Downloads · 30 days
0
ai-detection-demo-gemma_4_e2b
DarrenJiaImbue/ai-detection-demo-gemma_4_e2b
ai-detection-demo-gemma_4_e2b is a machine learning model from DarrenJiaImbue. Use it for the machine learning task on the model card, and read the license before you ship it in a product. The card lists the license as cc-by-nc-sa-4.0.
LoRA adapter (attention q/k/v/o, r=8, α=16) + linear-probe classification head for on-device AI-text detection, trained under deployment-exact fake-quant (static int8 activations + post-RoPE int8 KV cache) over the fr…
Access
Public
Updated Jul 14, 2026
Repo size
10.8 MB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors10.8 MB · 100%
At a glance
- License
- cc-by-nc-sa-4.0
- Access
- Public
- Created
- Jul 13, 2026
- Updated
- Jul 14, 2026
- SHA
- 2fd3a827
Base models
Model card
From the Hugging Face model README
gemma_4_e2b AI-text detector — training checkpoint
LoRA adapter (attention q/k/v/o, r=8, α=16) + linear-probe classification
head for on-device AI-text detection, trained under deployment-exact
fake-quant (static int8 activations + post-RoPE int8 KV cache) over the
frozen, dequantized google/gemma-4-E2B-it-qat-mobile-transformers text
tower.
This is the PEFT checkpoint the deployed LiteRT artifacts
(DarrenJiaImbue/gemma-4-e2b-ai-text-detector-v2)
were converted from. Training/eval/export code:
imbue-ai/ai-detection-demo
(gemma_4_e2b/). Build the base with python -m gemma_4_e2b.prepare_base,
then score with python -m gemma_4_e2b.inference.
Test-set results (device-faithful int8 proxy): human-vs-AI 0.996 accuracy, ternary 0.882.
- License
- cc-by-nc-sa-4.0
- Created
- Jul 13, 2026
- Updated
- Jul 14, 2026