remove DetachedEagleGPT model and handle all offline mode in the _DynamicEagleGPTModel#321

Merged

yeyu-nvidia merged 29 commits intomainfrom

yeyu/move_offline_eagle_to_online

Sep 18, 2025

Commits on Sep 18, 2025

remove detachedeagle and handle all offline mode in the eagle model
yeyu-nvidia
committed
apply coderabbit suggestion
yeyu-nvidia
committed
remove OfflineEagleDMRegistry
yeyu-nvidia
committed
apply suggestion to cover eagle1 case
yeyu-nvidia
committed
debug
yeyu-nvidia
committed
debug
yeyu-nvidia
committed
minor
yeyu-nvidia
committed
fix the bug in megatron import
yeyu-nvidia
committed
Added support for qwen3-next quantization and export (#323)

kinjalpatel27
authored and
yeyu-nvidia
committed
Update distill Megatron plugin (#319)

AAnoosheh
authored and
yeyu-nvidia
committed
Remove unused utilities for ModelOpt <0.29 MCore checkpoints (#322)

realAsma
authored and
yeyu-nvidia
committed
Upgrade TensorRT-LLM docker to 1.1.0RC2 (#327)

cjluo-nv
authored and
yeyu-nvidia
committed
[1/N] QATTrainer training workflow fixes and clean up; Added backend specific unitests; (#318)

realAsma
authored and
yeyu-nvidia
committed
import fix for torch 2.9 (#315)

i-riyad
authored and
yeyu-nvidia
committed
checkout main branch's fix for megatron+importer
yeyu-nvidia
committed
pad labels if it's 1 token shorter than input_ids
yeyu-nvidia
committed
Revert "pad labels if it's 1 token shorter than input_ids"
yeyu-nvidia
committed
pad labels if it's 1 token shorter than input_ids
yeyu-nvidia
committed
Fix issues of attention.core_attention.softmax_offset is None (#330)

yueshen2016
authored and
yeyu-nvidia
committed
update eagle example notebook (#314)

h-guo18
authored and
yeyu-nvidia
committed
Allow KD loss in val mode for MLM plugin (#331)

AAnoosheh
authored and
yeyu-nvidia
committed
Deprecate TRTLLM-build in examples (#297)

authored and
yeyu-nvidia
committed
Fix issue of attention.core_attention is None (#334)

yueshen2016
authored and
yeyu-nvidia
committed
Reinstate int8_sq support for vlm_example. (#333)

cjluo-nv
authored and
yeyu-nvidia
committed
Update News README.md (#336)

omrialmog
authored and
yeyu-nvidia
committed
[NVBug: 5525758] Update VLM-PTQ readme (#339)

cjluo-nv
authored and
yeyu-nvidia
committed
Slurm support for QAT Simplified Flow + Qwen3-8B recipe (#285)

jenchen13
authored and
yeyu-nvidia
committed
Merge branch 'main' into yeyu/move_offline_eagle_to_online
yeyu-nvidia
authored
Merge branch 'main' into yeyu/move_offline_eagle_to_online
yeyu-nvidia
authored

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

remove DetachedEagleGPT model and handle all offline mode in the _DynamicEagleGPTModel#321

remove DetachedEagleGPT model and handle all offline mode in the _DynamicEagleGPTModel#321
yeyu-nvidia merged 29 commits intomainfrom
yeyu/move_offline_eagle_to_online

Commits on Sep 18, 2025

remove detachedeagle and handle all offline mode in the eagle model

apply coderabbit suggestion

remove OfflineEagleDMRegistry

apply suggestion to cover eagle1 case

debug

debug

minor

fix the bug in megatron import

Added support for qwen3-next quantization and export (#323)

Update distill Megatron plugin (#319)

Remove unused utilities for ModelOpt <0.29 MCore checkpoints (#322)

Upgrade TensorRT-LLM docker to 1.1.0RC2 (#327)

[1/N] QATTrainer training workflow fixes and clean up; Added backend specific unitests; (#318)

import fix for torch 2.9 (#315)

checkout main branch's fix for megatron+importer

pad labels if it's 1 token shorter than input_ids

Revert "pad labels if it's 1 token shorter than input_ids"

pad labels if it's 1 token shorter than input_ids

Fix issues of attention.core_attention.softmax_offset is None (#330)

update eagle example notebook (#314)

Allow KD loss in val mode for MLM plugin (#331)

Deprecate TRTLLM-build in examples (#297)

Fix issue of attention.core_attention is None (#334)

Reinstate int8_sq support for vlm_example. (#333)

Update News README.md (#336)

[NVBug: 5525758] Update VLM-PTQ readme (#339)

Slurm support for QAT Simplified Flow + Qwen3-8B recipe (#285)

Merge branch 'main' into yeyu/move_offline_eagle_to_online

Merge branch 'main' into yeyu/move_offline_eagle_to_online