MiniCPM-SALA
MiniCPM Text Series
MiniCPM-SALA is a 9B-parameter model for ultra-long text. It uses a hybrid sparse and linear attention architecture and supports contexts of up to a million tokens.
Download the MiniCPM-SALA weights and prepare the corresponding runtime environment following the Deployment Guide. Resource requirements for long-context runs vary with your actual configuration, so check both your hardware and your inference parameters.
Key features
- Million-token context: Supports ultra-long context of up to 1M tokens
- Hybrid attention: 25% InfLLM-V2 sparse attention + 75% Lightning linear attention
- HyPE positional encoding: Handles ultra-long sequences efficiently
- 2.5x TTFT speedup: Compared with full-attention models, time to first token is reduced by 60%
Model downloads
| HuggingFace | ModelScope |
|---|---|
| MiniCPM-SALA | MiniCPM-SALA |
Developer Showcase
View all →MiniCPM TextEdgeHome HarnessA replayable, evaluable local control loop for a 1B on-device model, built in RustMiniCPM TextSqueezing a NAS: Hooking Up WorkBuddyRunning local AI on an N150 CPU: real-world uses and lessons learnedMiniCPM TextWhen a Catgirl Learns to Think, on Your PhoneWas this model made for catgirls? An on-device deployment testMiniCPM TextMaking a 1B Model a Good Math AgentTeXada’s on-device MiniCPM practice and lessonsMiniCPM TextBMB Agent: a MiniCPM-powered Personal AssistantLetting a 2B on-device model take over schedules and todos, with real-world lessonsMiniCPM TextSelf-Evolving Catgirl: Reasoning via On-Policy Self DistillationNo teacher, no new base model: math reasoning emerges from self-distillation