Yong Hoon Shin 77d906995c [KV sharing] Re-land Gemma3n model changes from #22628 (#24357) ...

Signed-off-by: Yong Hoon Shin <yhshin@meta.com>

2025-09-23 19:25:34 -07:00

..

layers

[fix]: add Arm 4bit fused moe support (#23809)

2025-09-24 01:32:22 +00:00

model_loader

[TPU] Deprecate xm.mark_step in favor of `torch_xla.sync (#25254)

2025-09-22 10:12:57 +00:00

models

[KV sharing] Re-land Gemma3n model changes from #22628 (#24357)

2025-09-23 19:25:34 -07:00

warmup

[Bug] Fix AttributeError: 'FusedMoE' object has no attribute 'w13_weight_scale'. Did you mean: 'w13_weight_scale_inv' (#25519)

2025-09-24 00:07:51 +00:00

__init__.py

[V0 Deprecation] Remove V0 sampling metadata (#25345)

2025-09-21 10:37:11 -07:00

custom_op.py

[V0 deprecation] Deprecate V0 Neuron backend (#21159)

2025-09-06 16:15:18 -07:00

parameter.py

[Core] Support weight_loader_v2 for UnquantizedLinearMethod (#23036)

2025-09-23 18:30:26 -06:00

utils.py

[OOT] Support sync_model_loading for OOT (#25126)

2025-09-19 05:41:53 +00:00