make e2e tests a bit faster by reducing test split size (#2522) [skip ci]

* [ci] make e2e tests a bit faster by reducing test split size

* use 10% split of alpaca dataset to speed up dataset loading/tokenization

* reduce gas 4->2 for most e2e tests

* increase val set size for packing
This commit is contained in:
Wing Lian
2025-04-12 07:24:43 -07:00
committed by GitHub
parent 51267ded04
commit de8a625dd7
24 changed files with 54 additions and 44 deletions

View File

@@ -62,7 +62,7 @@ class TestGemma2:
"chat_template": "gemma", # gemma2's template is same as gemma
"num_epochs": 1,
"micro_batch_size": 1,
"gradient_accumulation_steps": 4,
"gradient_accumulation_steps": 2,
"output_dir": temp_dir,
"learning_rate": 0.00001,
"optimizer": "adamw_bnb_8bit",
@@ -114,7 +114,7 @@ class TestGemma2:
},
"num_epochs": 1,
"micro_batch_size": 1,
"gradient_accumulation_steps": 4,
"gradient_accumulation_steps": 2,
"output_dir": temp_dir,
"learning_rate": 0.00001,
"optimizer": "adamw_bnb_8bit",