwen3.5-4b-JSON.Q5_K_M.gguf`	2.86 GB	model weights
`Frame2KG-Qwen3.5-4b-JSON.Q5_K_S.gguf`	2.78 GB	model weights
`Frame2KG-Qwen3.5-4b-JSON.Q6_K.gguf`	3.23 GB	model weights
`Frame2KG-Qwen3.5-4b-JSON.Q8_0.gguf`	4.17 GB	model weights
`mmproj-Frame2KG-Qwen3.5-4b-JSON.f16.gguf`	641.27 MB	multimodal projector
`mmproj-Frame2KG-Qwen3.5-4b-JSON.Q8_0.gguf`	349.90 MB	multimodal projector

Usage Notes

These files are intended for llama.cpp-compatible GGUF runtimes.

Scope

Frame2KG models are intended to convert image or frame content into structured graph-style outputs. The exact output format depends on the variant:

JSON variants target JSON-formatted Frame2KG output.
CT variants target compressed Frame2KG graph tokens.

Citation

If you use this model in your work, please cite the paper:

bibtex

@inproceedings{watson2026frame2kg,
  title = {Frame2KG: A Benchmark and Evaluation Toolkit for Interpretable Frame-to-Graph Generation},
  author = {Watson, Lewis N. and Strathearn, Carl and Mitchell, Kenny and Yu, Yanchao},
  booktitle = {Proceedings of the Fifteenth Language Resources and Evaluation Conference (LREC 2026)},
  month = {May},
  year = {2026},
  pages = {10912--10926},
  address = {Palma, Mallorca, Spain},
  publisher = {European Language Resources Association (ELRA)},
  editor = {Piperidis, Stelios and Bel, Núria and van den Heuvel, Henk and Ide, Nancy and Krek, Simon and Toral, Antonio},
  doi = {10.63317/4ys6kofrzoc5},
  url = {https://doi.org/10.63317/4ys6kofrzoc5}
}

Disclaimer

This model is provided as is, without warranties or guarantees of any kind, either express or implied. The authors make no representations regarding the accuracy, reliability, safety, suitability, or performance of the model or its outputs.

The model may generate incorrect, incomplete, or misleading results and should not be relied upon for critical, safety-sensitive, legal, medical, financial, or other high-stakes decisions. Use of this model is entirely at your own risk.

The authors accept no liability for damages, losses, or consequences arising from use, misuse, or inability to use the model.

lewiswatson/Frame2KG-Qwen3.5-4b-JSON

Frame2KG-Qwen3.5-4b-JSON