github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-23 06:01:37 -05:00
Code Issues 16.9k Packages Projects Releases 122 Wiki Activity

[GH-ISSUE #9293] xLSTM please! #83746

New Issue
Open
opened 2026-05-09 19:01:22 -05:00 by GiteaMirror · 0 comments
GiteaMirror commented 2026-05-09 19:01:22 -05:00
Owner
Copy Link

Originally created by @ai-bits on GitHub (Feb 22, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/9293

xLSTM is the 3rd of a kind (besides Transformers and Mamba) and its performace is touted to be superior by its creators at JKU (university in Linz on Danube), who are near and dear to my heart.

Sepp Hochreiter's LSTM (kind of better RNN and precursor) is quoted by all AI gods and used by biggies like DeepMind.
xLSTM has totally been flying under the radar so far (even compared to Mamaba).

I released Python code today for inference (streaming) at AI-bits/xlstm, CUDA bug pending.

Model size is 7b, but at FP32 it's 30GB.
(Quantization to at least FP16 wouldn't hurt - to run on 1 consumer Nvidia GPU, but may be complicated due to architecture,)
Inference basically is pip install xlstm (or -e .), Torch 2.5.1+cu124 and custom Transformers. (not yet merged)

Would love to see some eposure to this Snowwhite.
Thanks in advance
G.

Originally created by @ai-bits on GitHub (Feb 22, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/9293 xLSTM is the 3rd of a kind (besides Transformers and Mamba) and its performace is touted to be superior by its creators at JKU (university in Linz on Danube), who are near and dear to my heart. Sepp Hochreiter's LSTM (kind of better RNN and precursor) is quoted by all AI gods and used by biggies like DeepMind. xLSTM has totally been flying under the radar so far (even compared to Mamaba). I released Python code today for inference (streaming) at [AI-bits/xlstm](https://github.com/ai-bits/xlstm), CUDA bug pending. Model size is 7b, but at FP32 it's 30GB. (Quantization to at least FP16 wouldn't hurt - to run on 1 consumer Nvidia GPU, but may be complicated due to architecture,) Inference basically is `pip install xlstm` (or `-e .`), Torch 2.5.1+cu124 and custom Transformers. (not yet merged) Would love to see some eposure to this Snowwhite. Thanks in advance G.
GiteaMirror added the model label 2026-05-09 19:01:22 -05:00
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 model
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#83746
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?