[GH-ISSUE #8775] After launching DeepSeek-R1 with ollama, it was found that different questions got the same answer #67753

Closed
opened 2026-05-04 11:34:54 -05:00 by GiteaMirror · 3 comments
Owner

Originally created by @gallery2016 on GitHub (Feb 2, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/8775

What is the issue?

Thank you for your attention.
After launching DeepSeek-R1 with ollama, it was found that different questions got the same answer.

Below is my detailed environment and configuration

  1. Ollama client version is 0.5.7

  2. CentOS Linux release 7.9.2009 (Core)

  3. The GPU is: 8 * L40

  4. Driver Version: 550.54.15 CUDA Version: 12.4

  5. Configuration of ollam startup:
    export OLLAMA_LOAD_TIMEOUT=90m
    export GGML_CUDA_ENABLE_UNIFIED_MEMORY=1
    export OLLAMA_GPU_OVERHEAD=536870912
    export OLLAMA_FLASH_ATTENTION=1

ollama serve
ollama run deepseek-r1:671b

  1. ollama chat result with DeepSeek-R1; It was found that different questions got the same answer.

Question:
Please give some common commands for ollama

Answer:
think
Greetings! I'm DeepSeek-R1, an artificial intelligence assistant created by DeepSeek. I'm at your service and would be delighted to assist you with any inquiries or tasks you may have.

It was found that different questions got the same answer.
Is there some configuration problem?

OS

Linux

GPU

Nvidia

CPU

Intel

Ollama version

0.5.7

Originally created by @gallery2016 on GitHub (Feb 2, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/8775 ### What is the issue? Thank you for your attention. After launching DeepSeek-R1 with ollama, it was found that different questions got the same answer. Below is my detailed environment and configuration 1. Ollama client version is 0.5.7 2. CentOS Linux release 7.9.2009 (Core) 3. The GPU is: 8 * L40 4. Driver Version: 550.54.15 CUDA Version: 12.4 5. Configuration of ollam startup: export OLLAMA_LOAD_TIMEOUT=90m export GGML_CUDA_ENABLE_UNIFIED_MEMORY=1 export OLLAMA_GPU_OVERHEAD=536870912 export OLLAMA_FLASH_ATTENTION=1 ollama serve ollama run deepseek-r1:671b 6. ollama chat result with DeepSeek-R1; It was found that different questions got the same answer. Question: Please give some common commands for ollama Answer: think Greetings! I'm DeepSeek-R1, an artificial intelligence assistant created by DeepSeek. I'm at your service and would be delighted to assist you with any inquiries or tasks you may have. It was found that different questions got the same answer. Is there some configuration problem? ### OS Linux ### GPU Nvidia ### CPU Intel ### Ollama version 0.5.7
GiteaMirror added the bug label 2026-05-04 11:34:54 -05:00
Author
Owner

@robustini commented on GitHub (Feb 2, 2025):

Same issue here.
I create different questions and it is continuing to give me the same answer over and over again.

<!-- gh-comment-id:2629488992 --> @robustini commented on GitHub (Feb 2, 2025): Same issue here. I create different questions and it is continuing to give me the same answer over and over again.
Author
Owner

@Vizway commented on GitHub (Feb 3, 2025):

Well I don´t get any answer from any of the Deepseek models I run.

<!-- gh-comment-id:2630638114 --> @Vizway commented on GitHub (Feb 3, 2025): Well I don´t get any answer from any of the Deepseek models I run.
Author
Owner

@goactiongo commented on GitHub (Feb 6, 2025):

any update for this?

<!-- gh-comment-id:2638979558 --> @goactiongo commented on GitHub (Feb 6, 2025): any update for this?
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/ollama#67753