github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-07-16 05:05:37 -05:00
Code Issues 16.9k Packages Projects Releases 122 Wiki Activity

[GH-ISSUE #9339] Importing RomboUltima-32B Fails with "unknown data type: U8" in Ollama #31859

New Issue
Closed
opened 2026-04-22 12:37:23 -05:00 by GiteaMirror · 1 comment
GiteaMirror commented 2026-04-22 12:37:23 -05:00
Owner
Copy Link

Originally created by @ALLMI78 on GitHub (Feb 25, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/9339

I attempted to import the FINGU-AI/RomboUltima-32B model into Ollama following the official import guide. However, I encountered the following error during the conversion process:

ollama create --quantize q4_0 RomboUltima-32B gathering model components copying file sha256:beae19148f00ebadc740411c319b6f952a43d5ade14cd21a21fa9954651f11ee 100% copying file sha256:629a4f234c7f8c060fa1be3af3a35896f124f6c00f4ffba5f4dc02efbb99f0f6 100% copying file sha256:704c38755ae9efd59ca45507eb219ab1c5a6a11d204753cd25b50697b6234d33 100% copying file sha256:548d9472e15d43770f6b509e0231e7fab1defb9db0142d231c508c9570860b7f 100% copying file sha256:9c5ae00e602b8860cbd784ba82a8aa14e8feecec692e7076590d014d7b7fdafa 100% copying file sha256:c7288d09761d0cd79c915376cdb9812248ccc085b4229b98924edd119c492f46 100% copying file sha256:bc26c94becde9c5f4642a5d5591f17ce8366f29367fdb742a10d8545abe0eda2 100% copying file sha256:58b54bbe36fc752f79a24a271ef66a0a0830054b4dfad94bde757d851968060b 100% copying file sha256:c2b1a4e8797e0f7a853a6c6c23e73e33de5f922367e023c7f78b758c857dfaf1 100% copying file sha256:6676f091c8bc4d1b50146427cfde92073402866b87b6e39223227931b70083e9 100% copying file sha256:a5dae0102f342a3b6bc2a4d03f430ed83666dcab8879c9ee44c08a3d9095132d 100% copying file sha256:ca10d7e9fb3ed18575dd1e277a2579c16d108e32f27439684afa0e10b1440910 100% converting model Error: unknown data type: U8

Steps to Reproduce:
Download RomboUltima-32B files from Hugging Face. >>> https://huggingface.co/FINGU-AI/RomboUltima-32B/tree/main
Run ollama create --quantize q4_0 RomboUltima-32B.
Encounter the error unknown data type: U8.

System Information:
OS: Windows 10
Ollama Version: ollama --version → 0.5.12

GPU: RTX 4060 Ti 16GB | RAM: 32GB

Model Format: safetensors (from Hugging Face)

Additional Information:
The error suggests an unsupported U8 (uint8) data type. This issue might be caused by an unsupported format in safetensors.

Expected Behavior:
Ollama should successfully import and quantize the model or provide a clear error message if the format is unsupported.

Possible Solutions:
Confirm whether safetensors models containing uint8 tensors are unsupported.
Provide a conversion script that handles this issue.
Add support for uint8 tensors if feasible.

Question:
What is the simplest way (no gguf found) to import RomboUltima-32B into Ollama? If direct import is not possible, what alternative steps should be taken? Would appreciate any guidance on whether this is a bug, missing feature, or incorrect import method. Thanks!

Thanks!

Originally created by @ALLMI78 on GitHub (Feb 25, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/9339 I attempted to import the [FINGU-AI/RomboUltima-32B](https://huggingface.co/FINGU-AI/RomboUltima-32B) model into Ollama following the official [import guide](https://github.com/ollama/ollama/blob/main/docs/import.md#Importing-a-model-from-Safetensors-weights). However, I encountered the following error during the conversion process: `ollama create --quantize q4_0 RomboUltima-32B gathering model components copying file sha256:beae19148f00ebadc740411c319b6f952a43d5ade14cd21a21fa9954651f11ee 100% copying file sha256:629a4f234c7f8c060fa1be3af3a35896f124f6c00f4ffba5f4dc02efbb99f0f6 100% copying file sha256:704c38755ae9efd59ca45507eb219ab1c5a6a11d204753cd25b50697b6234d33 100% copying file sha256:548d9472e15d43770f6b509e0231e7fab1defb9db0142d231c508c9570860b7f 100% copying file sha256:9c5ae00e602b8860cbd784ba82a8aa14e8feecec692e7076590d014d7b7fdafa 100% copying file sha256:c7288d09761d0cd79c915376cdb9812248ccc085b4229b98924edd119c492f46 100% copying file sha256:bc26c94becde9c5f4642a5d5591f17ce8366f29367fdb742a10d8545abe0eda2 100% copying file sha256:58b54bbe36fc752f79a24a271ef66a0a0830054b4dfad94bde757d851968060b 100% copying file sha256:c2b1a4e8797e0f7a853a6c6c23e73e33de5f922367e023c7f78b758c857dfaf1 100% copying file sha256:6676f091c8bc4d1b50146427cfde92073402866b87b6e39223227931b70083e9 100% copying file sha256:a5dae0102f342a3b6bc2a4d03f430ed83666dcab8879c9ee44c08a3d9095132d 100% copying file sha256:ca10d7e9fb3ed18575dd1e277a2579c16d108e32f27439684afa0e10b1440910 100% converting model Error: unknown data type: U8` Steps to Reproduce: Download RomboUltima-32B files from Hugging Face. >>> https://huggingface.co/FINGU-AI/RomboUltima-32B/tree/main Run ollama create --quantize q4_0 RomboUltima-32B. Encounter the error unknown data type: U8. System Information: OS: Windows 10 Ollama Version: ollama --version → 0.5.12 GPU: RTX 4060 Ti 16GB | RAM: 32GB Model Format: safetensors (from Hugging Face) Additional Information: The error suggests an unsupported U8 (uint8) data type. This issue might be caused by an unsupported format in safetensors. Expected Behavior: Ollama should successfully import and quantize the model or provide a clear error message if the format is unsupported. Possible Solutions: Confirm whether safetensors models containing uint8 tensors are unsupported. Provide a conversion script that handles this issue. Add support for uint8 tensors if feasible. Question: What is the simplest way (no gguf found) to import RomboUltima-32B into Ollama? If direct import is not possible, what alternative steps should be taken? Would appreciate any guidance on whether this is a bug, missing feature, or incorrect import method. Thanks! Thanks!
GiteaMirror commented 2026-04-22 12:37:24 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Feb 25, 2025):

llama.cpp (and by extension, ollama) doesn't support the U8 data type. You can only convert to GGUF format from models with data in float16, bfloat16 or float32 formats. You can't convert models that are already quantized to a non-GGML format. In order to import this model, you would need to re-quantize (eg, with pytorch or bitsandbytes) to a supported data type and then import.

<!-- gh-comment-id:2681737405 --> @rick-github commented on GitHub (Feb 25, 2025): llama.cpp (and by extension, ollama) doesn't support the U8 data type. You can only convert to GGUF format from models with data in float16, bfloat16 or float32 formats. You can't convert models that are already quantized to a non-GGML format. In order to import this model, you would need to re-quantize (eg, with pytorch or bitsandbytes) to a supported data type and then import.
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#31859
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?