github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-22 13:42:25 -05:00
Code Issues 16.9k Packages Projects Releases 122 Wiki Activity

[GH-ISSUE #15519] Add MERaLiON-2 multimodal audio model support #9919

New Issue
Open
opened 2026-04-12 22:45:52 -05:00 by GiteaMirror · 0 comments
GiteaMirror commented 2026-04-12 22:45:52 -05:00
Owner
Copy Link

Originally created by @SiruiHe on GitHub (Apr 12, 2026).
Original GitHub issue: https://github.com/ollama/ollama/issues/15519

MERaLiON-2 is a speech-text model from A*STAR Singapore. It pairs a localized Whisper-large-v3 encoder with a Gemma-2 decoder, and covers English, Mandarin, Malay, and Tamil reasonably well, with a focus on Singapore's multilingual context. There are 3B and 10B variants.

The inference support landed in llama.cpp on April 11 (build b8762, ggml-org/llama.cpp#21756). GGUF files are on HuggingFace:

The architecture adds one new projector type (PROJECTOR_TYPE_MERALION) to the existing mtmd framework, following the same pattern as Ultravox and Voxtral. It needs llama.cpp at b8762 or later to run.

I noticed Ollama is currently pinned to ec98e200 from December 2025, so this would require a llama.cpp bump to pick up. Posting here mainly to flag interest and check if there is a preferred process for getting this into the library. Happy to help with a Modelfile or anything else on our end.

Originally created by @SiruiHe on GitHub (Apr 12, 2026). Original GitHub issue: https://github.com/ollama/ollama/issues/15519 MERaLiON-2 is a speech-text model from A*STAR Singapore. It pairs a localized Whisper-large-v3 encoder with a Gemma-2 decoder, and covers English, Mandarin, Malay, and Tamil reasonably well, with a focus on Singapore's multilingual context. There are 3B and 10B variants. The inference support landed in llama.cpp on April 11 (build b8762, ggml-org/llama.cpp#21756). GGUF files are on HuggingFace: - 3B: https://huggingface.co/SiruiHe/MERaLiON-2-3B-GGUF - 10B: https://huggingface.co/SiruiHe/MERaLiON-2-10B-GGUF The architecture adds one new projector type (PROJECTOR_TYPE_MERALION) to the existing mtmd framework, following the same pattern as Ultravox and Voxtral. It needs llama.cpp at b8762 or later to run. I noticed Ollama is currently pinned to ec98e200 from December 2025, so this would require a llama.cpp bump to pick up. Posting here mainly to flag interest and check if there is a preferred process for getting this into the library. Happy to help with a Modelfile or anything else on our end.
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
amd
api
app
bug
build
cli
cloud
compatibility
context-length
create
docker
documentation
embeddings
feature request
feedback wanted
good first issue
gpt-oss
gpu
harmony
help wanted
image
install
intel
js
launch
linux
macos
memory
mlx
model
needs more info
networking
nvidia
ollama.com
performance
pull-request
Mirrored from GitHub Pull Request
 python
question
registry
rendering
thinking
tools
top
vulkan
windows
wsl
No Label
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
GiteaMirror ninjasurge
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/ollama#9919
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?