# Help to choose a model for compact summarization

**URL:** </static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fhelp-to-choose-a-model-for-compact-summarization%2F116359%3E%5C
**Category:** 🤗Hub\
**Created:** [November 8, 2024, 8:44am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fhelp-to-choose-a-model-for-compact-summarization%2F116359 "2024-11-08T08:44:04Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![KirGear](https://sea2.discourse-cdn.com/hellohellohello/user_avatar/discuss.huggingface.co/kirgear/32/34208_2.png) [@KirGear](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2FKirGear)%5C
**Post date:** [November 8, 2024, 8:44am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fhelp-to-choose-a-model-for-compact-summarization%2F116359%2F1 "2024-11-08T08:44:04Z")

</div>

Which language model should be used to summarize a small number of small comments into a single sentence? For example, if this is the input:

“week 1: agent 375 has a good performance  
week 2: agent 375 did 3 mistakes in Monday  
week 3: agent 375 has the best performance in the team”

I want to have output like this:

“agent 375 performed very well, except few mistakes”

I also want this model to be small enough to run locally with 8GB of RAM.

---

<div class="post-metadata">

**Author:** ![John6666](https://sea2.discourse-cdn.com/hellohellohello/user_avatar/discuss.huggingface.co/john6666/32/27664_2.png) [@John6666](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2FJohn6666)%5C
**Post date:** [November 8, 2024, 9:40am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fhelp-to-choose-a-model-for-compact-summarization%2F116359%2F2 "2024-11-08T09:40:10Z")

</div>

There are many models that can be used for text summarization tasks, but with 8GB of RAM, it’s tough to use a general-purpose model larger than 3B. I think it would be better to find a small model that has been adjusted specifically for summarization, or to try a general-purpose model of 2B or less.  
There is no way to search by size, so there are also large models mixed in. The T5 models are large.

> **[Models - Hugging Face](https://huggingface.co/models?library=transformers&language=en&sort=downloads&search=summar)**
>
> We’re on a journey to advance and democratize artificial intelligence through open source and open science.

> **[Qwen/Qwen2.5-1.5B-Instruct · Hugging Face](https://huggingface.co/Qwen/Qwen2.5-1.5B-Instruct)**
>
> We’re on a journey to advance and democratize artificial intelligence through open source and open science.