cyankiwi/Qwen3.6-35B-A3B-AWQ-4bit
Image-Text-to-Text • 36B • Updated • 1.49M • 74
MHL
Zipper112
·
AI & ML interests
None yet
Recent Activity
liked a model about 14 hours ago
cyankiwi/Qwen3.6-35B-A3B-AWQ-4bit liked a model 16 days ago
sapientinc/HRM-Text-1B liked a model about 1 month ago
Qwen/WebWorld-14BOrganizations
None yet
liked a
model about 14 hours ago
liked a
model 16 days ago
liked a
model about 1 month ago
liked a
dataset about 1 month ago
upvoted a paper about 1 month ago
liked a
dataset about 1 month ago
liked 2
models about 2 months ago
reacted to SeanLee97's post with 🔥 about 2 months ago
Post
8201
Our lab recently released a paper where we introduce ShadowPEFT, a new Parameter-Efficient Fine-Tuning (PEFT) paradigm tailored for edge computing scenarios.
Unlike traditional approaches such as LoRA and its variants, which inject trainable parameters directly into the weights of Transformer, requiring tight coupling with the backbone.
ShadowPEFT instead enhances the frozen large base model by adding a lightweight, centralized, pretrainable, and detachable Shadow network.
This shadow network operates in parallel with the base model, delivering learned corrections to each decoder layer. Because the shadow module is architecturally decoupled from the backbone, it can be independently trained, stored, and deployed, benefiting edge computing scenarios and edge-cloud collaboration computing.
- HF Paper: ShadowPEFT: Shadow Network for Parameter-Efficient Fine-Tuning (2604.19254)
- GitHub: https://github.com/ShadowLLM/shadow-peft
- HF Collection: https://huggingface.co/collections/shadow-llm/shadow-peft-models
Unlike traditional approaches such as LoRA and its variants, which inject trainable parameters directly into the weights of Transformer, requiring tight coupling with the backbone.
ShadowPEFT instead enhances the frozen large base model by adding a lightweight, centralized, pretrainable, and detachable Shadow network.
This shadow network operates in parallel with the base model, delivering learned corrections to each decoder layer. Because the shadow module is architecturally decoupled from the backbone, it can be independently trained, stored, and deployed, benefiting edge computing scenarios and edge-cloud collaboration computing.
- HF Paper: ShadowPEFT: Shadow Network for Parameter-Efficient Fine-Tuning (2604.19254)
- GitHub: https://github.com/ShadowLLM/shadow-peft
- HF Collection: https://huggingface.co/collections/shadow-llm/shadow-peft-models
- 7 replies
liked 2
models about 2 months ago
liked 6
models 2 months ago
LiquidAI/LFM2.5-350M
Text Generation • 0.4B • Updated • 88.4k • 324
netflix/void-model
Video-to-Video • Updated • 940
Hcompany/Holo3-35B-A3B
Image-Text-to-Text • 35B • Updated • 15k • 358
chromadb/context-1
Text Generation • 21B • Updated • 1.82k • 410
mlx-community/LongCat-Next-4bit
Text Generation • 69B • Updated • 110 • 6
fishaudio/s2-pro
Text-to-Speech • 5B • Updated • 147k • 1.01k
liked 3
models 3 months ago
