[GH-ISSUE #449] Different codellama models - confused #210

Closed
opened 2026-04-12 09:44:01 -05:00 by GiteaMirror · 3 comments
Owner

Originally created by @abulka on GitHub (Sep 1, 2023).
Original GitHub issue: https://github.com/ollama/ollama/issues/449

What is the difference between

  • ollama run codellama:7b
  • ollama run codellama:7b-instruct
  • ollama run codellama:7b-code

for general purpose (not python specialised) programming?

Originally created by @abulka on GitHub (Sep 1, 2023). Original GitHub issue: https://github.com/ollama/ollama/issues/449 What is the difference between - ollama run codellama:7b - ollama run codellama:7b-instruct - ollama run codellama:7b-code for general purpose (not python specialised) programming?
Author
Owner

@mchiang0610 commented on GitHub (Sep 1, 2023):

@abulka sorry that it wasn't clear.

  • codellama:7b is the same as 7b-instruct, we alias it so that it'll be simple for people just wishing to run codellama.

instruct is for general purpose - this is where you can just 'chat' or have a dialog with it.

code - is made for code completion. So if you pass in partial unfinished code, it'll try to complete it. This is usually made for apps, code editors etc.

Hope this helps! Let me know if I've answered your question, and how we can make it clear on the page. Did you find it in the Ollama model library?

<!-- gh-comment-id:1702006687 --> @mchiang0610 commented on GitHub (Sep 1, 2023): @abulka sorry that it wasn't clear. - codellama:7b is the same as 7b-instruct, we alias it so that it'll be simple for people just wishing to run codellama. instruct is for general purpose - this is where you can just 'chat' or have a dialog with it. code - is made for code completion. So if you pass in partial unfinished code, it'll try to complete it. This is usually made for apps, code editors etc. Hope this helps! Let me know if I've answered your question, and how we can make it clear on the page. Did you find it in the Ollama model library?
Author
Owner

@abulka commented on GitHub (Sep 1, 2023):

Thanks for the explanation! Would be useful if this info was added to https://ollama.ai/blog/run-code-llama-locally at least.

I'm subscribed to the Ollama newsletter and probably heard about it there - or possibly via the TLDR newsletter.

<!-- gh-comment-id:1702403722 --> @abulka commented on GitHub (Sep 1, 2023): Thanks for the explanation! Would be useful if this info was added to https://ollama.ai/blog/run-code-llama-locally at least. I'm subscribed to the Ollama newsletter and probably heard about it there - or possibly via the [TLDR newsletter](https://tldr.tech/changeset).
Author
Owner

@technovangelist commented on GitHub (Sep 1, 2023):

Hey there, the overview for the model on ollama.ai (https://ollama.ai/library/codellama) is the best place to go to find out, but I didn't make it easy to translate the name of the model variant to the actual tags. Now I have an example for each one.
CleanShot 2023-09-01 at 09 57 01

<!-- gh-comment-id:1703058352 --> @technovangelist commented on GitHub (Sep 1, 2023): Hey there, the overview for the model on ollama.ai (https://ollama.ai/library/codellama) is the best place to go to find out, but I didn't make it easy to translate the name of the model variant to the actual tags. Now I have an example for each one. ![CleanShot 2023-09-01 at 09 57 01](https://github.com/jmorganca/ollama/assets/633681/f624e54d-02e1-49ed-990e-32ed12ad69fa)
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/ollama#210