[GH-ISSUE #1312] trouble with deepseek-coder #47191

Closed
opened 2026-04-28 03:25:18 -05:00 by GiteaMirror · 4 comments
Owner

Originally created by @niknoproblems on GitHub (Nov 29, 2023).
Original GitHub issue: https://github.com/ollama/ollama/issues/1312

I'm having trouble getting this model to run on mac m1 16gb ram:
ollama run deepseek-coder:6.7b-base-q8_0

but this model work without any troubles:
ollama run neural-chat:7b-v3.1-q8_0

it has more weights and bigger file size.

Originally created by @niknoproblems on GitHub (Nov 29, 2023). Original GitHub issue: https://github.com/ollama/ollama/issues/1312 I'm having trouble getting this model to run on mac m1 16gb ram: ollama run deepseek-coder:6.7b-base-q8_0 but this model work without any troubles: ollama run neural-chat:7b-v3.1-q8_0 it has more weights and bigger file size.
Author
Owner

@deanrie commented on GitHub (Dec 1, 2023):

I'm experiencing the same problems, deepseek stopped running, although everything worked fine a couple of days ago, I thought it might be worth updating the ollama version, but it also gives an error 500, this problem is not only with deepseek, it occurs for all models.
I am using m2 16gb.

<!-- gh-comment-id:1835694535 --> @deanrie commented on GitHub (Dec 1, 2023): I'm experiencing the same problems, deepseek stopped running, although everything worked fine a couple of days ago, I thought it might be worth updating the ollama version, but it also gives an error 500, this problem is not only with deepseek, it occurs for all models. I am using m2 16gb.
Author
Owner

@deanrie commented on GitHub (Dec 1, 2023):

I just tried the deepseek-llm:7b-chat-q8_0 model, there are no problems with it, everything works fine.

<!-- gh-comment-id:1836342289 --> @deanrie commented on GitHub (Dec 1, 2023): I just tried the deepseek-llm:7b-chat-q8_0 model, there are no problems with it, everything works fine.
Author
Owner

@niknoproblems commented on GitHub (Dec 5, 2023):

Yeah, I had no problems with deepseek-llm:7b-chat-q8_0 model either. but deepseek-coder-6.7b-instruct.Q8_0 model is better for code

<!-- gh-comment-id:1840624498 --> @niknoproblems commented on GitHub (Dec 5, 2023): Yeah, I had no problems with deepseek-llm:7b-chat-q8_0 model either. but deepseek-coder-6.7b-instruct.Q8_0 model is better for code
Author
Owner

@pdevine commented on GitHub (Mar 12, 2024):

Sorry about the slow response. You probably want the instruct model and not the base model.

% ./ollama run deepseek-coder:6.7b-instruct-q8_0
pulling manifest
pulling b6dfda3810bc... 100% ▕█████████████████████████████████████████████████████████████████▏ 7.2 GB
pulling a3a0e9449cb6... 100% ▕█████████████████████████████████████████████████████████████████▏  13 KB
pulling 8893e08fa9f9... 100% ▕█████████████████████████████████████████████████████████████████▏   59 B
pulling 8972a96b8ff1... 100% ▕█████████████████████████████████████████████████████████████████▏  297 B
pulling 733c8610b9f5... 100% ▕█████████████████████████████████████████████████████████████████▏  483 B
verifying sha256 digest
writing manifest
removing any unused layers
success
>>> write me hello world in golang
Sure, here is a simple "Hello World" program written in Go language:

```go
package main

import "fmt"

func main() {
    fmt.Println("Hello, World!")
}

You can run this code by copying it into a file with a .go extension and running the file through the Go compiler (go run filename.go). Make sure you have installed Go on your system to compile and run the program.


I'm going to go ahead and close out the issue, but feel free to reopen it if you're still having issues. The `base` model is sentence completion model, so more appropriate for hooking up a code editor.
<!-- gh-comment-id:1989730018 --> @pdevine commented on GitHub (Mar 12, 2024): Sorry about the slow response. You probably want the `instruct` model and not the `base` model. ``` % ./ollama run deepseek-coder:6.7b-instruct-q8_0 pulling manifest pulling b6dfda3810bc... 100% ▕█████████████████████████████████████████████████████████████████▏ 7.2 GB pulling a3a0e9449cb6... 100% ▕█████████████████████████████████████████████████████████████████▏ 13 KB pulling 8893e08fa9f9... 100% ▕█████████████████████████████████████████████████████████████████▏ 59 B pulling 8972a96b8ff1... 100% ▕█████████████████████████████████████████████████████████████████▏ 297 B pulling 733c8610b9f5... 100% ▕█████████████████████████████████████████████████████████████████▏ 483 B verifying sha256 digest writing manifest removing any unused layers success >>> write me hello world in golang Sure, here is a simple "Hello World" program written in Go language: ```go package main import "fmt" func main() { fmt.Println("Hello, World!") } ``` You can run this code by copying it into a file with a `.go` extension and running the file through the Go compiler (`go run filename.go`). Make sure you have installed Go on your system to compile and run the program. ``` I'm going to go ahead and close out the issue, but feel free to reopen it if you're still having issues. The `base` model is sentence completion model, so more appropriate for hooking up a code editor.
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/ollama#47191