deepspeedai / deepspeedai/DeepSpeed

[REQUEST] How to use int8 quantization inference without training?

Open
#4,023 4 comments 1 reaction 0 assignees View on GitHub

Nobody has claimed this yet.

enhancement
Dominant language
Python
Stars
43.1k
Forks
5k
Avg merge
4d 15h
Merged PRs (30d)
112

Description

Hi, I read the docs about zero_quant, but it seems to require extra training.

And in deepspeed.init_inference, the dtype can be set to int8, but the code does nothing for int8. https://github.com/microsoft/DeepSpeed/blob/master/deepspeed/inference/engine.py#L521

Is is possible to quantize an exisiting llm model directly and inference?

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start by reading the zero_quant documentation and deepspeed/inference/engine.py around line 521 to understand the existing quantization and dtype paths. Determine whether direct int8 quantization of an existing model is supported, and document or implement a defined inference path with validation that the converted model runs without extra training.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
machine-learning, performance
Issue type
Feature
Difficulty
5/5
Estimated time
Over a week
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
20/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.