cmedina/agenx-lora-training
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

fix: set gradient_checkpointing=False (FSDP handles it)

Browse Source
This commit is contained in:
Christian Medina
2026-07-02 09:38:51 -04:00
parent 361a58addc
commit f1e016c8af
1 changed files with 1 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
train.py
View File

@@ -168,7 +168,7 @@ def train(config_path):
eval_strategy=config.get("eval_strategy", "steps"), eval_strategy=config.get("eval_strategy", "steps"),
eval_steps=config.get("eval_steps", 100), eval_steps=config.get("eval_steps", 100),
bf16=True, bf16=True,
gradient_checkpointing=True, gradient_checkpointing=False, # FSDP activation_checkpointing handles it
fsdp=True, fsdp=True,
fsdp_config={ fsdp_config={
"sharding_strategy": "SHARD_GRAD_OP", "sharding_strategy": "SHARD_GRAD_OP",