423A35C7/LLaMA-Factory
1
0
Fork 0
mirror of https://github.com/hiyouga/LLaMA-Factory.git synced 2025-08-22 21:52:51 +08:00
Code Issues Packages Projects Releases Wiki Activity

improve data process logger

Browse Source 
Former-commit-id: a851056229f37391023627180b5712ed64ae3528
This commit is contained in:
hiyouga 2024-05-18 22:02:42 +08:00
parent c53e626c9a
commit 519d2511ae
3 changed files with 7 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
src/llamafactory/data/aligner.py
View File

@ -149,7 +149,7 @@ def convert_sharegpt(
chosen[dataset_attr.role_tag] not in accept_tags[-1] chosen[dataset_attr.role_tag] not in accept_tags[-1]
or rejected[dataset_attr.role_tag] not in accept_tags[-1] or rejected[dataset_attr.role_tag] not in accept_tags[-1]
): ):
logger.warning("Invalid role tag in {}.".format(messages)) logger.warning("Invalid role tag in {}.".format([chosen, rejected]))
broken_data = True broken_data = True
prompt = aligned_messages prompt = aligned_messages

5
src/llamafactory/data/preprocess.py
View File

@ -77,6 +77,7 @@ def preprocess_supervised_dataset(
for i in range(len(examples["prompt"])): for i in range(len(examples["prompt"])):
if len(examples["prompt"][i]) % 2 != 1 or len(examples["response"][i]) != 1: if len(examples["prompt"][i]) % 2 != 1 or len(examples["response"][i]) != 1:
logger.warning("Dropped invalid example: {}".format(examples["prompt"][i] + examples["response"][i]))
continue continue
if processor is not None: if processor is not None:
@ -129,6 +130,7 @@ def preprocess_packed_supervised_dataset(
input_ids, labels = [], [] input_ids, labels = [], []
for i in range(len(examples["prompt"])): for i in range(len(examples["prompt"])):
if len(examples["prompt"][i]) % 2 != 1 or len(examples["response"][i]) != 1: if len(examples["prompt"][i]) % 2 != 1 or len(examples["response"][i]) != 1:
logger.warning("Dropped invalid example: {}".format(examples["prompt"][i] + examples["response"][i]))
continue continue
messages = examples["prompt"][i] + examples["response"][i] messages = examples["prompt"][i] + examples["response"][i]
@ -178,6 +180,7 @@ def preprocess_unsupervised_dataset(
for i in range(len(examples["prompt"])): for i in range(len(examples["prompt"])):
if len(examples["prompt"][i]) % 2 != 1: if len(examples["prompt"][i]) % 2 != 1:
logger.warning("Dropped invalid example: {}".format(examples["prompt"][i] + examples["response"][i]))
continue continue
if processor is not None: if processor is not None:
@ -224,6 +227,7 @@ def preprocess_pairwise_dataset(
for i in range(len(examples["prompt"])): for i in range(len(examples["prompt"])):
if len(examples["prompt"][i]) % 2 != 1 or len(examples["response"][i]) < 2: if len(examples["prompt"][i]) % 2 != 1 or len(examples["response"][i]) < 2:
logger.warning("Dropped invalid example: {}".format(examples["prompt"][i] + examples["response"][i]))
continue continue
if processor is not None: if processor is not None:
@ -285,6 +289,7 @@ def preprocess_kto_dataset(
for i in range(len(examples["prompt"])): for i in range(len(examples["prompt"])):
if len(examples["prompt"][i]) % 2 != 1 or len(examples["response"][i]) < 2: if len(examples["prompt"][i]) % 2 != 1 or len(examples["response"][i]) < 2:
logger.warning("Dropped invalid example: {}".format(examples["prompt"][i] + examples["response"][i]))
continue continue
if processor is not None: if processor is not None:

2
src/llamafactory/train/kto/workflow.py
View File

@ -62,7 +62,7 @@ def run_kto(
trainer.save_metrics("train", train_result.metrics) trainer.save_metrics("train", train_result.metrics)
trainer.save_state() trainer.save_state()
if trainer.is_world_process_zero() and finetuning_args.plot_loss: if trainer.is_world_process_zero() and finetuning_args.plot_loss:
plot_loss(training_args.output_dir, keys=["loss", "eval_loss"]) plot_loss(training_args.output_dir, keys=["loss", "eval_loss", "train/rewards/chosen"])
# Evaluation # Evaluation
if training_args.do_eval: if training_args.do_eval: