Commit Graph

4 Commits

Author SHA1 Message Date
Wing Lian
830e9f7eaf automatically enable tf32 if supported (#3473) [skip ci]
* automatically enable tf32 if supported

* update fixtures

* handle only when True

* Address CR comments

* address readability from pr comment

* simplify
2026-03-16 23:47:00 -04:00
VED
a806704e94 moe quant patch for merge miss match (#3483)
* moe quant patch for merge miss match

* lint

* revert test + fix moe patch

* comment fixxes

* e2e tests

* mismatch fixx tested

* mis match fix wwith vllm compatablity + test

* comment lint

* fix: missing os import, duplicate no op

* chore: simplify comments

---------

Co-authored-by: NanoCode012 <nano@axolotl.ai>
2026-03-15 22:10:30 -04:00
NanoCode012
6c8c73e5a4 fix(validation): add validation for lora target linear with quantize experts (#3461)
* fix: add validation for lora target linear with quantize experts

* chore: fix lint

* chore: comment

* fix: missing link on readme
2026-03-06 09:19:05 -05:00
NanoCode012
945c8aeb10 Fix: quantize and target moe layers in transformers v5 for adapters and many misc fixes (#3439)
* fix: saving clones state dict

* fix: apply fix for only CP mode

* fix: add dropout check when using lora target param

* fix: re-add patch from transformers PR #39866

* feat: add moe quant to test by ved

* fix: try match target param properly end with

* fix: clear cache per param quant

* fix: attempt on-load quantize experts instead of post-load

* fix: attempt disable async load

* chore: add log

* chore: adjust log

* fix: remove cuda alloc for moe and enable async load

* chore: remove leftover logs

* chore: add extra empty cache

* fix(doc): clarify support

* fix: handle fsdp2 for paramwrapper dtensor

* feat: attempt to quant experts in 8bit mode too

* feat: attempt to release bf16 experts from vram

* feat: upgrade cce

* fix: fsdp2 init_sharded_param load int8/uint4 dtensor as
require_grad=true on init

* fix: remove unnecessary gc and empty cache

* Revert "fix: remove unnecessary gc and empty cache"

This reverts commit 1d54518990.

* fix: do not call full_tensor on non-dtensors

* fix: attempt to address fsdp2 with quant exp high loss

* fix: attempt lora quant experts wrong dim

* fix: ensure require_grad patch applied for lora 8bit

* fix: attempt lora 8bit fsdp2

* fix: attribute access on save for lora 8bit fsdp2

* fix: wrong weight attrib access

* chore(refactor): add config, re-arrange position of patches, clean
comments

* feat: add example docs

* chore: cherry pick trinity fixes from PR 3399

* chore: comments refactor; add guards

* fix: guard using wrong key

* fix: mamba save does not accept main process param

* fix: guard prevent double hook

* fix: move gc to upper scope

* chore: add comment on proxy forward patch

* fix: add comment to clarify

* feat: add test idempotency

* fix: AttributeError: `e_score_correction_bias` is not an nn.Parameter

* fix: AttributeError: 'NoneType' object has no attribute 'to'

* fix: update docs on cpu_ram_efficient_loading
2026-03-03 10:06:23 -05:00