Comments (7)
The warmup used every time when model is loaded can help triton to tune the best configuration that can make inference as fast as possible. But since it's kindly like grid search, hyper-params are manually set (hard coded), it may have no effect for some card. So now I've added warmup_triton
to from_quantized api, one can set it to False to skip warmup stage and then model loading can be fast.
I will also find a way to let users to cache the best triton configuration and save as a file so that one can warmup only once.
from autogptq.
Will close this issue for loading time problem of .from_quantized
has been fixed. Feel free to reopen or raise a new issue if you still encounter similar problem.
I will also find a way to let users to cache the best triton configuration and save as a file so that one can warmup only once.
For this I will add into a backlog as a future work.
from autogptq.
Hi, when using triton with from_quantized, there is a auto_tune_warmup is executed. Maybe I should make the warmup as an option for users.
from autogptq.
Is the "warm up" needed after you saved the model once?
from autogptq.
Is the "warm up" needed after you saved the model once?
from autogptq.
Interested in the response to @philschmid's question as well. If so, what are the implications of not using the warm up?
from autogptq.
Thank you will give it a try later this week!
from autogptq.
Related Issues (20)
- [BUG] Llama 3 8B Instruct - `no_inject_fused_attention` must be true or else errors out HOT 8
- Why doesn't AutoGPTQ quantize lm_head layer? HOT 5
- What magnitude of avg loss indicates a relatively good result for a quantization model HOT 6
- Why LLaMA3-8B after GPTQ test in wikitext2 so bad? HOT 8
- [PR Ready for Review] [FEATURE] Extend Support for Phi-3
- [FEATURE] Backport vllm expanded Marlin kernel to autogptq. HOT 1
- [DEPRECATION] Discussion on Fused attention and QiGEN HOT 5
- Llama-3 8B Instruct quantized to 8 Bit spits out gibberish in transformers `model.generate()` but works fine in vLLM? HOT 5
- [BUG]safetensors_rust.SafetensorError: Error while deserializing header: MetadataIncompleteBuffer
- [Question] Differences in quantization logic compared to AWQ
- [FEATURE] ADD SUPPORT DeepSeek-V2
- [BUG] ARM installation error
- [BUG] ROCm installation and building broken
- Target modules [] not found in the base model. Please check the target modules and try again.
- [BUG] Cannot install from source
- [BUG] Following the quant_with_alpaca.py example but keep getting "You shouldn't move a model that is dispatched using accelerate hooks." and the model is never saved.
- [FEATURE] Models that support MOE do GPTQ
- [FEATURE] Add marlin24 support
- How to select between different kernels?
- Question about data shape difference between quantization and forward
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 autogptq.