Lincoln Stein 1788f4f6ec Merge branch 'main' into feat/ernie-image-gguf

2026-10-10 14:52:40 -04:00

..

gguf

feat(ernie-image): load GGUF transformers

2026-10-09 23:07:07 +02:00

scripts

feat - Migrate to Transformers 5.5.4 (#9248)

2026-06-29 22:01:20 +00:00

sdnq

fix(mm): stop the SDNQ guards from aborting folder-encoder identification (#9561)

2026-09-24 01:45:45 +00:00

__init__.py

Move requantize.py to the quatnization/ dir.

2024-08-26 20:17:50 -04:00

block_scale_tiles.py

feat(quantization): decode MXFP8 block scales instead of refusing them

2026-09-17 23:52:50 +02:00

bnb_cast_notice.py

fix(quantization): install the int8 cast notice filter without importing bitsandbytes

2026-10-09 07:27:32 +02:00

bnb_llm_int8.py

fix(quantization): install the int8 cast notice filter without importing bitsandbytes

2026-10-09 07:27:32 +02:00

bnb_nf4.py

Install sub directories with folders correctly, ensure consistent dtype of tensors in flux pipeline and vae

2024-08-26 20:17:50 -04:00

dequantizing_linear.py

fix(model-cache): reserve the GGUF dequantization transient in every node

2026-10-08 22:32:52 +02:00

fp8_scaled.py

fix(fp8): keep the matmul probe off the forward path, probe CUDA for storage

2026-09-24 02:21:20 +02:00

int8_convrot.py

test(z-image): record why the header spelling of int8 is refused here

2026-09-18 18:02:01 +02:00

load_plan.py

refactor(loaders): reserve through one function instead of assembling the sum

2026-09-26 08:38:47 +02:00

nvfp4.py

Merge branch 'feat/loader-conformance-driver' into feat/loader-conformance-matrix

2026-09-19 00:01:26 +02:00