['QEff.finetuning'] Changing some params from training config to model config#747

Merged

quic-akuruvil merged 28 commits intoquic:ft_experimentalfrom

tchawada:ft_config

Feb 5, 2026

Commits on Feb 5, 2026

General disagg fix for prefill-only model (#698)

ochougul
authored and
tchawada
committed
Evaluating the values of CCL lists for different scenarios (#710)

vjanfaza
authored and
tchawada
committed
Updated finetune docs for MULTI NODE Training (#717)

quic-akuruvil
authored and
tchawada
committed
Adding support for multi-node DDP training (#708)

smedhe
authored and
tchawada
committed
HOTFIX: changes in alpaca and grammar dataset utils (#724)

smedhe
authored and
tchawada
committed
Added default NPI file (#657)

quic-akuruvil
authored and
tchawada
committed
Removed OpenGVLab/InternVL2_5-1B and OpenGVLab/InternVL3_5-1B (#736)

quic-rishinr
authored and
tchawada
committed
Qeff versioning (#741)

quic-rishinr
authored and
tchawada
committed
Fix for Qwen 2.5 VL with subfunction (#733)

abhishek-singh591
authored and
tchawada
committed
Fixed torch patch for subfunction with VLMs (#750)

abhishek-singh591
authored and
tchawada
committed
Fixing SW issue in Gemma3 (#740)

qcdipankar
authored and
tchawada
committed
Adding Full document for hf_based finetuning stack
tchawada
committed
moving some params from trainconfig to modelconfig
tchawada
committed
Updating config manager so it include all params from master config
tchawada
committed
updating Config Manager to add CLI flags into file
tchawada
committed
updating Config Manager to add CLI flags into file
tchawada
committed
Updating config manager
tchawada
committed
Changing documentation and config manager
tchawada
committed
Updating config_manager
tchawada
committed
Updating config
tchawada
committed
Updating documentation
tchawada
committed
making torch_compile=False by default
tchawada
committed
[QEff.finetuning] Adding sample config and ReadMe file (#692)

committed
moving some params from trainconfig to modelconfig
tchawada
committed
Updating config manager so it include all params from master config
tchawada
committed
Changing documentation and config manager
tchawada
committed
Removing create_trainer_config from config_manager.py
tchawada
committed
Removing create_trainer_config from config_manager.py
tchawada
committed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Comments

['QEff.finetuning'] Changing some params from training config to model config#747

['QEff.finetuning'] Changing some params from training config to model config#747
quic-akuruvil merged 28 commits intoquic:ft_experimentalfrom
tchawada:ft_config

Commits on Feb 5, 2026

General disagg fix for prefill-only model (#698)

Evaluating the values of CCL lists for different scenarios (#710)

Updated finetune docs for MULTI NODE Training (#717)

Adding support for multi-node DDP training (#708)

HOTFIX: changes in alpaca and grammar dataset utils (#724)

Added default NPI file (#657)

Removed OpenGVLab/InternVL2_5-1B and OpenGVLab/InternVL3_5-1B (#736)

Qeff versioning (#741)

Fix for Qwen 2.5 VL with subfunction (#733)

Fixed torch patch for subfunction with VLMs (#750)

Fixing SW issue in Gemma3 (#740)

Adding Full document for hf_based finetuning stack

moving some params from trainconfig to modelconfig

Updating config manager so it include all params from master config

updating Config Manager to add CLI flags into file

updating Config Manager to add CLI flags into file

Updating config manager

Changing documentation and config manager

Updating config_manager

Updating config

Updating documentation

making torch_compile=False by default

[QEff.finetuning] Adding sample config and ReadMe file (#692)

moving some params from trainconfig to modelconfig

Updating config manager so it include all params from master config

Changing documentation and config manager

Removing create_trainer_config from config_manager.py

Removing create_trainer_config from config_manager.py