Downloads · 30 days
0
legalchat-v11-lora
Agnes999/legalchat-v11-lora
legalchat-v11-lora is a machine learning model from Agnes999. Use it for the machine learning task on the model card, and read the license before you ship it in a product.
This is the deployed LegalChatModel adapter for mlx-community/Qwen2.5-7B-Instruct-Uncensored-4bit.
Access
Public
Updated Jul 18, 2026
Repo size
138 MB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors185 MB · 100%
At a glance
- Access
- Public
- Created
- Jul 18, 2026
- Updated
- Jul 18, 2026
- SHA
- afc35472
Model card
From the Hugging Face model README
V11 specialist LoRA adapter
This is the deployed LegalChatModel adapter for
mlx-community/Qwen2.5-7B-Instruct-Uncensored-4bit.
- Fine-tuning type: LoRA, rank 8, 16 layers
- Context used for training: 4,096 tokens
- Selected validation: V11 outperformed the preceding V10 adapter on the untouched test set (loss 2.681 vs 2.698; perplexity 14.601 vs 14.844)
- Privacy audit: the model-visible corrective datasets contain no local paths, candidate identifiers, private filenames, chat transcripts, or lower-mark submitted prose
Only the final adapters.safetensors and its configuration are published.
Intermediate checkpoints and the rejected V12 experiment are intentionally
excluded.
An adapter is not a legal authority or an accuracy guarantee. Runtime official source checking, authority repair, privacy filtering, and release supervision remain necessary.
- Created
- Jul 18, 2026
- Updated
- Jul 18, 2026