github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-06 08:02:14 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #11796] Type code for MXFP4 Quantization different from llama.cpp #33584

New Issue
Closed
opened 2026-04-22 16:26:35 -05:00 by GiteaMirror · 1 comment
GiteaMirror commented 2026-04-22 16:26:35 -05:00
Owner
Copy Link

Originally created by @chaserhkj on GitHub (Aug 7, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/11796

What is the issue?

I just discovered that type code used for MXFP4 quantization defined in ollama is different then defined in llama.cpp:

f2e9c9aff5/fs/ggml/ggml.go (L303)

Defined in llama.cpp:

MXFP4   = 39

This is causing fine-tuned gpt-oss models converted using llama.cpp unable to be properly loaded by ollama, see logs below.

Relevant log output

time=2025-08-07T21:42:35.398Z level=INFO source=server.go:175 msg=offload library=rocm layers.requested=-1 layers.model=25 layers.offload=25 layers.split="" memory.available="[16.0 GiB]" memory.gpu_overhead="0 B" memory.required.full="2.2 GiB" memory.required.partial="2.2 GiB" memory.required.kv="192.0 MiB" memory.required.allocations="[2.2 GiB]" memory.weights.total="1.2 GiB" memory.weights.repeating="680.5 MiB" memory.weights.nonrepeating="586.8 MiB" memory.graph.full="256.0 MiB" memory.graph.partial="256.0 MiB"
gguf_init_from_file_impl: tensor 'blk.0.ffn_down_exps.weight' has invalid ggml type 39 (NONE)
gguf_init_from_file_impl: failed to read tensor info
[GIN] 2025/08/07 - 21:42:35 | 500 |  148.708508ms |   192.168.110.1 | POST     "/api/chat"
[GIN] 2025/08/07 - 21:43:38 | 200 |    5.979947ms |   192.168.110.1 | GET      "/api/tags"
llama_model_load: error loading model: llama_model_loader: failed to load model from /root/.ollama/models/blobs/sha256-aa130697e8a9d565bf6d4eb72efa65005269e69705232aab42e39c1fd10276ea

llama_model_load_from_file_impl: failed to load model

OS

Docker

GPU

AMD

CPU

AMD

Ollama version

0.11.3

Originally created by @chaserhkj on GitHub (Aug 7, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/11796 ### What is the issue? I just discovered that type code used for MXFP4 quantization defined in ollama is different then defined in llama.cpp: https://github.com/ollama/ollama/blob/f2e9c9aff5f59b21a5d9a9668408732b3de01e20/fs/ggml/ggml.go#L303 [Defined](https://github.com/ggml-org/llama.cpp/blob/50aa9389014bba2dd12234132aa6b8ca3601a17f/gguf-py/gguf/constants.py#L2730C1-L2730C17) in llama.cpp: ```cpp MXFP4 = 39 ``` This is causing fine-tuned gpt-oss models converted using llama.cpp unable to be properly loaded by ollama, see logs below. ### Relevant log output ```shell time=2025-08-07T21:42:35.398Z level=INFO source=server.go:175 msg=offload library=rocm layers.requested=-1 layers.model=25 layers.offload=25 layers.split="" memory.available="[16.0 GiB]" memory.gpu_overhead="0 B" memory.required.full="2.2 GiB" memory.required.partial="2.2 GiB" memory.required.kv="192.0 MiB" memory.required.allocations="[2.2 GiB]" memory.weights.total="1.2 GiB" memory.weights.repeating="680.5 MiB" memory.weights.nonrepeating="586.8 MiB" memory.graph.full="256.0 MiB" memory.graph.partial="256.0 MiB" gguf_init_from_file_impl: tensor 'blk.0.ffn_down_exps.weight' has invalid ggml type 39 (NONE) gguf_init_from_file_impl: failed to read tensor info [GIN] 2025/08/07 - 21:42:35 | 500 | 148.708508ms | 192.168.110.1 | POST "/api/chat" [GIN] 2025/08/07 - 21:43:38 | 200 | 5.979947ms | 192.168.110.1 | GET "/api/tags" llama_model_load: error loading model: llama_model_loader: failed to load model from /root/.ollama/models/blobs/sha256-aa130697e8a9d565bf6d4eb72efa65005269e69705232aab42e39c1fd10276ea llama_model_load_from_file_impl: failed to load model ``` ### OS Docker ### GPU AMD ### CPU AMD ### Ollama version 0.11.3
GiteaMirror added the bug label 2026-04-22 16:26:35 -05:00
GiteaMirror commented 2026-04-22 16:26:37 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Aug 7, 2025):

#11714, #11710

<!-- gh-comment-id:3166048022 --> @rick-github commented on GitHub (Aug 7, 2025): #11714, #11710
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 bug
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#33584
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?