Skip to content

fix(Qwen3-VL-8B): replace deprecated tokenizer arg in SFTTrainer with processing_class and set eos_token in SFTConfig #253

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
Saad1926Q wants to merge 2 commits into
base: main
Choose a base branch
from
+10 −5
Open

fix(Qwen3-VL-8B): replace deprecated tokenizer arg in SFTTrainer with processing_class and set eos_token in SFTConfig #253

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
3c957df
fix(Qwen3-VL-8B): replace deprecated tokenizer arg with processing_cl…
Saad1926Q Apr 24, 2026
ff6540a
revert: restore model_created_at.csv to original state
Saad1926Q Apr 24, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
3 changes: 2 additions & 1 deletion nb/Kaggle-Qwen3_VL_(8B)-Vision.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -615,7 +615,7 @@
"\n",
"trainer = SFTTrainer(\n",
" model = model,\n",
" tokenizer = tokenizer,\n",
" processing_class = tokenizer,\n",
" data_collator = UnslothVisionDataCollator(model, tokenizer), # Must use!\n",
" train_dataset = converted_dataset,\n",
" args = SFTConfig(\n",
Expand All @@ -638,6 +638,7 @@
" dataset_text_field = \"\",\n",
" dataset_kwargs = {\"skip_prepare_dataset\": True},\n",
" max_length = 2048,\n",
" eos_token = tokenizer.tokenizer.eos_token,\n",
" ),\n",
")"
]
Expand Down
3 changes: 2 additions & 1 deletion nb/Qwen3_VL_(8B)-Vision.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -615,7 +615,7 @@
"\n",
"trainer = SFTTrainer(\n",
" model = model,\n",
" tokenizer = tokenizer,\n",
" processing_class = tokenizer,\n",
" data_collator = UnslothVisionDataCollator(model, tokenizer), # Must use!\n",
" train_dataset = converted_dataset,\n",
" args = SFTConfig(\n",
Expand All @@ -638,6 +638,7 @@
" dataset_text_field = \"\",\n",
" dataset_kwargs = {\"skip_prepare_dataset\": True},\n",
" max_length = 2048,\n",
" eos_token = tokenizer.tokenizer.eos_token,\n",
" ),\n",
")"
]
Expand Down
3 changes: 2 additions & 1 deletion original_template/Qwen3_VL_(8B)-Vision.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -588,7 +588,7 @@
"\n",
"trainer = SFTTrainer(\n",
" model = model,\n",
" tokenizer = tokenizer,\n",
" processing_class = tokenizer,\n",
" data_collator = UnslothVisionDataCollator(model, tokenizer), # Must use!\n",
" train_dataset = converted_dataset,\n",
" args = SFTConfig(\n",
Expand All @@ -611,6 +611,7 @@
" dataset_text_field = \"\",\n",
" dataset_kwargs = {\"skip_prepare_dataset\": True},\n",
" max_length = 2048,\n",
" eos_token = tokenizer.tokenizer.eos_token,\n",
" ),\n",
")"
]
Expand Down
3 changes: 2 additions & 1 deletion python_scripts/Kaggle-Qwen3_VL_(8B)-Vision.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -235,7 +235,7 @@ def convert_to_conversation(sample):

trainer = SFTTrainer(
model = model,
tokenizer = tokenizer,
processing_class = tokenizer,
data_collator = UnslothVisionDataCollator(model, tokenizer), # Must use!
train_dataset = converted_dataset,
args = SFTConfig(
Expand All @@ -258,6 +258,7 @@ def convert_to_conversation(sample):
dataset_text_field = "",
dataset_kwargs = {"skip_prepare_dataset": True},
max_length = 2048,
eos_token = tokenizer.tokenizer.eos_token,
),
)

Expand Down
3 changes: 2 additions & 1 deletion python_scripts/Qwen3_VL_(8B)-Vision.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -235,7 +235,7 @@ def convert_to_conversation(sample):

trainer = SFTTrainer(
model = model,
tokenizer = tokenizer,
processing_class = tokenizer,
data_collator = UnslothVisionDataCollator(model, tokenizer), # Must use!
train_dataset = converted_dataset,
args = SFTConfig(
Expand All @@ -258,6 +258,7 @@ def convert_to_conversation(sample):
dataset_text_field = "",
dataset_kwargs = {"skip_prepare_dataset": True},
max_length = 2048,
eos_token = tokenizer.tokenizer.eos_token,
),
)

Expand Down
Loading