Comments (1)
AFAIK, vLLM will reuse the KV cache of one token only if all preceding token matches. So
- Yes. The KV cache of
system prompt + 'you are a clever assistant'
will be reused. - No. If the system prompt is not identical, the preceding token does not match and the KV cache of all future tokens will not be reused.
- It will not be immediately erased but may be evicted sometime in the future.
Sadly I am not familiar with --chat-template
so I am not sure about the last question.
from vllm.
Related Issues (20)
- [Usage]: distributed inference with kuberay HOT 4
- [Feature]: Health check for restart policy HOT 1
- [Bug]: Shape error encountered in speculative decoding when `enable_lora=True`
- [RFC]: Add control panel support for vLLM HOT 4
- [Bug]: single lora request error make all processing requests error
- [Usage]: gpu memory usage when using tensor parallel HOT 1
- [Installation]: Do we have the plan to update the pip package installation method for the CPU backend.
- [Usage]: why can't I set gpu nums while use "tensor_parallel_size"?
- [Bug]: assert parts[0] == "base_model" AssertionError HOT 2
- [Bug]: Too strict version requirement on `tiktoken` HOT 3
- [Bug]: `CohereForAI/c4ai-command-r-v01`OSError: [Errno 12] Cannot allocate memory HOT 1
- [Misc]: When is the planned date for the next release? HOT 1
- v0.4.3 Release Tracker HOT 12
- [Usage]: Passing a guided_json in offline inference
- [Usage]: Profiling Prefill and Decode Phases Separately HOT 2
- [Feature]: Support for Falcon-11B model (Falcon 2) HOT 1
- [Bug]: llm_engine_example.py (more requests) get stuck
- [Bug]: Cannot use FlashAttention-2 backend because the flash_attn package is not found HOT 8
- [Bug]: Can't use offline inference embedding HOT 2
- [Misc]: can vllm support long content inference like 800k HOT 1
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from vllm.