ros/LLaMA-Factory
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

[misc] fix packing and eval plot (#7623)

Browse Source
This commit is contained in:
hoshi-hiyouga
2025-04-07 18:20:57 +08:00
committed by GitHub
parent 5115dc8c7f
commit c3c0efbaa0
70 changed files with 288 additions and 194 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
tests/train/test_sft_trainer.py
View File

@@ -27,10 +27,10 @@ from llamafactory.train.sft.trainer import CustomSeq2SeqTrainer
DEMO_DATA = os.getenv("DEMO_DATA", "llamafactory/demo_data")
TINY_LLAMA = os.getenv("TINY_LLAMA", "llamafactory/tiny-random-Llama-3")
TINY_LLAMA3 = os.getenv("TINY_LLAMA3", "llamafactory/tiny-random-Llama-3")
TRAIN_ARGS = {
"model_name_or_path": TINY_LLAMA,
"model_name_or_path": TINY_LLAMA3,
"stage": "sft",
"do_train": True,
"finetuning_type": "lora",
@@ -41,6 +41,7 @@ TRAIN_ARGS = {
"overwrite_output_dir": True,
"per_device_train_batch_size": 1,
"max_steps": 1,
"report_to": "none",
}