LFM2.5-Encoders for Fast Long-Context Inference on CPU
• 30
AI & ML interests
A new generation of foundation models from first principles.
Recent Activity
View all activity
Articles
Collection of post-trained and base LFM2.5 models.
Japanese-tuned LFMs.
-
LiquidAI/LFM2.5-1.2B-JP-202606
Text Generation • 1B • Updated • 23.7k • 68 -
LiquidAI/LFM2.5-1.2B-JP-202606-GGUF
Text Generation • 1B • Updated • 2.55k • 21 -
LiquidAI/LFM2.5-Audio-1.5B-JP
Audio-to-Audio • 1B • Updated • 178 • 68 -
LiquidAI/LFM2.5-Audio-1.5B-JP-GGUF
Audio-to-Audio • 1B • Updated • 2.19k • 23
LFM2-VL is our first series of vision-language models, designed for on-device deployment.
Library of task-specific models: https://www.liquid.ai/blog/introducing-liquid-nanos-frontier-grade-performance-on-everyday-devices
-
LiquidAI/LFM2.5-VL-450M
Image-Text-to-Text • 0.4B • Updated • 80.3k • 206 -
LFM2.5-VL-450M WebGPU
📹60Live video captioning and object tracking in your browser
-
LiquidAI/LFM2.5-VL-1.6B
Image-Text-to-Text • 2B • Updated • 56.6k • 309 -
LFM2.5-VL-1.6B WebGPU
🧠92In-browser vision-language inference with LFM2.5-VL-1.6B
LFM2 is a new generation of hybrid models, designed for on-device deployment.
End-to-end audio foundation model, designed for low latency and real-time conversations
Collection of post-trained and base LFM2.5 models.
Library of task-specific models: https://www.liquid.ai/blog/introducing-liquid-nanos-frontier-grade-performance-on-everyday-devices
Japanese-tuned LFMs.
-
LiquidAI/LFM2.5-1.2B-JP-202606
Text Generation • 1B • Updated • 23.7k • 68 -
LiquidAI/LFM2.5-1.2B-JP-202606-GGUF
Text Generation • 1B • Updated • 2.55k • 21 -
LiquidAI/LFM2.5-Audio-1.5B-JP
Audio-to-Audio • 1B • Updated • 178 • 68 -
LiquidAI/LFM2.5-Audio-1.5B-JP-GGUF
Audio-to-Audio • 1B • Updated • 2.19k • 23
-
LiquidAI/LFM2.5-VL-450M
Image-Text-to-Text • 0.4B • Updated • 80.3k • 206 -
LFM2.5-VL-450M WebGPU
📹60Live video captioning and object tracking in your browser
-
LiquidAI/LFM2.5-VL-1.6B
Image-Text-to-Text • 2B • Updated • 56.6k • 309 -
LFM2.5-VL-1.6B WebGPU
🧠92In-browser vision-language inference with LFM2.5-VL-1.6B
LFM2 is a new generation of hybrid models, designed for on-device deployment.
LFM2-VL is our first series of vision-language models, designed for on-device deployment.
End-to-end audio foundation model, designed for low latency and real-time conversations