Downloads · 30 days
8
3% of all-time downloads
aletheias-phoenix-wright-v3.1-rank24-adapter
Jazhyc/aletheias-phoenix-wright-v3.1-rank24-adapter
aletheias-phoenix-wright-v3.1-rank24-adapter is a text generation model from Jazhyc. Use it when you need the model to write or continue text. It is set up for peft. The card lists the license as mit.
This is the rank-24, alpha-48 Qwen/Qwen3.5-9B LoRA used by the Phoenix Wright 3.1 Aletheia's Quest submission. It was trained on all 2,877 parsed, label-consistent varied-deception privileged reasoning summaries for o…
All-time downloads
250
Public
Repo size
349 MB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors175 MB · 100%
At a glance
- Task
- Text Generation
- Library
- peft
- License
- mit
- Access
- Public
- Created
- Jul 23, 2026
- Updated
- Jul 28, 2026
- SHA
- b07c0e29
Try a prompt
Base models
Model card
From the Hugging Face model README
Phoenix Wright 3.1 rank-24 deception adapter
This is the rank-24, alpha-48 Qwen/Qwen3.5-9B LoRA used by the Phoenix Wright
3.1 Aletheia's Quest submission. It was trained on all 2,877 parsed,
label-consistent varied-deception privileged reasoning summaries for one epoch
with AdamW at 5e-5 and effective batch size 32. It targets the attention and
MLP projection modules and emits a concise evidence summary followed by
Prediction:0 or Prediction:1.
Matched local validation against the rank-16 predecessor showed a small gain, but all available varied-deception training rows are Qwen-family. This adapter should therefore not be treated as evidence of cross-family generalization or as a general-purpose truth classifier.
The model is published for reproducibility of the matching competition notebook. License: MIT.
- Task
- Text Generation
- Library
- peft
- License
- mit
- Created
- Jul 23, 2026
- Updated
- Jul 28, 2026