Towards Economical Inference: Enabling DeepSeek's Multi-Head Latent Attention in Any Transformer-based LLMs
Paper
• 2502.14837 • Published
• 3
MHA2MLA
The MHA2MLA model published in the paper "Towards Economical Inference: Enabling DeepSeek's Multi-Head Latent Attention in Any Transformer-Based LLMs"
-
-
OpenMOSS-Team/Llama-2-7B-MLA-d_kv_16
Text Generation • 6B • Updated • 1 -
OpenMOSS-Team/Llama-2-7B-MLA-d_kv_32
Text Generation • 6B • Updated • 2 -
OpenMOSS-Team/Llama-2-7B-MLA-d_kv_64
Text Generation • 7B • Updated • 1 -
OpenMOSS-Team/Llama-2-7B-MHA-d_kv_256
Text Generation • 7B • Updated • 8 • 3 -
OpenMOSS-Team/SmolLM-135M-MLA-d_kv_8
Text Generation • 0.1B • Updated • 2 • 1 -
OpenMOSS-Team/SmolLM-135M-MLA-d_kv_16
Text Generation • Updated -
OpenMOSS-Team/SmolLM-135M-MLA-d_kv_32
Text Generation • 0.1B • Updated • 2 -
OpenMOSS-Team/SmolLM-135M-GQA-d_kv_128
Text Generation • 0.1B • Updated • 2 -
OpenMOSS-Team/SmolLM-360M-MLA-d_kv_8
Text Generation • 0.3B • Updated • 3 -
OpenMOSS-Team/SmolLM-360M-MLA-d_kv_16
Text Generation • 0.3B • Updated • 2 -
OpenMOSS-Team/SmolLM-360M-MLA-d_kv_32
Text Generation • 0.4B • Updated • 5 -
OpenMOSS-Team/SmolLM-360M-GQA-d_kv_128
Text Generation • 0.4B • Updated • 1 -
OpenMOSS-Team/SmolLM-1B7-MLA-d_kv_8
Text Generation • 2B • Updated • 2 -
OpenMOSS-Team/SmolLM-1B7-MLA-d_kv_16
Text Generation • 2B • Updated • 3 -
OpenMOSS-Team/SmolLM-1B7-MLA-d_kv_32
Text Generation • 2B • Updated • 2 -
OpenMOSS-Team/SmolLM-1B7-MHA-d_kv_128
Text Generation • 2B • Updated • 2 -
OpenMOSS-Team/MHA2MLA-corpus-llama2
Updated • 37 -
OpenMOSS-Team/MHA2MLA-corpus-smollm_v1
Updated • 32 -
OpenMOSS-Team/MHA2MLA-corpus-mistral-v0_1
Updated • 31 -
OpenMOSS-Team/MHA2MLA-corpus-qwen2
Updated • 6 -
OpenMOSS-Team/MHA2MLA-corpus-qwen1_5
Updated • 49