tocmo0nlord/axolotl
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

Accelerate 1.8.1 and BNB 0.46.0 update (#2815)

Browse Source 
* update accelerate to v1.8.0

* update bnb also

* fix multigpu ci timeout

* fix test set size

* use latest accelerate 1.8.1

* disable default dtype
This commit is contained in:
Wing Lian
2025-06-28 15:29:19 -04:00
committed by GitHub
parent a1a740608d
commit 81893c775c
11 changed files with 32 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
tests/e2e/multigpu/solo/test_flex.py
View File

@@ -54,6 +54,7 @@ class TestPackedFlex:
"gradient_accumulation_steps": 2,
"gradient_checkpointing": True,
"output_dir": temp_dir,
"dataset_prepared_path": temp_dir + "/last_run_prepared",
"learning_rate": 0.00001,
"optimizer": "adamw_torch_fused",
"lr_scheduler": "cosine",

2
tests/e2e/multigpu/solo/test_grpo.py
View File

@@ -309,6 +309,7 @@ def oai_gsm8k_transform(cfg, *args, **kwargs):
"warmup_steps": 10,
"val_set_size": 0.0,
"output_dir": temp_dir,
"dataset_prepared_path": temp_dir + "/last_run_prepared",
"learning_rate": 0.0001,
"optimizer": "adamw_torch_fused",
"lr_scheduler": "cosine",
@@ -400,6 +401,7 @@ def oai_gsm8k_transform(cfg, *args, **kwargs):
"warmup_steps": 10,
"val_set_size": 0.0,
"output_dir": temp_dir,
"dataset_prepared_path": temp_dir + "/last_run_prepared",
"learning_rate": 0.0001,
"optimizer": "adamw_torch_fused",
"lr_scheduler": "cosine",