supplymind / scripts /hf_train_supplymind_roles.py

Commit History

Align warehouse GRPO prompt with SFT
0ac1e98
verified

rishavutk commited on

Expose GRPO generation batch size
87004db
verified

rishavutk commited on

Add 4bit GRPO training knobs
bbb6307
verified

rishavutk commited on

Allow configurable GRPO base model
130e9e4
verified

rishavutk commited on

Tighten role training scaffold
a2144da

Rishav commited on

Add SFT warm start training pipeline
be8d222

Rishav commited on

Disable Trackio checkpoint sync
aa7d416

Rishav commited on

Tune GRPO completion length
cce2fba

Rishav commited on

Make GRPO config version tolerant
15084c8

Rishav commited on

Fix HF training config resolution
d77a60c

Rishav commited on

Harden HF training startup
1cd6456

Rishav commited on

Add role-specific training scores
37049ad

Rishav commited on

Add training progress logs
a7160ed

Rishav commited on

Fix training smoke task id
5ca7724

Rishav commited on

Add HF role GRPO training job
3f1eabc

Rishav commited on