QJev-1.5B-LoRA / README.md
devankit7873's picture
fix: establish correct model tree hierarchy in README metadata
1bb506d verified
|
Raw History Blame Contribute Delete
3.13 kB
---
license: apache-2.0
base_model: devankit7873/QJev-1.5B
library_name: peft
tags:
- transformers
- peft
- lora
- qlora
- pytorch
- qwen
- qwen2
- decision-making
- system-one
- reasoning
- jev
- jevbench
- joint-expected-value
- classification
- safetensors
- agent
- fast-inference
datasets:
- SargeDev/jev-distill-corpus-v3
metrics:
- accuracy
- tvd
- kl_divergence
pipeline_tag: text-classification
---
<div align="center">
# โšก QJev-1.5B-LoRA: Lightweight PEFT Adapter (~20 MB)
### Joint Attention Adapter + MLP Decision Head for `Qwen/Qwen2.5-1.5B-Instruct`
[![Hugging Face](https://img.shields.io/badge/Hugging%20Face-QJev--1.5B--LoRA-FFD21E?style=for-the-badge&logo=huggingface&logoColor=black)](https://huggingface.co/devankit7873/QJev-1.5B-LoRA)
[![GitHub](https://img.shields.io/badge/GitHub-QJev--1.5B-181717?style=for-the-badge&logo=github)](https://github.com/AnkitNayak-dev/QJev-1.5B)
[![JevBench](https://img.shields.io/badge/JevBench-71.5%25%20Score-8A2BE2?style=for-the-badge)](https://jevbench.github.io/)
[![License](https://img.shields.io/badge/License-Apache%202.0-2EA44F?style=for-the-badge)](LICENSE)
---
### ๐Ÿ“ฆ Part of the QJev Model Family:
- **Flagship Standalone (Fused):** [devankit7873/QJev-1.5B](https://huggingface.co/devankit7873/QJev-1.5B) (3.08 GB)
- **Lightweight LoRA Adapter:** [devankit7873/QJev-1.5B-LoRA](https://huggingface.co/devankit7873/QJev-1.5B-LoRA) (20 MB)
- **4-Bit Quantized:** [devankit7873/QJev-1.5B-4bit](https://huggingface.co/devankit7873/QJev-1.5B-4bit) (1.20 GB, 1.07 GB VRAM)
---
</div>
## ๐Ÿ’ก Overview
**`QJev-1.5B-LoRA`** is the lightweight **PEFT / LoRA adapter release** for edge, mobile, and modular LLM deployments.
Instead of downloading the entire 3GB transformer weights, developers who already cache `Qwen/Qwen2.5-1.5B-Instruct` can download just **20 MB** to equip their base model with high-speed System 1 decision-making capabilities.
- **Adapter Size:** 17.5 MB (`adapter_model.safetensors`, targeting `q_proj`, `k_proj`, `v_proj`, `o_proj`, rank 16, alpha 32)
- **Decision Head Size:** 3.18 MB (`decision_head.safetensors`, 1536 $\to$ 512 $\to$ 10)
- **JevBench Calibrated Score:** **71.5%** (Identical performance to full merged model)
---
## ๐Ÿš€ Quickstart: Super Friendly 3-Line Decision
```python
from qjev import QJev
# 1. Load the adapter in 1 line:
model = QJev.from_pretrained("devankit7873/QJev-1.5B-LoRA")
# 2. Decide in 31 ms!
result = model.decide(
context="Production database primary replica CPU spiked to 100% and connection pool is exhausted.",
choices=[
"Trigger immediate automated failover to standby replica",
"Restart the primary database daemon immediately",
"Double connection pool limits without scaling compute",
"Silence alerts and monitor for next 15 minutes"
]
)
print(f"๐Ÿ‘‰ Recommended: {result['best_choice']} ({result['confidence']}%)")
for choice, prob in result['probabilities'].items():
print(f" [{prob:>5.2f}%] {choice}")
```
---
## ๐Ÿ“„ License
Released under the **Apache 2.0 License**. Base model governed by Alibaba Cloud Qwen License.