Downloads · 30 days
0
LoRA_Qwen3-4B-Thinking-2507_from_base
amine-khelif/LoRA_Qwen3-4B-Thinking-2507_from_base
LoRA_Qwen3-4B-Thinking-2507_from_base is a machine learning model from amine-khelif. Use it for the machine learning task on the model card, and read the license before you ship it in a product.
Collection of LoRA adapters extracted from Qwen/Qwen3-4B-Thinking-2507 using different rank and epsilon parameters.
Access
Public
Updated Aug 12, 2025
Repo size
6.7 GB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors33.6 GB · 100%
At a glance
- Access
- Public
- Created
- Aug 11, 2025
- Updated
- Aug 12, 2025
- SHA
- 124168a6
Model card
From the Hugging Face model README
Qwen3 LoRA Adapters Collection
Collection of LoRA adapters extracted from Qwen/Qwen3-4B-Thinking-2507 using different rank and epsilon parameters.
Structure
adapters/
├── rank8_eps0.01/
├── rank8_eps0.1/
├── rank16_eps0.01/
└── ...
Usage
from peft import PeftModel
from transformers import AutoModelForCausalLM
# Load base model
base_model = AutoModelForCausalLM.from_pretrained("Qwen/Qwen3-4B-Base")
# Apply specific LoRA adapter
lora_model = PeftModel.from_pretrained(base_model, "adapters/rank16_eps0.10")
Parameter Guide
- Rank: Lower = more efficient, Higher = potentially better quality
- Epsilon: Threshold for singular value filtering (lower = more components kept)
- Created
- Aug 11, 2025
- Updated
- Aug 12, 2025