Downloads · 30 days
0
gqa-moe-topk
lathatechie99/gqa-moe-topk
gqa-moe-topk is a machine learning model from lathatechie99. Use it for the machine learning task on the model card, and read the license before you ship it in a product.
This model uses Grouped Query Attention (GQA) with Mixture of Experts.
Access
Public
Updated Nov 15, 2025
Repo size
5.1 GB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.bin125 MB · 99%
At a glance
- Access
- Public
- Created
- Nov 15, 2025
- Updated
- Nov 15, 2025
- SHA
- b91cafd6
Model card
From the Hugging Face model README
GQA-MoE Transformer
This model uses Grouped Query Attention (GQA) with Mixture of Experts.
Architecture
- Query heads: 4
- KV heads: 2
- Experts: 4
- Top-K: 2
Key Features
- Custom GQA implementation (not nn.MultiheadAttention)
- Top-K routing for experts
- Encoder-Decoder architecture for summarization
- Created
- Nov 15, 2025
- Updated
- Nov 15, 2025