[GH-ISSUE #4423] LLAVA1.6 performance huge drop after export/import using ModelFile #49274

Open
opened 2026-04-28 11:04:23 -05:00 by GiteaMirror · 0 comments
Owner

Originally created by @vai-minzhou on GitHub (May 14, 2024).
Original GitHub issue: https://github.com/ollama/ollama/issues/4423

What is the issue?

Using ModelFile to change the temperature of LLAVA 1.6 or importing LLAVA from a GGUF file significantly reduces performance. This issue likely arises because the import operation treats the LLAVA 1.6 model as LLAVA 1.5 or earlier, ignoring the image-splitting into four parts. Please address this bug.

OS

Linux

GPU

Nvidia

CPU

AMD

Ollama version

Any version

Originally created by @vai-minzhou on GitHub (May 14, 2024). Original GitHub issue: https://github.com/ollama/ollama/issues/4423 ### What is the issue? Using ModelFile to change the temperature of LLAVA 1.6 or importing LLAVA from a GGUF file significantly reduces performance. This issue likely arises because the import operation treats the LLAVA 1.6 model as LLAVA 1.5 or earlier, ignoring the image-splitting into four parts. Please address this bug. ### OS Linux ### GPU Nvidia ### CPU AMD ### Ollama version Any version
GiteaMirror added the bug label 2026-04-28 11:04:23 -05:00
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/ollama#49274