llama.cpp
convert : ability to lazy-load safetensors remotely without downloading to disk
#12820
Merged

convert : ability to lazy-load safetensors remotely without downloading to disk #12820

ngxson
ngxson gguf util : add SafetensorRemote
2507c8e2
github-actions github-actions added python
ngxson fix style
7f61d0b2
ngxson convert: add --remote option
08ecbbe3
ngxson
compilade
compilade commented on 2025-04-08
compilade convert : allow using lazy remote tensors
3a3682de
ngxson
ngxson commented on 2025-04-08
compilade
ngxson
ngxson correct metadata.name
df95a3a7
ngxson
ngxson small style fix
4f657627
ngxson
ngxson ngxson marked this pull request as ready for review 1 year ago
ngxson ngxson requested a review from compilade compilade 1 year ago
compilade
ngxson
ngxson support HF_TOKEN
b584e397
ngxson
ngxson commented on 2025-04-08
ngxson
csabakecskemeti
compilade convert : use writeable buffer for remote lazy tensors
78094fc4
ngxson
compilade convert : fix flake8 lint regarding lamdba assigment
4c0170e2
compilade
ngxson multithreaded download
42fc895a
csabakecskemeti
ngxson multithread: print debug
63f0604a
ngxson fix style
c8760ccb
ngxson
ngxson
csabakecskemeti
ngxson
csabakecskemeti
csabakecskemeti
ngxson Revert "multithreaded download"
2e535f6f
compilade
compilade commented on 2025-04-09
ngxson bring back _get_request_headers
e8b7d263
ngxson
compilade
compilade approved these changes on 2025-04-10
ngxson ngxson merged 64eda5de into master 1 year ago
ddh0
ngxson
ngxson
ddh0

Login to write a write a comment.

Login via GitHub

Reviewers
Assignees
No one assigned
Labels
Milestone