Downloads · 30 days
11
3% of all-time downloads
Slotherynn/legal-chatbot-qwen-sft
legal-chatbot-qwen-sft is a text generation model from Slotherynn. Use it when you need the model to write or continue text. It is set up for transformers. The card lists the license as apache-2.0.
This is a historical supervised fine-tuning experiment based on Qwen2.5-3B-Instruct. Despite its repository name, it was trained on a general Indonesian instruction dataset, not a curated legal QA dataset, and must no…
Downloads · 30 days
11
3% of all-time downloads
All-time downloads
412
Public
Parameters
3.1B
6.2 GB on disk
Likes
0
Public
Click a slice to open those files.
.safetensors6.2 GB · 100%
From the Hugging Face model README
This is a historical supervised fine-tuning experiment based on Qwen2.5-3B-Instruct. Despite its repository name, it was trained on a general Indonesian instruction dataset, not a curated legal QA dataset, and must not be treated as an Indonesian legal expert.
Ichsan2895/alpaca-gpt4-indonesian, licensed CC BY-SA 4.0 and
derived from FreedomIntelligence/alpaca-gpt4-indonesian.2e-4.The model was evaluated on 60 reviewed Indonesian legal cases using the same frozen retrieval results, prompt, deterministic decoding, and validator as the base and GRPO models.
| Metric | Result |
|---|---|
| Attempted generations | 43 |
| Valid grounded outputs | 0 |
| Citation precision | Not evaluable |
| Faithfulness | Not evaluable |
| Answer relevance | Not evaluable |
| Abstention accuracy | 0.2167 |
| Mean generation latency | 13.72 s |
| Peak VRAM (4-bit load) | 2686.36 MB |
| Loaded footprint | 1916.96 MB |
Every attempted generation failed the required grounded-output contract. The model is rejected for the production RAG application; fine-tuning showed no quality improvement over the base model.
The project retains this model and its training notebook only as a learning artifact. The application uses a deterministic extractive fallback instead.